Feature articles

Introduction

[specs-box]

VRayPattern is a 3DS Max plugin made by iCube, publishers of the well-received MultiScatter and VRayScatter plugins. At first glance VRayPattern appears to be another entry into the burgeoning proxy scattering market, already well served by the aforementioned and competitors such as Carbon Scatter, Forest Pack Pro and ScatterFX. It is tempting to ask what VRayPattern, with its comparatively minimal feature-set, can offer to differentiate itself. Upon closer inspection however, we find a tool that takes a rather different approach to multiplying geometry, sidestepping proxies to allow a literally limitless amount of geometry to be rendered.

The plugin exploits a feature of the V-Ray SDK that enables the use of geometry as a texture, somewhat akin to vector displacement, but instead of 2D maps it enables any source mesh to be distributed onto a target surface. This unique approach lends itself to situations that require dense geometry that either tiles seamlessly or in a regularly modulated pattern.

To appreciate the uses of VRayPattern it is helpful to think of it as a 3D displacement texture map. This explains its limitations - a lack of advanced randomising or distribution algorithms - and also its main strengths: fast rendering, the ability to deform the source geometry to follow a target surface, and low memory usage.

First Impressions and Workflow.

To create a new VRayPattern object you simply navigate to the V-Ray geometry roll-out and create a new instance in the viewport. Its unassuming UI can be found in the modify panel just like any other object. Using it couldn’t be simpler: a PDF manual is provided, but will require little more than a cursory glance to get up and running. Part of VRayPattern’s efficacy is due to its ease of use and speed to set up.

There are two key components of the interface: base and surface parameters. Using these, the workflow is very straightforward, simply pick base geometry from either the scene or a separate file, and identify the surface over which it should be distributed. There are additional settings to control the cropping, orientation and tiling of the base geometry, restrict the distribution to particular map channels, and some features to create 2D previews, but little else. This is not a tool with a steep learning curve.

To see how well the plug in performs we’ll explore each of the settings in the context of a simple test scene. The objective was to use the plugin for as many elements as possible. In the final render only the road and the roof were not created with the plugin.

Creating the Geometry

The most time-consuming aspect of using VRayPattern is creating suitable geometry. There are two conditions this must fulfil. Firstly, in order that the pattern’s edges are not obvious, the base geometry must tile seamlessly or have obtrusive joints designed-in. Secondly, the surface must be UVW unwrapped with minimal distortion. Since the pattern’s height scales proportionately to its footprint, overly stretched UVWs will result in geometry unexpectedly jutting from the surface.

VRayPattern has no part to play in this process directly but it’s worth noting since the creation of appropriate meshes is by far the most labour-intensive step. However these are likely to have been necessary regardless of whether this plug in is part of your workflow.

Setting up the Pattern (Base Parameters)

Once created, your patterns can be either stored directly in the scene or in a separate file. This is a useful feature that allows you to build a library of reusable assets, especially in a large studio where a central repository is desirable.

When storing your patterns in an external file there are a few limitations to be aware of. The library browser does not use a standard Windows file selector, which by itself is not a problem, but the replacement lacks features we’re accustomed to and is slow to navigate. The reason for a custom selector is to allow VRayPattern to remember the location of the library - set by clicking “save library path”. This is a great idea, with a single click you can access your library. However, in order to set the path, 3DS Max must be running as an Administrator and more irritatingly, it wouldn’t store folders with spaces in the name. Fortunately if your library location isn’t going to change, you need only run as Administrator once. The problem can probably be resolved with some judicious permissions tinkering.

Once you’ve negotiated these limitations the library is a very useful feature. Each file can have a 160 x 100 pixel JPG assigned to it for easy identification, and it allows for much more organised and less cluttered scene management.

When the base geometry has been picked, a crop box surrounds its extents. This is used to sample the geometry to be tiled, anything outside the box is ignored. The size is fully customisable using spinners or by scaling its sub object. This is essential for objects where the edges are ill-defined such as a grass pattern where leaving the crop box at the defaults would result in gaps between the tiles.

When creating the test scene I found that any boundary geometry that ran parallel to the crop box could often be lost using the default size settings, even when the edges are well-defined (This can be seen on the cornice in the final render). To fix this problem it’s a simple a case of slightly increasing the size of the crop box on the appropriate axis.

By default, the crop box is oriented to the XY world space. This is easily changed by rotating the crop box sub-object, unless your pattern is in an external file where rotating the crop box doesn’t appear to work. In these cases it’s best to ensure the pattern file is aligned correctly to avoid frustration later but it would be better if the orientation could be changed directly from within the scene file.

A crucial setting is the shift spinner that adjusts a plane on the pattern object to determine the position to be aligned to the host surface. Shift was crucial in the test scene to ensure that the different building components worked seamlessly, unfortunately it doesn’t work quite as advertised.

I’ve created an image to explain the anomaly. In this quick test an object was created with clear marks at 25% intervals. Logically, if the shift setting is half the height of the pattern it should intersect the surface at its mid point. What we actually see however is the pattern intersecting at a quarter of its height. If however, shift is equal to the height of the object it will intersect half-way. It seems the intersection point is always half the value of the shift measurement. This will probably be ironed out in future releases - and once realised it was easy to remedy - but it might cause some confusion to new users.

A simple toggle allows you to switch between texturing the pattern using its own or the surface’s UVW coordinates. This feature allows for texture variations across the whole surface and goes some way to allowing variety in the pattern. In the test scene this is used on the grass to introduce a subtle hue shift across the meadow.

Materials must be applied to the VRayPattern object and not the pattern geometry itself. The plugin does not automatically adopt the material of the pattern object - though it would be nice if this were included as an option - unless it is saved in an external file. 

Setting up the Surface (Surface Parameters)

The surface section is used to select and control the pattern’s distribution. The plugin is able to easily control the number of tile iterations using simple UV spinners or real-world measurements. A useful feature is the ability to choose map channels to allow for different tiling configurations on a single surface, such as offsetting pattern’s joints or rotating the UV coordinates.

The surface settings are very easy to set up and use, though currently each pattern can only be applied to a single object. To have the same pattern on multiple surfaces you must either duplicate the pattern or attach all the surfaces together to make a single poly object. The ability to pick a list of surfaces would be a far more elegant solution.

Specific material IDs can be targeted to enable patterns to apply to a part of a surface. This can be seen in the test scene where the facade is broken into IDs to identify different stories, windows, cornices, etc. Moreover some randomisation is possible. The crop box can be regularly subdivided using the random segment count controls. VRayPattern will then pick one segment at random as the pattern object. By including blank spaces it's possible to create a more irregular scatter pattern. The chimneys, flowers, and open windows in the test scene use this principle and it’s a very simple way of adding some variety to your renders.

Generating Previews

In the Preview section, the plugin allows you to control the size of the icon in the viewports and create preview textures. The latter is invaluable when working out pattern tiling, since there is no other way to see the end result without performing time-consuming renders.

Clicking the “Make Preview Texture” button will render a black and white preview. Once saved, it is necessary to create a new material, and apply it to the surface. Though not a major or lengthy procedure, I must confess that I only did this when it was absolutely necessary. An option to automatically create a simple material and apply it to the surface would be a valuable time saver. Doubly so if the UV tiling of the VRayPattern objects and the bitmap were wired together to remove the need to update them both separately.

G-Buffer and Animation Support

In the test renders, VRayPattern had no issues supporting G-Buffer channels and render elements.

Since VRayPattern works on texturing principles, animation of the surface geometry is supported, however at present, animation of the pattern geometry is not. The animation state is sampled at frame 0 irrespective of the active timeline so there is also currently no easy workaround. If the technology behind VRayPattern is capable of supporting this, it would be an excellent additional feature.

Conclusion

When originally mentioned on the Chaos Group forums back in 2010, VRayPattern was described as “Grass Sea”, which gave an indication of its initial intended purpose and put it in direct competition with scattering plugins and one-click solutions like Autograss. But by allowing the plugin to use any geometry, operate like a texture, and render limitless polygon counts, iCube created a seemingly peerless tool that has the potential to change our approach to modelling and texturing repetitive geometry.

At present VRayPattern is a solid release that has many possible applications. Most of the bugs and quirks are mere irritations that once understood, do not hinder the functionality of a plugin that I continue to find many uses for, particularly as a rough-and-ready panelling tool.

If you find yourself regularly creating vast swathes of repetitive geometry and want low memory consumption and easy set-up, VRayPattern will prove a wise investment. If iCube can resolve the shift plane problem and enable the easy application of a pattern to multiple surfaces, then it could progress from very useful, to invaluable.

By Pablo Hadis and Jason Addy

Introduction

CG architectural visualization is a recent innovation in the long history of architecture, which spans more than 9,000 years (according to some theorists). However, it has quickly become one of the most dynamic fields in computer graphics. Linked to the construction industry, the need for realistic project previews has resulted in (or stimulated the) development of a great number of related software, dedicated to tasks such as the creation of plants, grass, snow, parquet, roofs, terrain, urban environments, parametric models, skies, crowds, etc. It has also spawned a long list of related assets and information, including textures, models, materials, HDRI maps, tutorials, conferences, websites and more.

The progressive refinement and specialization of these tools is a testimony of the rapid evolution of the field, which presents the habit of segmenting reality into a dual interior/exterior classification, a trait inherited from one of architecture's primary goals. The influence of architectural visualization in 3D computer graphics is so strong that it has affected the growth and development of render engines themselves, and brought a new wave of real-time incarnations aimed straight at Archviz projects. It is also, of course, having a significant influence on architectural practice, and through it, on our own environment.

In our quest to analyze the history and evolution of the CG Industry, we've had the great pleasure of interviewing the founders and chief creatives at Neoscape, a pioneering studio in the field of architectural visualization that has gained worldwide recognition. Neoscape was founded in 1995 (the same year as Blur Studio) and has grown and expanded steadily ever since, witnessing and participating in the evolution of the field.

Join us once again as we take a look at the past, present and future of the Archviz Industry. Enjoy!

Neoscape interviewees:

- Principal and Founding Partner, Rob MacLeod
- Principal and Founding Partner, Rod MacLeod
- Principal and Founding Partner, Nils Norgren
- Principal and Chief Creative Officer, Rodrigo Lopez
- Associate Principal and Director of Visualization, Carlos Cristerna

The Interview

Please tell us a bit about how Neoscape was born. What made you decide to leave Parsons Brinckerhoff and found Neoscape? How did you come up with the name?

In the early ’90s, the three of us were working for Parsons Brinckerhoff – I was based out of NYC, and Rod and Nils were both working out of the Denver office. Together, we started formulating a plan to start a visualization business that would bring us back into working on architecture and real estate development projects. We worked on a business plan for about a year before giving notice and launching Neoscape.

An interesting part of starting a business is coming up with just the right name. Neoscape was born around the Norgren family kitchen table, and Nils’ father Phil was instrumental in us choosing that moniker for our business.

How did you choose the location for Neoscape's first office?

We actually started the business in Cambridge, MA with an office in Central Square. The location was chosen due to its proximity to many of the top architectural firms in the region, as well as to MIT and Harvard (where we hoped to gain access to a deep talent pool). After nine years in Cambridge, we relocated to South Boston. The move was predicated on us having outgrown our space and wanting to create a workspace that would not only allow for physical growth, but that would also reflect the changing nature of our business. Boston has been a great location for us in terms of growth and in attracting top talent.

What difficulties did you face when starting the company?

Hmm…where to begin? In hindsight, much of what we undertook and ultimately accomplished during the first few years was very difficult. Had we known just how hard, we may never have taken that crucial first step. But we were young, full of big ideas, and eager to create something lasting. One of the principles that helped us overcome some of these difficulties was creating an environment that allowed us to have fun at work. We knew there would be long hours; we knew there would be low pay; we knew there would be technology hurdles—but at the end of the day, we wanted to create a studio that had a unique culture and allowed us to grow a business by adding like-minded artists to the company.

Some of the biggest challenges revolved around us being a pioneer in the field of architectural visualization. Half the battle was educating our potential customers about the benefits of 3D visualization and animation.

Also, as a three-person company we often found ourselves in a vicious cycle of working extremely hard to win work, and then once the projects were landed, focusing all of our attention on executing them. This cycle gets old very quickly, but in those days it was absolutely necessary to build a client base. It was typical to work 100-hour weeks during the first couple of years.

When did you find yourselves needing to start expanding the team?

After the first year, we knew that there was way too much work for the three of us to handle alone. We had been hosting a user group as a way to network and meet local artists, and it was from that pool that we hired our first employee – an artist who still works for us. Early on, we were given an important piece of advice: always hire people who are better than you.This particular individual is perhaps the single most talented artist we’ve ever met. The most amazing imagination, knowledge of art history, and incredible hand skills. Our challenge was to help him transition his talents to a new set of digital tools. He ultimately helped elevate the artistry of our business. We also hired an office manager at that time to help manage all of the administration that comes with running a business.

How did the company grow through the years, what were the key factors in stimulating this continued growth?

Sales volume was a key to growing the business. In the beginning, our growth was about adding more 3D artists, animators and account execs. Eventually, we expanded our service offerings to include interactive, design and branding, which required artists with new and varied skill sets.

At its largest, how many people big is the production studio? How many projects are handled at the same time on average?

At its largest, the studio was about 60 people. We are now 55 people split between two offices – 40+ in Boston and 10 in NYC. The number of projects is variable and ranges anywhere from 20 to 30 of varying sizes and durations.

You founded Neoscape in 1995, and prior to that had been working on CG visualization for a number of years. How has the architectural visualization industry changed through these years? How has the profession of CG artists in Archviz evolved?

We’ve seen a large number of changes in the industry over the years: Photoreal Vs Art, GI, unbiased rendering, render engine battles, etc. The evolution of the profession has expressed itself in many ways: technically, artistically, and in industry maturity. Technically, the hardware has evolved in every way. The computing power available now greatly outstrips anything available when we started in 1995, an expected and predictable growth. The software and techniques also have evolved—often in more unexpected ways. The ability for massive scenes, (almost) unlimited polygons, global illumination, distributed rendering, 3d capabilities in compositing applications…the list goes on and on. Artistically, the profession has evolved as well, taking its cues from more established media such as traditional architectural illustration, architectural photography, filmmaking, fashion, product rendering, etc. This evolution is evident in high levels of work. Photo-real is only the starting point of a good rendering—artistic considerations like composition, subject matter, perspective, visual balance, contrast, color, and coincident line consideration (among others) all make a successful rendering. The most interesting evolution is the way in which the entire industry has matured. The profession has gone from rare, specialized, and esoteric to mainstream. Only in the last few years are there education paths for students to enter into the profession. It is now customary and expected that some sort of CG is created for large developments, for everything from conceptual design through to marketing materials. The profession is respected, and the artists are seen as much more than “button pushers.” Just like a great camera can’t make a great photographer, good software and computers can’t make a great CG artist. Artistic talent, eye, experience and thirst for excellence generally trump technical knowledge—all of which are necessary to get ahead in the industry today.

How has the production process at Neoscape changed through the years?

In a nutshell: deadlines are shorter, and clients are less and less easily impressed. As a result, we have now come full circle to give new value to the product: artistry. Marketing value as such is no longer about a rendering looking like something or looking real. Architectural rendering before CG was pure art, then it became gimmicky, and now it’s time we make it art again.

What is the largest project Neoscape has done to date?

We’ve completed many large-scale projects both in scope and size, but the one that came as a surprise for all of us at the time was a huge project for Callison in the Middle East, “Marsa Al Seef.” Like many of the projects being built in that area, the existing site was ocean, and developers were going to pump sand to make land for it. Marsa Al Seef was made up of more than 100 buildings, each about 40 to 50 stories high, plus two iconic towers and villas. This was a city from scratch. At the time, we weren’t sure how we were going to render all of these buildings—we didn’t know if the render farm could handle it, or how the process of building the whole city would work since there was very little design for the buildings. We created mostly everything from sketches we received from the architect. Back then, we didn’t have a plug-in to manage high-density landscaping or big volumes of polygons. We simply had to come up with some creative ways of putting this city together with some fake GI and a lot of faith that it would render. We ended up producing a 5-minute animation and 10 renderings. It was a huge learning experience for all of us. Large projects don’t seem nearly as daunting after Marsa Al Seef.

Neoscape's main focus has been architectural visualization, but you're now also producing other type of work. Please tell us more about this.

We focus primarily on architectural visualization, but we do take work outside of our traditional strengths when opportunity arises. We have corporate clients that have become some of our biggest clients, but for the most part what we create for them is very similar to the work for our traditional clients. We have also expanded the range of services that we offer to our core clients. Clients that may have in the past only thought of us for architectural visualization, now ask us to create and develop multifaceted marketing assets that include photography, branding and design, as well as interactive presentations, mobile apps, and web design. We have been getting much more involved in iOS development, which we target towards our core audience, although there is no reason that what we create couldn’t be used outside our normal circles. We tend to follow the philosophy of “new services to old clients and old services to new clients.”

What are your thoughts on the amount of work being outsourced to large studios in other countries? Has this negatively affected the industry?

As the architectural visualization industry developed over the years, we started to see more and more people creating digital visualizations. The software got better—faster and easier to use—prompting the emergence of many more studios capable of producing “renderings”. This evolution applied as well to many architectural firms who began to rely on their own artists for 3D viz. This shift definitely made a dent in the industry and, as expected, competition grew and prices shifted. Outsourcing became just one more part of that process. Now we find that more and more clients who tried outsourcing are coming back to us not only because of the lower quality of work, but also because of the “brain damage” from working with these facilities. So, has it affected the industry negatively? Perhaps at some level. But even though the shift was inevitable, clients noticed the big difference in service and quality of work—both of which they grew to miss, prompting a return to smaller agencies. Now, a lot of these outsource studios are taking on work that doesn’t require the level of detail or the investment to create what is needed for a particular delivery.

Has Neoscape taken on any real-time rendering technology?

No, we have been asked by some of our clients to use this technology to advertise their products, but that has been the only instance (aside from Unreal Engine way back when) where we have actually used a real-time unbiased render engine for production of still imagery. The technology has a lot of potential but is not there yet –at least not for the way we do things right now. We think once GPUs become cheaper and faster this could be something used every day to preview, manipulate light and materials with real-time feedback, and even create interactive applications where you can move things around and design.

What is Neoscape's go to software combination?

3ds Max, Photoshop, After Effects, and Final Cut are our work horses, but there are many other programs we use for particular tasks: PFTrack, RealFlow, Mocha, Adobe Creative Suite, etc.

Does Neoscape rely on any key plugins or scripts on a daily basis?

Plug-ins are something of a double-edged sword. We are very careful when and how we use them. Plug-ins need to pass a series of tests amongst production and our Technical Director before we use them. Basically, we’ll use a plug-in if we think it will consistently save time and money. To give you an example, one plug-in we just got recently (and that we use a lot) is “Forest Pack”. Of course, the great help it provides doesn’t come without its pains. We’ve found along the way that not only are there kinks, but there are quite simply limitations to its use. We also write our own “plug-in” scripts to tackle otherwise boring and repetitive tasks.

How much time do you dedicate to the different parts of the CG process in a given project?

Modeling, texturing, and lighting go hand-in-hand, so I would give them on average the same amount of time. As for compositing, we end up giving our film team very little time to put things together (sometimes just a day or two). Ideally, we’d like to give them more time, but in the process of accommodating our clients’ last-minute requirements and short deadlines, we end up having to eat into our own schedule. And, as always happens, the guys at the end of the process are the ones that suffer. In an ideal scenario, we would devote a reasonable and equal amount of time to all tasks in this order: modeling, lights and materials, compositing/editing, and color correction.

Do the artists at Neoscape lean more towards specialization or general arts skills?

We have both in our studio. We have people that are very good at certain things (as is the nature of things), but we encourage all of our artists to learn and become proficient in all the arts or skills. We all like to be involved in the process from the creative conception to the final color grading, and we have a very flexible team to get all of this done.

Most of our artists are “generalists”, as they’re called today, but when things need to get done quickly, we know who to go to, who will get it done most efficiently, and also who will teach new generations of artists that specific craft. Another key component is the fact that we ask our artists what they like to do and what they would like to learn, and that helps shape our artists.

What are your thoughts on Backburner?

Backburner served us well for the past 10 years—not without its problems or limitations—but 10 years and hundreds of thousands of frames is not a bad record! That being said, the program was no longer supported by Autodesk; their support was “don’t use it.” But it worked well enough for a while.

Interestingly enough, Autodesk recently tweaked the program, and when we switched to Max 2012 with the new Backburner we encountered lots of errors. Basically, we couldn’t render anything big or have a long list of jobs in line without Backburner crashing. So we’re testing other options right now, like Deadline, and so far so good. As with plug-ins, we are being very careful not to expect that another program will be the solution to the problem just by installing it. We are thoroughly testing it to make sure we don’t open a new can of worms.

Do you see the industry changing in ways that are promising? Which are the most interesting technological developments that you see coming up for the future?

There are many ways in which we see the industry evolving. The expectations of our clients will continue to rise, which requires us to continue to innovate and evolve our process and products. Things like HD/2K, stereo, immersive presentations, crafted experiences all will become mainstream. In our experience, often the evolution can be unpredictable and happen in fits and starts. We have been expecting real-time tools to become common, but it has not materialized yet in our experience. This seems to be more a matter of the ability to craft a narrative and control the story than quality or ease of use. There are innovations on the technical and artistic side that will allow for more visually rich productions, such as more 3d tools in compositing and paint applications. Better animation integration, from passing animation between applications to better mocap. These advancements will also allow for richer media, better visualizations, and more innovative solutions for clients. And it’s important to remember that even with innovation and evolution there will also always be the need for traditional renderings.

As mentioned before, we also see the industry changing back to make architectural illustration an art form again. The time of the gimmick is over; the time of the “I use this program vs. I use this other program” will be over. Exploring will always be there, and new technologies will arrive, but how much better can a pencil get? There are plenty of them in different sizes, shapes, brands, etc., but they all will work to make a good drawing if you get it in the right hands. Software will soon be invisible and will become the “pencil”, and so you will concentrate on getting it into the right hands to create the best art to help your clients meet their needs.

Introduction

[specs-box]

Bitmap2Material is a software developed by Allegorithmic that allows users to generate seamlessly tiling materials that include normal maps, height maps, specular maps, etc., all derived from a single bitmap image (usually a photo).

The main idea behind Bitmap2Material (B2M) is to speed up shader and texture production, and help users create realistic shaders natively inside their 3D software of choice. Currently 3DS Max 2012, Maya 2012, Flame and Unity 3.4+ are supported.

B2M is a small part of what Allegorithmic offers with its robust Substance toolset. This review briefly touches on some of the other tools as they are related to B2M workflows, but it's mainly focused on B2M for 3DS Max 2012.

Why Bitmap2Material?

If you're an experienced Max user, why would you need to add another software to your toolset for creating shaders? The answer is simple: B2M offers some features that make it a great addition to the 3DS Max pipeline. The most obvious one is, of course, being able to extract multiple maps from a single image. Plugging an image into B2M and having it generate maps for normals, specular, bump, etc., makes it a great time saver. It also eliminates the need of having to go into a secondary software to create these maps. Being able to do it inside the 3DS Max interface provides a smooth and quick way to build your shaders.

Another worthy feature is the ability to generate seamlessly tiled textures. We all have run into the problem at one time or another where you need to tile a texture but upon doing so, you realize you can see the seams. B2M offers a 'MakeItTile' option that offers two methods for removing visible seams. There are various methods out there that allow one to create seamlessly tiled textures, but again, being able to do so right inside Max is great.

The software also offers a great amount of control over your shader. B2M is part of Allegorithmic's Substance toolset, which gains you access to a 'Global Textures Settings' feature. This provides extensive controls for setting the resolution of the texture being used, fine tuning normals, specularity, AO, heightmap, diffuse, etc.

Last but not least, the Substance Player comes as part of the install with B2M. Substance Player is a full 3D viewer that users can use to preview what their shaders will look like before having to build them in 3DS Max. Using this viewer is a quick and easy way to test out textures, and it features the ability to import FBX meshes so you can get a real sense of how the shader will render when used in Max.

User Interface and Customization

B2M uses Allegorithmic's Substance file format, .sbsar, which is the output resulting from Allegorithmic's Substance System, an application framework that allows users to create and export predefined Substances, or shaders for use in 3DS Max, Maya, Flame, and Unity. These Substances can be given custom attributes and parameters that users can adjust once inside their 3D package of choice. B2M is essentially a Substance preset that allows for the creation of normal maps, specular maps, bump maps, etc., from one single image.

B2M's UI and workflow are very intuitive. The software operates in a similar way to other Max maps in the material editor. You start by selecting the Substance map, and then loading in the Bitmap2Material.sbsar file. Once the Bitmap2Material Substance is loaded, inputs show up on your Substance node which allow for connecting to the bitmap of you choice. You just have to bring in the bitmap file and connect it to the Bitmap_Diffuse input on the Substance node.

Putting B2M to the test

Version 1.5 of B2M provides key updates in the area of detail control. With this new version you now have the ability to do things like sharpen edge detail on bump, height, and normal maps, and also use custom maps for bitmap height, and bitmap details. These new features give the user more control for fine tuning.

To test B2M's performance, I started off by creating two rather simple scenes. I approached the first scene as a test in getting used to the UI and the B2M workflow. The second scene was more directed towards pushing B2M's performance to see how much detail it could pull from textures that did not have a great deal of contrast.

For the initial scene, I chose a good candidate to get a quick sense of how well B2M works: a 3.5k image that had a good amount contrast and depth to it. I created a box, applied a material, and then gave it some bump and specularity by plugging the specular, normal, and bump channels outputted by the B2M Substance map into the corresponding map channels. Without any adjustments or fine tuning I hit render and sure enough the normal, bump, and specularity maps were working fine. Not bad for creating these additional maps on the fly.

With the amount of control one has over things like normal recognition angle and smoothing, specular levels, and more, one gets a real sense of how powerful B2M can be. Having the ability to access all these parameters inside Max speeds up your workflow significantly. It allows you to adjust your shaders quickly, and without having to update or change multiple maps one at a time.

For the second scene, I quickly threw together some primitives and lights, and proceeded to build some materials. This is where I wanted to see how B2M would handle pulling bump, normal, height, and specularity maps, from textures that varied in contrast. I decided to use a total of three materials in this scene, each with increasing contrast. For the first I chose a concrete texture which offered little contrast or variation in colour, the second was a rock pathway which gave me some nice variations in colour and contrast to play with, and for the third I chose a dry, cracked dirt texture which has some really nice contrast in it.

For the most part B2M handled these images as I had expected; textures with more contrast and color variations offered the most flexibility when it came to building a more realistic shader. The concrete texture which had little contrast, didn't perform as well and provided me with maps that didn't have as much detail as the other two did. As you can judge from the render on the right above, B2M can add a lot of detail to your shaders simply by utilizing one texture map.

Fine Control

One key area to note, is the fine tuning you have access to when using B2M. There are multiple controls for dialing in various channels like, diffuse, specular, AO, and normals, which allow you to really get in there and create a look to your shader that works best for you. There is also the option of using the height map generated by B2M to drive some displacement which can bring even more detail to your geometry.

MakeItTile

For any of us who have had to create seamless textures, we know it usually involves taking our texture into a program such as Photoshop, creating a new seamless version of that texture, and then bringing it back into our 3d scene.

With B2M, Allegorithmic has eliminated these extra steps. MakeItTile offers two methods for erasing visible seams. An Offset method, which as its name implies offsets the tiles so that the seams don't match up, and a Splat method, which uses the idea of 'patching' your tiled map seams with bits and pieces grabbed from the texture itself.

Both methods yield good results and help eliminate visible seams when repeating your textures. Being able to complete this function in 3DS Max, and on the fly, is a really great feature. If you use a lot of textures that are not tiled by default then I highly suggest giving B2M a look. It saves countless steps of going back and forth between 3DS Max and your image editing program of choice.

Texture Resolution Tools

Another feature that comes in handy when using B2M is the texture resolution or size parameters that enable you to switch resolution on the fly. This allows the user to lower the size of textures by 1/2, 1/4, 1/8, etc., which is useful for preview renders where you don't need to see the full resolution texture. Being able to decrease resolution when needed without having to create proxy textures is a really nice feature. Rendering objects up close could use the full texture resolution, while objects in the back could use a copied material with lower texture resolution.

Substance Player

Finally, let's take a closer look at the Substance Player, which comes as part of the B2M installation. As mentioned earlier, B2M is part of the bigger Substance framework, and that also includes the Substance Player, a complete 3D viewer that allows you to preview what a Substance will look like before you build it. Loading the Bitmap2Material.sbsar file into the Substance Player will create a 3D cube, upon which you can drag and drop your texture onto. Once in the player, you can adjust the normal, the specularity, etc., to get a real sense of how your Substance will look once rendered in 3DS Max. The Substance Player also allows you to export the texture maps created by B2M, which gives you the ability to use them in another package. Another useful tool in case you need to send a set of textures to someone who does not have access to B2M.

Conclusions

B2M offers some really great workflow improvements to the shading process in 3DS Max. Having the ability to create realistic shaders simply by using a single image is enough reason to purchase the tool. The amount of control offered by B2M allows you to fine tune the look of your shaders and make changes to textures and map settings, all without having to leave the Max interface. Additional tools like being able to seamlessly tile textures, and the ability to change the resolution of your textures on the fly are great features that one gets to use frequently.

There are no serious drawbacks to this tool that I could find, though something that could be improved is the UI's speed when swapping out textures or changing texture resolution in the material editor. When performing these tasks the UI can become a bit slow. Also, what may be regarded as an obstacle for adopting the plugin is that it is currently only available for 3DS Max 2012, so users of older versions of 3DS Max are not covered at this time. The documentation seems to have improved since v1.0; it's now easy to read and follow. However, it would be better if one could access the help documents via a compiled HTML file through 3DS Max's 'Help' menu instead of having to browse manually for the PDF file.

One thing that seemed to be missing was the Substance Preset toolbar. When reading through the B2M manual, there is reference to the Substance Bonus Tools which are supposedly included in the installer. After installing twice and poking around the 3DS Max install directory I was unable to locate any scripts that gave me this feature. Not having used this feature, I cannot comment on whether or not your B2M files can be saved as presets for loading at a later time.

That being said, if you are using 3DS Max 2012, I highly recommend Bitmap2Material, as it has proven itself to be a great asset to improve the daily shader workflow. It offers some really great controls and a simple interface for creating very realistic shaders, all at a reasonable price. With the amount of time it saves, and the tools it offers, it is a very good investment for anyone who creates shaders on a regular basis.

Introduction

In our aim to complement the daily news with context and analysis we have the pleasure to present readers with another article that provides a broader look at the patterns and trends of the changing CG ecosystem. In this interview we focus on Mexico, and its place in the Latin American landscape. We talk to Paco Navarro, CEO and co-founder of prominent Guadalajara-based studio Exodo Digital Workshop. Paco has worked in the CG field in Mexico for many years and has been a first-hand witness of the industry's evolution in the region. He shares his experiences of building Exodo, the studio's growth process and his views on the local CG industry's past, present and future.

The Interview

How was Exodo founded and why did you choose that name for the company?

Four of the five founding partners originally worked together at another animation studio in Mexico. That’s were we initially met. But each of us left that place at one point or another for different reasons. We took different paths, some started working at advertising agencies, others at public relations organizations or animation studios. We stayed in touch, though, and at a certain point we decided to create this new company. We called it Exodo (“Exodus”), because the word means “migrating to a better place”, and that’s exactly what was happening to us.

Exodo has grown a lot through the years. How did this growth take place and what changes has it brought to the company?

Exodo started with 3 people, and quickly grew to 5 during the first year. We then hired specialized artists: a concept artist and a modeler (raising the staff to 7). In the third year we managed to break into the market of Mexico City, which brought a big growth to the company, raising the number to 12 people. At the end of that year we produced a series of 16 videos for covering the Football World Cup in South Africa, and grew to 25. Through this past year we reduced the staff to 21 people, because we managed to implement an animation production pipeline that has made us much more efficient, optimizing our production time.

Four of the five initial partners now work as area supervisors, and one of them, who used to dedicate to animating particles and physics simulations, serves as managing director.

Which plugins and scripts for Max do you use frequently?

We use few plugins, mostly for rendering and animation, such as LPM (L Pass Manager script) and Asymmetric Bridge, to move data between applications, because we animate some projects using XSI.

How have customer’s needs evolved since the early days of Exodo?

First and foremost in the complexity of the things they ask for, and also now 90% of the jobs we are getting are related to character animation. This is because we have shifted the focus from a “generalist” studio to one dedicated to character animation.

How has the production process at Exodo changed through the years to meet these requests?

We have put an emphasis in improving the revision processes in the pipeline, because the key aspect of creating a character is developing its personality, and sometimes the agency has a very different idea from the client regarding what the end result should be like. So we’ve created very flexible, non-destructive processes, in order to be able to quickly go back to previous versions and work on sequences that the client has specifically requested.

How has the 3D artist profession evolved in Mexico?

It has evolved a lot. In my particular case, when I started there wasn’t the amount of online learning resources that one can find now. It was a matter of going through books and spending a great amount of hours in front of the computer practicing. We ended being generalists because we had to complete all the different tasks a project required, from beginning to end. But now, with so many schools available, and with all the resources one can find on the net, the 3D professional has changed from someone who does all CG tasks to someone who specializes in a key area of these processes.

Tell us a bit about the latest work produced at Exodo. What challenges did they bring to the studio?

We’ve been working on Zaztun, a short film that will be 4 minutes long (approx.). It’s based on a free interpretation of some Mayan legends. We’ve had this project in our hands for some years, due to it being an internal project that we’ve been working on our “free time”, and this free time has been diminishing each year. The most important goal of the piece is to show the capabilities of the studio in developing photorealistic cinematics.

Time Warrior is a science fiction feature film directed by Joaquin Rodríguez and produced by Jaguar Films’ Mike Halverson. Working on the VFX for this project was a big challenge, because it was the first time we had taken so many shots from a single project. Production took place in Las Vegas, and since it was and independent movie we had to adapt our processes in order to fit the budget and timing estimated by the production company. We’re talking here about more than 130 VFX shots that require from on-site supervision to greenscreens, seamless integration of CG characters with live action footage, as well as some 100% CG scenes involving mocap, dynamics, etc. The biggest challenge in this case was managing the volume of information and processes which were taking place concurrently, and it’s in this area that our company improved a lot. All this work has done in 5 months under the direction of Jaguar Film’s Joaquin Rodríguez and Mike Halverson. We also invited a good friend of ours to participate in this project, Jaime Jasso, who worked as Associate Art Director and Lead Matte Artist.

Plaza del Sol is a job we’re especially fond of. It’s a spot we made for a shopping center here in Guadalajara. We were in charge of doing the redesign for the characters, and last year we managed to create the 20” Christmas spot in just 3 weeks. This was thanks to taking advantage of all we had learned during the production of Time Warrior, and the re-utilization of certain assets that we had from previous versions. This spot is the 3rd one we’ve made for the local shopping center. We also produce the designs for their print campaigns (around 8 campaigns a year, with 3 graphic pieces each).

Z14 is another short film we’re working on. The idea came from a discussion about how illogical it was that an extraterrestrial ship would crash at Rosswell after having traveled many light years to reach Earth. The reply to the question was “shit happens”, and from there we decided to try to find out what could possibly be the most irrelevant cause of why such a thing could take place. The script has gone through various stages, we’re currently on the 3rd version, but the characters remain the same. We intend to turn it into a TV series with various episodes, but for the time being we’re focusing on creating a finished short with the quality we are used to producing.

What is your view of the current state of the CG field in Mexico and Latin America? Where is it headed?

The CG Industry in Mexico is in an embryonic state, we're taking the first steps to get together with other companies in the field and organize a common global strategy. One of the key elements we're aiming for is showing the world that Mexico has the capabilities for making quality productions. We haven't been able to enter the US and Canada markets as much as we could, mainly because local companies and the Mexican CG Industry itself needs to mature. We need to have more artists with quality training.

On the other hand, some individual companies have experienced significant growth. And not only in Mexico. Argentina has managed to attract many projects, as well as Chile and Colombia. But the one who is miles ahead is Brazil, due to the huge size and strength of its internal market.

I believe the Latin American CG Industry is headed towards offering quality work with a new cultural and visual approach, as well as the development of new production methods, due to the fact that companies from these countries are used to coming up with great output that's produced with scarce resources.

What are the advantages of producing CG in Mexico?

The main advantage is the quality vs cost ratio. We can't compete with Asia regarding costs, but we have other clear advantages in comparison, such as our time zone, geographic proximity, connectivity, roads and transportation that facilitate the coordination of projects with various countries.

Regrettably Mexico has had some bad press due to the increase in organized crime, but thanks to the net this can be fortunately put aside, since most of the work is done through online meetings and collaboration. And if there's need, we have the added advantage of being able to take a plane and being in Los Angeles, Miami or Vancouver in a few hours.

What are the things you like most about Mexican culture and Mexico itself?

Mexico is a spectacular country. Creativity and visual richness permeate the whole place. You can find the most creative solutions to all kinds of problems, and this is very valuable when at a certain point one may not have the ideal material, economic or human resources.

Mexican CG workers are very capable, focused and committed. In our sector, there's always a tendency to look for ways to improve and excel. All this helps local institutions, too, since they are pushed by the artists to improve their CG-related courses .

Any funny or interesting anecdotes from your work at Exodo?

Our day to day work at Exodo is a lot of fun. We've organized tournaments of COD, pool, darts, etc. This has allowed us to build a group where people understand each other well. Also, some of our illustrators draw cartoon versions of unexpected situations that take place at the studio. Such as when Paco Zamudio took the habit of disconnecting monitors if they were left on and he became the Monitors' Elf. We've had lots funny situations at Exodo.

What pending debts does the studio have?

To keep growing as a studio and attract challenging projects that make us push the quality bar always higher.

Anything you’d like to mention that we haven’t asked you about?

Exodo is always looking for talent and new projects. So we're open to receiving demo reels and budget requests at any time.

Introduction

[specs-box]

Mudbox 2012 is a digital sculpting software developed by Autodesk, designed to detail, shape and paint objects imported from other applications. It was originally conceived by artists at Weta Digital, while the studio was working on The Lord of the Rings. Mudbox's developers founded Skymatter in New Zealand in 2005, and Autodesk acquired the company's assets in 2007. Since then, the software has continued to evolve, gaining 3D paint tools and a considerably improved toolset.

In the latest 2012 version development has concentrated on improving the 3D painting system, making it more robust and adding UV-less workflow capabilities with the introduction of Ptex support. Posing tools have received new features and the software has been streamlined with many workflow improvements. Another major enhancement is the support for Linux 64-bit, which probably makes it a more desirable choice for Linux-oriented studios and users.

Note: all models illustrating this article have been created by Olli Sorjonen.

Performance Improvements

When put to test Mudbox 2012 seems to have noticeable performance improvements under its hood. Complex objects clearly load faster. They were really cumbersome to load in earlier versions, you sometimes had to wait for minutes before you got a requester saying that the object you were trying to load was a high resolution one, and that loading could only be continued after clicking a button.

In Mudbox 2012 larger objects and textures can be displayed, sculpted and painted more smoothly. Saving speed has also been improved, and there seems to be less file corruption problems.

A new Validate Meshes function helps check the integrity of objects that are imported into Mudbox, so you'll get a better idea if the object will present any problems when you start sculpting or painting on it. In earlier versions of Mudbox it was really annoying to find out late in the process that you had bad geometry: you basically had to take a step back and fix the base mesh or bake the details to another mesh, so the addition of this feature is a very welcome improvement. It would be nice however to have the faulty geometry shown visually (as colored/highlighted vertices and polygons) instead of a list of vertex numbers and other text information. This would allow for finding and fixing the problem more easily.

Editable Stencils

Since 1.x, Mudbox has been able to import images to be used as stencils. In Mudbox 2012 the creation of stencils has been improved a lot: images copied to the clipboard can now be used as stencils and you can also deform the stencil plane with brushes, which makes it much easier to apply facial details or patterns with intricate and continuous textures on your model surfaces, without having to rely on Photoshop to warp and adjust the textures all the time.

Single-Step interoperability with 3ds Max, Maya and Softimage

New in Mudbox 2012 is the ability to use "Send To" in the file menu to send an object to be edited in 3ds Max, Maya or Softimage. Installation of this feature is quite transparent and no manual copying of plugin files or other such setup is required. You can send a new scene to the software in question, update the existing scene, update only the textures and you can also add objects to the current scene. A few export options also provide control on how texture and sculpt layers are handled. Objects can be brought back to Mudbox 2012 with the same ease.

The manual is bit vague regarding the actual functionality of the "Send To" feature, you may have to figure out by yourself what can and can't be done to the model in order to have it preserve the sculpted details properly in Mudbox.

New Brushes and Brush Options

Mudbox 2010 and 2011 added some brushes that tried to catch up with the brush functionality of ZBrush 4, but in Mudbox 2012, instead of adding brushes, both significant and small improvements have been made to the painting and sculpting toolset. You can now apply a single brush stroke across multiple objects and new falloff options have been added - Falloff based on Facing Angle lets you adjust the effect of a brush based on the difference of surface and camera angle, this can be used to adjust behavior of the brush close to the silhouette edges and creases on a object, for example.

Grab brush performance has been improved and mirror limitations in earlier version have been eliminated, according to Autodesk. Two new modes of operation have been added to the Grab brush - You can now use Grab Silhouette to manipulate the silhouette edge of a model and use Follow Path option to pull tentacle-like shapes from an object, very similar to the SnakeHook in ZBrush 4.

Ptex and Create UVs

Mudbox 2012 extends UV mapping-related features in a major way. It now offers UV creation tools and support for a UV-less workflow with Ptex. Developed by Walt Disney Animation Studios, Ptex (Per-Face Texturing) is an open source system that provides the ability to paint textures without UV mapping. Ptex assigns a texture to each base face and no user created UV coordinates are required. The setup for Ptex painting is very easy, you just select Ptex Setup in the Mesh menu and you're ready to go. Not having to go through manual UV creation is especially great for sketching and trying out various concepts.

Ptex textures can be currently exported in native Ptex format, which can then be rendered in software that supports the Ptex format - currently support in Autodesk products includes VrayPtex (3ds Max, Maya), Mankua's Stripes (3ds Max), RenderMan (Maya) and the Maya Hardware Renderer v2.0. As the Ptex format is quite new and lacks support in many commonly-used applications (such as Photoshop), Mudbox can bake Ptex textures into UV coordinates and export them as typical image files.

The new Create UVs option automatically produces UV mapping for an object. The results, however, which are similar to ZBrush's automatic mapping, are not usable for manual painting in Photoshop. No tools exist currently for modifying UV mappings, but in some situations you can use it to avoid manual UV creation outside of Mudbox 2012, which is a great improvement from earlier versions.

Transfer Details

The new Transfer Details tool (in the Mesh menu) lets you transfer properties from a high resolution object to another with different topology, point order and UV mapping. This can be very useful if the topology of the object you are working on has been changed for some reason, so you can use this tool to transfer all the surface details and textures quite easily to an updated object.

Paint Layers and Blend Modes

Combining different layers of paint is now much easier and powerful with the help of over 20 new additional blending modes, all of which are very similar to Photoshop's. Paint layers can now have masks that let you hide and reveal parts of a layer, and you can select a blending mode for the layers, which makes it a lot easier to apply logos, cracks and other details to surfaces in a non-destructive way.

Transparency (Opacity channel) support has also been finally added - transparent surfaces can now be created with the new Opacity channel paint layer and an opacity option has been added to material properties, so you can now easily visualize surfaces such as eye glasses and translucent membranes in your scenes, which is a great improvement for character and vehicle design, among other things.

Improved layer support makes it easier to seamlessly import and export layers between Mudbox 2012 and Photoshop. You can now export multiple paint layers from your scene using the Export All Paint Layers option. The Solo option lets you isolate selected paint layers, automatically toggling off all the other layers. The Solo as Diffuse option, in turn, lets you display selected layers as flat color to make it easier when painting opacity, bump and other layer channels.

However, when you use the Export Channel to PSD command, you can't export a layer's transparency to Photoshop. Luckily, you can export and import layer masks with the Import Layer Mask and Export Selected commands. It would have been a lot easier, though, if you could just export the whole set of layers and masks to Photoshop automatically.

Multiple paint layers can now also be transferred from one object to another using the Transfer Paint Layers option in Extract Texture Maps window.

Large Texture Datasets

Mudbox 2012 improves support for large resolution textures and has better tile management. This lets you work with ease on complex objects such as vehicles or aircraft that might have hundreds of different textures, without being limited by your system's or graphics card's memory.

The Subscription Advantage Pack has added a “Gigatexel Engine”, which enables painting of huge textures in Mudbox. Since this feature is only available to those on subscription and will be accessible by the public at large in Mudbox 2013 it was left out of the scope of this review.

Rebuild Subdivision Levels

The new Rebuild Subdivision Levels tool adds a feature similar to ZBrush's Reconstruct Subdiv, allowing users to recreate lower subdivision levels for imported high resolution subdivision surface objects. Rebuild Subdivision Levels works fast and manages to do pretty much the same as ZBrush's equivalent tool, though sometimes there are problems with polygons on the symmetry axis: Rebuild Subdivision Levels can create a non-mirrored structure on the center line. On a side note, it would be nice to see Autodesk add support for exporting files with a ZBrush-compatible point order, so that you could swap work between the two software easily.

Region, Lasso selections and Freeze

Region and Lasso selection modes have been added. This makes it much easier to use selections and masking during the sculpting process. It used to take a lot of time to paint the Freeze, Mask and Selections in Mudbox 2011 and earlier versions.

Freeze can now be performed using either a sculpted surface layer or painted details. In the first case Freeze will be performed using relative displacement of vertices, in the later case it will be done according to the luminance of the paint layer.

Multiple joints and pose presets

Mudbox used to be seriously lacking in posing tools functionality, as there was none before Mudbox 2011. That version fixed the issue a bit by adding single joint deformation and the ability to import pre-made skeletons. Now it's possible to create multiple joints instead of just one joint, with support for symmetrical joint pair creation. You can also now add joints to imported skeletons.

So, in theory, you can now construct a whole skeleton inside Mudbox, but it's not that straightforward. Getting the hierarchy correct is sometimes tricky, so you may prefer to import skeletons via FBX from 3ds Max and other Autodesk software.

A pose presets feature has also been added, this lets you store different poses and then use them later. Posing is based on weights and is non-destructive. This makes it easy to test your object in the required key poses and also to sculpt hard to reach areas. Sculpting can also be performed while an object is being posed and changes are carried to the other poses correctly.

One thing that could have been automated is the storage of the initial base pose, so that you can return to when needed. Currently you have to this manually.

In any case, the posing improvements are a welcome and great addition. The existing usability issues, however, make the posing work somewhat cumbersome and it's not really a favorite task to perform in Mudbox 2012.

Interface improvements in 2012

Since its early Skymatter days, Mudbox development has focused on usability and ease of use, and Autodesk has stayed true this path.

More recently, Mudbox has been released as part of the Maya Entertainment Creation Suite, and during the last few versions a Maya-style visual appearance and menu functionality has been implemented in the software (probably with the intention to simplify this connection).

Mudbox 2012 UI improvements aren't huge, but some streamlining has been done. When you press the space bar, a "hot box" menu pops up, which allows you to perform quick selection of all main items using icons.

The right button menu has been changed to a new marking menu system, similar to Maya's. These context-sensitive menus offer easy access to various Mudbox tools with less cursor movement, which is especially useful if you use a tablet.

The Object List and Layers menus are still the same they were in 2010 and 2011, actually almost the same as in 1.x. You can't sort, group select or perform any typical mass operations you would expect to be able to do utilizing this kind of list, which is somewhat disappointing and old fashioned.

Another issue that may be bugging some users is the lack of ability to properly save the workspace - if one would like to store open windows, panels and their positions, this is not currently possible. So you have to manually repeat the operations every time you boot up Mudbox 2012. This is especially irritating if you work on a project that requires to have many windows open, like having many ImagePlanes or Stencils property panels open for quick access to visibility. Also, when importing new models to the scene, some "Properties" panels you might have open get "disconnected" and you have to re-open them, which is very annoying.

In addition, some preferences don't get stored properly, they reset every time you restart Mudbox. Camera sensitivity settings are one of such example.

Image planes seem to have quite a few issues - if you change Mudbox's window aspect ratio, the ImagePlanes drift off their place. This is problematic if you're working on a likeness or trying to accurately place details. Also, when you zoom the viewport in, the ImagePlane exhibits erroneous behavior, moving in a way that makes you quickly lose control over it and a restart of Mudbox is needed.

Another probably more irritating issue with ImagePlanes is the lack of ability to re-locate ImagePlane and Stencil files applied to cameras if you've moved them to another location on disk. In this case, placement and everything else is pretty much lost. When the scene is loaded after you have OK'ed all the warnings about missing files, you don't have any filenames or other info left in the properties (unless you've happened to rename your camera to store the image name).

License and price issues

As CG work is a quite global business, it's somewhat annoying, especially for freelancers and small companies, that Autodesk offers different pricing based on regions around the world.

The result is that Mudbox 2012 costs a lot more to purchase in Europe than in the USA, for example. The difference in price is actually quite big - at time of writing, a full license costs $601 (equals to 464€) when bought from Amazon.com and the cheapest price found from Amazon.co.uk is £829 (993€). The difference is over 520 euros between European and USA users. It is even more glaring if one would buy the license from a local reseller, a quick check showed that the prices start from 1000€ and go upwards. When compared to the pricing policies of ZBrush (Pixologic) and 3D-Coat (Pilgway), this feels somewhat harsh.

Conclusions

The addition of Ptex support, improved posing and painting features already make version 2012 an upgrade worth purchasing. Even though the upgrade costs almost as much as ZBrush and 3D-Coat combined, Mudbox 2012 has some features that make it hard to be without. The proper perspective in viewport, plus the simplicity of the user interface make it a good candidate for being the primary 3D sculpting and painting application.

Good and easy interoperability with 3ds Max, Maya, Softimage and Photoshop is another strong area that makes Mudbox 2012 stand out from the competition. Also, the Texture Extraction and Transfer Details tools are top notch quality and have features that cannot be performed in the competing software with the same ease.

However, some bugs and known issues have been lurking in Mudbox for ages, and they still haven't been fixed. Also, it's noticeable that Autodesk has adopted a policy of more likely documenting a "known issue" than fixing it - in the latest service pack, which at the time of this writing was SP3, the release notes document lists one fixed bug, but the limitations and notes list contains roughly 15 issues, and in addition there are some other known limitations listed later in the document. Not to mention the numerous undocumented ones. Some bugs have also crept into "Export Selection", as the OBJ format export is a lot slower in 2012 than it was in 2011. The "Send to" command works fast, but some people might prefer or need the traditional export method for various reasons, using ZBrush might not be the least among them.

Also, when it comes to features, Mudbox 2012 has always lacked the ability to create base meshes or new geometry, so you have to import them from outside. The selection of pre-made meshes really doesn't help either, and makes one wonder why there's a tree stump and other seemingly random templates, instead of some more useful basic geometric shapes.

With the release of Zbrush 4 R2, when it comes to geometry creation, the already existing gap between Mudbox and ZBrush has grown bigger, as ZBrush already had many tools to generate new geometry and now the added DynaMesh makes it even better. Hopefully this will be addressed by Autodesk in the next version of Mudbox, as it would improve vastly Mudbox's usability for prototyping and sketching.

Introduction

I don't remember the exact circumstances, but sometime in late 2009 I got word that there was this interesting new movie being prepared in Norway. It was a mockumentary about trolls titled The Troll Hunter, written by a debut director, André Øvredal, and it would be the first time a Norwegian movie would need photo-realistic 3D characters. This was surprising since VFX-heavy movies are rare up here, and it was made more exciting since it concerned our cultural heritage. I mean, how often is there a chance to get paid to do big trolls? Even though I have wanted to move away from VFX work, I knew this was a project I had to be part of!

Through a fair bit of luck and lobbying (begging on knees), I eventually found myself bidding on the VFX work. There were three big troll sequences in the movie, and the production wanted separate companies handling each sequence. I eventually won the bid for the so-called Ringlefinch sequence, about a predator troll living under a bridge, who is confronted by the Troll Hunter. I suggested I could help with the character design, and ended up doing the final design on the Ringlefinch and another troll in the movie (The Mountain King).

The Challenge

The Ringlefinch sequence was the simplest VFX sequence in the film, but it had some challenging aspects to it. Even though it was only ten shots, the longest shot was 900 frames long. It was photographed hand-held at night, with some zooming (which is hard to track). There was some interaction, the troll was going to eat a sheep and try to eat the troll hunter. And that meant I had to do two digitals stunts as well, so there were three photoreal characters in total.

I quickly figured out I couldn't do this all by myself, and I didn't want to either! For the animation, I hired Atle Blakseth. He is the fastest animator I know, and he actually ended up doing animation for the other companies as well. We did not share office, so I had to come up with some kind of pipeline that enabled parts of the project to be kept online.

And the final challenge was the software. My base package is 3ds Max, which is a great all-round 3D application, but out of the box it isn't the ideal software for doing team-based animation work. There is no easy way to reference animation rigs, so if you want to change the underlying rig while keeping the animation, you have a problem. And 3ds Max has no text-based file format where you can build multiple scenes from a collection of assets. These two issues force you into a very linear workflow. Also, animators really seem to hate 3ds Max! (If you've ever used the fast rigs and the excellent graph editor in Maya, you probably understand why.) But 3ds Max has a lot of good things going for it; it has an incredible selection of plugins and scripts, the wonderfully flexible stack, and a high-level scripting language which makes for quick and easy development.

A minimalistic 3DS Max creature pipeline

Fighting myself out of a linear workflow was the most important thing for this project. I did not want to end up with a situation where the director might want a change on the character, and we had to scrap both the rig and the animation just because, say, there was some topology change on the arms. That wouldn't be very professional! Also, my animator kept needing extra features in the rig while we were animating (like FK/IK switching that actually worked, ho-hum!), another reason to get some kind of reference system working.

De-coupling the rig and the animation is something I have wanted to build into my rigging system. I have a set of maxscripts that I use to build and animate my animation rigs. It all starts with loading a helper template that I fit into my character, and that the rig is built around. The only parts of the rig that can be animated are custom attributes on spline objects suffixed with "_CTL", nothing else. Knowing that, it's very easy to find the animation data in the scene. For The Troll Hunter, I created a simple text-based animation file format that would store the keys that are put on these controller objects. I then made a window where I could save out all the animation in a scene, character or single controller, clear everything - and then load it back in. Expanding from that, I wrote a script that collected new versions of all assets in a scene, found the most recent animation, and then saved an updated file for the animator. This was a perfect little pipeline, and was exactly what we needed to be able to update the rigs as we went along.

I have used Dropbox on smaller advertising jobs, and we used it to keep all files related to the animation online. Now the animator could work together with the director in his office, while I sat at home grabbing animation files as they were ready. Dropbox does this neat thing where it automatically backups deleted and overwritten files, which was a nice safety for us.

Troll Hunter images © Filmkameratene 2010

A tools greenhouse

As mentioned, I have my own simple rigging system that I keep expanding on every character project I do. I also have a script collection that I call the Super Tools that I call my "tools greenhouse" - as they will keep growing as long as I find time to care for them!

I thought I'd write about some of the scripts that have helped me the most.

One of my favourite scripts is called Super Morph Splitter. It takes a morph target, and splits it into a left and right morph target using a custom distance value. When these two left and right morphs are combined, they will look like the original morph target. I think most people would agree that it's much easier to create a symmetrical face expression. So with this script I can model a nice looking smile, but then let the computer figure out how it should look separated into right and left sides.

Another handy script I've programmed is the Super Attach to Closest Bone. This script breaks an object apart, and parents each polygon to the bone that is closest to them. I started rigging for the troll long before I modeled it, and with this script I could test out the rig on my high-resolution Mudbox model months before I did the final troll topology. I was surprised that the script created a better bind result than the 3ds Max skin modifier, especially on fingers. I'm interested in transfering the math into a skinning tool in the future.

And finally I thought I'd mention Super Attach to Surface. This script takes a selection of objects, finds out the nearest point on a mesh, and then locks the object to that point on the mesh. I used this script to attach the bushes and rocks to the back of the troll.

I intend to release most of these scripts to the public at some point, but I need some new character jobs to be able to make them presentable. I hope that will happen sometime next year.

Building and detailing the assets

I started building and rigging the stunt sheep, or Shaun as he is called. He was my guinea pig for testing out the animation workflow on the troll, what hair sim to use and how to light and render the scenes. He and the troll were built using what I guess is the most common process at the moment. First a quick mockup was made in Mudbox, this model was then re-topologized using Polyboost (now built into Max) and Wrapit. I then did the UVs in Unwrella, before more detailed sculpting and texture painting was done in Mudbox. I painted several paint layers and masks, and combined these in Nuke to make the final textures for the characters. I set this up in Nuke so that each UV tile was a separate frame.

This was the first time I had to combine everything I've done in 3D into one project, and make my most detailed character to date. Even though this was a feature film, I had to keep things manageable for one person. My philosophy is that every polygon adds up, and that extra polygons will lead to more time being spent doing UV's, building morph targets, skinning and simulation. So I kept the polygon count extremely low on these objects. The troll was around 3400 polygons, with about two thirds of those polygons in the face. (I did 3D-models for video games in the early nineties, this was the usual polygon count back then.) The poly count turned out to be a little too low when I started doing muscle sims, but for the most part it was fine.

The troll hunter in armor was modeled by Atle, and textured using photos taken on the set.

Iterating through the shots

My animator started animating as soon as the rigs and rough versions of the characters were ready. I then detailed the characters and worked on light and rendering in parallel to the animation, giving the animator updated characters and scenes as I went along.

Halfway through the project, my simple pipeline started taking shape. Here's how it worked:

1. The scene builder script will collect the latest versions of all animation assets, and build a maxfile for my animator, which is then put onto Dropbox.

2. When a new version of the animation is ready, the animation is simultaneously exported back to the scene builder, and baked out for the next step.

3. Muscle simulation is done on a simple muscle rig, where the muscles affect the baked version of the characters. The resulting simulated mesh is baked once more.

4. The render scenes are put together using the final textured and displaced character, with baked animation from previous step. The hair is simulated in this step, and the character is lit and rendered in multiple passes.

Simulation

One of the many benefits about being small, is that you can quickly change your way of working, and play with different software solutions on every project. I try not be too attached to my software, and I'm always looking for something that makes me solve things faster and with higher quality. And I'm always happy to find procedural solutions that also look good. Good-looking procedural animation is particularly great, since it gives you "free" and realistic looking animation.

On this project I wanted to see if I could find a better solution for hair, I've found the built-in solution in 3ds Max and almost all hair plugins to be both time consuming to setup and unstable to simulate. Hair Farm had just been released, and proved to be quicker and better than every other solution I have tested. The simulation part was especially good, with a gravity-compensating feature that lets you do hair simulations with no run-up frames.

Muscles was something I didn't think I would find time to do, but turned out to be a lot less setup work than I thought. I used SkinFX to build the muscle geometry, and SkinFX has a skin jiggle feature that I used on the troll belly. SkinFX doesn't do muscle dynamics, so those were done with the Flex modifier built into 3ds Max. The great thing about Flex is that it is real-time (something I wish more of the 3ds Max simulation stuff would be, simulations should challenge your creativity, not your patience), but it can be unreliable. If the character is running fast, the muscles could be hanging a metre behind him! Also, I never got Flex collisions working reliably, so I couldn't have the muscles stopping when hitting bone, which meant muscles would suddenly move too much into the body. I ended up compensating for these two issues by keyframing the flex influence by hand - which would have been totally unacceptable if there had been double the amount of shots.

I also set up some cloth simulation for all the belts, straps and mud flaps on the troll hunter armor. These were simple polygon strips, simulated with Max cloth. The detailed render-mesh was skin wrapped to these strips.

There was also two smaller simulation issues that had to be solved. I used Rayfire for breaking apart the bridge as the troll hand reaches for the sheep. That was only visible for a couple of frames, and perhaps overkill to do. But it took only an hour or two to setup in Rayfire, and it would probably had taken just as long modeling the bridge debris by hand. I used Particle Flow to blow up the Ringlefinch at the end of the sequence. I emitted chunks of meat and a spray of blood from a troll-sized plane under the bridge. The meat chunks was made from an instanced sphere with a noise deformer. When the chunks hit the water, they spawn flat planes with video footage of splashes. That turned out to be amazingly convincing, even though it was set up very quickly!

Troll Hunter images © Filmkameratene 2010

Rendering

The Brazil rendering system has been my renderer of choice for ten years now. It's an extremely fast raytracer, and is very efficient with multi-million poly scenes. GI-rendering was the best solution for this project, since most of the light hitting the camera is bounced. And I knew I would rather render it properly instead of spending days faking a spotlight based lighting setup. But I was unsure of how I could render hair and GI in one pass, since there is no hair rendering primitive in Brazil. I tried two other renderers, and while both could do hair and GI, none could match the speed and stability of Brazil. My solution would be to render everything except the hair, and then have shadow casting polygon hair that was invisible to the camera. The hair was then rendered separately using scanline, with a shadow pass that let me integrate the two.

Also, I found that the combination of GI, area lights and scattering was heavy on the renderer. To speed things up, I rendered a non-bounced pass with scattering. I then rendered the bounced light with simple white materials, without scattering. Finally in compositing, I multiplied this white render with a flat textured render, and added it to the non-bounced pass. The result was impossible to distinguish from a full beauty render, and all passes rendered in only fifteen minutes per frame.

This might sound complex, but it was really fast and simple to set up. Kudos to RPManager for making pass setups like these an easy thing to improvise.

I should also mention that Dave Baker (of Maxplugins.de) created two shaders for this project. He made a shader that rendered XYZ coordinates as RGB colours, and a shader that combined two materials using the lighten method from Photoshop (this way of combining materials works great for sub-surface scattering).

Bringing it all together

Compositing is the carrot at the end of the stick. After all the tiresome modeling, texturing and rigging, there's finally a part of the project where changes can be implemented at the speed of thought. Doing the final compositing is my second favourite part of any project, right after design and conceptualizing. Each shot had about 15-20 passes that had to come together in compositing, most of them dealing with separating key light from bounce, and integrating of the separately rendered hair. I also rendered out a bounce of helper passes and masks, usually three passes in one render split into RGB. For example, one pass had sharp speculars in the red channel, soft speculars in the green channel and the reflection intensity in the blue channel.

The Troll Hunter was shot on digital video, and matching the lens distortion, grain structure and ccd characteristics was a bit tricky. The lens distortion was analyzed with Nuke and the same distortion applied to the 3D elements. I ended up pulling out the grain, and then replacing it with grain that looked very similar (there was even more grain put onto the film in mastering, so there went my finely tuned imitation!). And I analyzed each RGB channel, and matched their respective glow and offset characteristics. These things can't be seen unless you pixel peep onto the full resolution frames, but if they were missing you would feel that something was wrong. Depth of field and motion blur was also done in Nuke, carefully tweaked to match the results from the camera and lens combination that shot the scene.

Return of the Ringlefinch!

That could have been the end of my work on the Ringlefinch, but there was one little thing I just had to do. Ever since I was told that the troll was missing one arm, I kept imagining him fighting with Darth Vader, and getting his arm chopped off, like Luke Skywalker did in the famous sequence in The Empire Strikes Back. I knew that I just had to make that scene at one point. I found time in-between projects, and managed to do this fake teaser that I brought with me at the End User Event this year.

Now this was all the "fun" of the Troll Hunter, only 100 times quicker! Here is a quick run-through of how it was made: I modeled a simple set based on blueprints from Empire Strikes Back, and eye-matched the cameras in each shot to that model. The animation was then rotoscoped (and this time I did it myself), by copying the extremes in each movement, and then carefully interpolating the rest to match. I added a 50 frame buffer in between each shot for the simulations to ease in and out of each animated shot. Then the troll hair and floppy bits on the armor was simulated. I could not do muscle simulation on this version, unfortunately. I had to scale the troll down a bit, and that messed up the flex simulations.

After that I picked the cleanest frame from each shot of the film, and cloned away the original characters, in order to comp the CG elements as replacements - I did that after animating the characters, so that I knew exactly where to focus my retouching. I set up the light in each shot matching the specular highlights on Darth Vader's helmet in the original movie, and did successive renders where I kept improving the placement and intensity of the lights. For passes, I changed my workflow and converted the hair to polys and rendered almost everything in one brutal beauty pass. The only other pass was a separate one for the lightsabers, since they needed more accurate motion blur. The reason why I could raytrace hair this time was because I turned off GI on the teaser. I could have rendered with GI, but then I couldn't have reached my one-weekend render deadline. On the next project I want to combine hair and GI though.

Finally, everything was comped together in Nuke, the grain and (minor) camera moves were put back in, and the teaser was done! As a last minute change I also flipped the images to be in continuity with Empire Strikes Back. I'm not sure if I should have done that, since people keep commenting on my "mistake" of messing up which arm was chopped off...

Afterword

Congratulations if you made it this far! I hope you found this to be an inspiring read. The Troll Hunter was a landmark project for my portfolio, and one of the few projects that have let me use the full range of my skillset. I feel privileged to have worked on the film, and I wish every project could be like this. I get a kick reading about directors like Gareth Edwards, who not only writes, photographs and directs the feature film Monsters, but then goes ahead and does all the VFX work as well! I would love to do something similar in the future.

It's pretty cool that software and hardware today can let a small group of artists do realistic 3D that 15 years ago would have been the domain of big VFX companies. Who can dare imagine what can be done 15 years from now?

Introduction

[specs-box] Anima is a stand-alone crowd simulation software developed by AXYZ-design. It's intended for exterior and interior architectural visualization, and it helps bring more dynamism and liveliness to scenes that usually consist of mostly static objects with camera movement. Anima focuses on creating big masses of animated people and provides various tools for the task.

The software comes with plugins for 3DS Max and C4D that integrate and import projects created with Anima into these 3D applications.

First Impressions

The software has a small learning curve. After quick exploration to understand the basics, a user manual helps one get through the first learning phase and takes less than a couple of hours to complete. The interface is quite easy to learn, though viewport navigation requires a bit of getting used to - zoom in/out controls and the movement inside the viewport react a little bit too fast, and currently there's no way to change this. It would be nice also to have an option to customize the interface as well as to create your own shortcuts, or maybe load familiar presets for major 3D applications.

Creating Crowds

The first step is to import a basic 3D model of a structure. In this case it was an OBJ format model, which was initially exported from Max. Anima can read OBJ files, as well as other well-known formats such as 3DS and LWO.

The next step is to create an actor's path. To do that there are 4 different tools available:

1. Spline Path - defines a path where people will be located and walk on.

2. Stairs Path - defines a stairs path that actors will use to go up or down.

3. Fixed Path - spreads actors around a surface who will stand at one point without walking, with a choice of several basic poses such as sitting on a chair, leaning on a wall, waiting in line etc.

4. Escalator Path - defines a path of animated stairs that actors will use to go up or down.

An interesting concept in Anima is that there is no ability to control the Z axis, something that ends up revealing as a big advantage. The paths are created in the scene only in the X and Y axes, while the height is determined automatically by the imported 3D model. This is a great time-saver, there is no need to adjust the spline and lay it down exactly on the surface; it's all done automatically by snapping according to the model's height. When creating the stairs or the escalators, their height is also determined by the surface they are created on.

After building and defining the path/stairs/escalator comes the part of adding the actors that will move through them. Adding an actor is easy, all there is to do is to drag and drop the desired one on the path - that's it. The actor starts walking immediately along the path. You can add any number of actors to this path and it will update in real-time.

Anima comes with only 2 actors, but their outfit can be modified and randomized with just a few clicks. This way you can get a good number of actors that look different. Having 2 actors could be more than enough for long and medium shots where you won't notice the similarity between them. It's a different story though in closer shots. In this later case you'll probably need to expand your actor's library, which can be done by buying actors from the AXYZ store.

The animation of actors walking, moving on the stairs and on the escalators is very convincing. Climbing or getting off from the escalators looks great, too.

Anima also allows you to define various animated actions for each actor, which can take place while the actor is either walking or standing in his place. These actions include smoking, talking, waving his/her hands and many more. The actions add more variety to the overall feel of the scene if you are using many actors in your simulation.

An impressive feature in Anima is the "clone" modifier. It lets you create a great number of actors on a single path very quickly. In just a few seconds you can have a crowd of people walking on your scene. After creating the spline path and adding, let's say, 3 actors to that path, the "clone" modifier is selected. Then the overall amount of actors is set, something like for example, 200 people. Anima calculates and creates a crowd of people based on the 3 actors chosen earlier. The animation of each actor and the color of their outfits can be randomized. No less impressive is the quick calculation and response time with this amount of actors – scrolling the timeline slide bar or just pressing play makes these 200 people walk smoothly and in real-time.

Another impressive feature is the Crowd Avoidance Engine. When this feature is turned on, Anima calculates the course of each actor and when two or more actors gets very close to each other it automatically changes each course in order to avoid collision. This calculation is automatic and within seconds it's updated in the viewport, depending on the amount of actors you are using.

Anima lets the user define groups that consist of paths and actors, to help work in an organized way. You can give names to each group and turn them on/off. But the real advantage of this feature is that these groups are also translated into 3DS Max as layers and dummy objects, which can be very helpful with heavy-weight projects.

Anima is ideal for working in long and medium shots. In tests we did at the studio the quality of the people in terms of models, textures and animation is excellent. They look very realistic and in long shots they can be easily confused with live action footage. On the other hand, in closer shots the realism decreases. You can increase the quality of the textures to be high resolution for close up shots, but you still won't reach the live action realism you got in the long shots. Animation is also great in close shots, but again, it naturally looks less good than it did in wider shots.

Working with 3DS Max

In order to import Anima projects into Max, one must first install the required plugin. After that you can import and work on the original .ani file that was created with Anima. Even though you can't modify or make changes to the paths or the location of the actors, integration works well. You can go back to Anima at any time, make changes, save the project, and refresh it in 3DS Max.

When importing the scene into Max, the actors are located exactly at the right point, on top of the original Max model exported earlier as an OBJ file. No further adjustments are needed. Along with the actors, their texture and materials are also imported automatically.

As opposed to Anima, scrolling the timeline slide bar in Max is slower and requires a lot of RAM, depending on the amount of actors. Performance varied when testing on different machines. In one case, when using a great number of actors it was impossible to work, due to the amount of freezes experienced. Only after removing actors' textures one could work freely and scroll the timeline smoothly. However, when testing on a different computer with 300 actors, including textures, things ran perfectly smooth. According to the developers the variation in performance depends on the video card's power, and is related to how Max manages texture maps, because it unloads much of its work on the video card's memory and processing. However, there's an easy workaround in case one has an older video card: the 'display properties' option allows you to display the actors in the viewport as boxes. This worked great and solved the issue every time.

Conclusion

Anima provides a fantastic solution for adding animated people to 3D scenes, a task that was previously avoided at our studio due to the huge complexity it involves. The tools are easy to learn and the real-time working environment helps make the creation process spontaneous and productive. It's a pleasure to see that there is finally an easy way to create quality 3D animated people and crowd simulation behavior, something that had been missing from the architectural field for too long. We are definitely going to use Anima in upcoming projects at Studio Aiko.

Introduction

[specs-box]

Topogun is a standalone retopologization and map baking application developed by PixelMachine SRL. Topogun's purpose is to make retopologization and resurfacing easy, and to enable users to bake normal and other maps from high resolution 3D meshes on lower resolution versions, in order to mimic the appearance and features of the original mesh.

Version 2.0 improves vastly on the previous 1.x version and adds many new features - including better symmetry support and new intuitive tools that help create geometry faster.

It should be noted that the version tested in this review is not the final 2.0 version, but a beta, so there might be some feature limitations or performance issues that will not exist in the final version. The current beta is, however, readily functional and available to those who purchase Topogun.

Why Topogun?

Why would you choose Topogun for retopologization or map baking instead of doing the task in a more traditional modeling software? The answer seems obvious - the purpose-specific tool should, in theory, perform better as it lacks the possible overhead a generic tool has.

Topogun offers almost a full pipeline for retopologization and map baking, with the exception of UV creation tools (which might not be a minor issue for some users). You can also use Topogun 2.0 for retopologizing your concept sculpt and then use subdivision tools to create a hi-res, wrapped mesh that captures the details automatically and then continue sculpting on the cleaned up hi-res mesh in your sculpting application.

After you're finished with the sculpted hi-res mesh, you can build a low resolution mesh in Topogun 2.0 for real-time use if needed, go outside of Topogun to UV unwrap it, then get back to Topogun, and bake all maps, which can then be taken to Photoshop or other painting application for creating the finished textures.

Supported Platforms

Topogun 2.0 is a standalone application and when ready, according to information on Topogun's website, will be available for Windows, Mac OS and Linux operating systems. Licenses are cross-platform so after acquiring a license, the software can be used on any of the three operating systems - this is great if you work on a few different platforms.

The software is available in both 32-bit and 64-bit versions. Most users will likely turn towards the 64-bit version which allows access to much more memory and increased performance. Topogun utilizes multi-threading, this is put into action in baking and subdivision modules. According to the author, the modules are fully multi-threaded and up to 256 CPU cores are supported.

File Format Support

Topogun uses an XML-based format for saving its own files and scenes; for importing objects Topogun supports .OBJ and also .PLY file formats, which includes support for per-vertex color information. The PLY format is used in various software that handle 3D scanned data (Geomagic's products for example). Retopologized objects can be exported in .OBJ format.

All the .OBJ files that were tried out for this review loaded fine, with correct smoothing groups, scale and everything as expected. Objects exported from 3ds Max, ZBrush and Mudbox seemed to load without any problems.

User Interface and Customization

The first impression is that the UI is pretty fluid and fast to use, there is no lagging or sluggish rendering when UI items are moved and activated.

Most of the on screen buttons are classic, subdued, professional looking, containing mostly text. A slightly longer tool tip description is also shown when you float the mouse cursor over buttons. Topogun supports a few different viewport navigation models, so this may help you feel at home a bit more easily. Menus are clearly laid out in a very logical fashion, most of the commands could be found easily right from the start.

Many UI items can be customized: keyboard shortcuts, viewport colors, the Topogun skin (though Topogun 2.0 beta only had a “Default” skin), you can even drag buttons on top of the viewport area, where they'll stay floating.

General preferences include many options - vertex snap settings, mouse interaction models (which simulate viewport navigation styles found in Max and other major packages), render settings, camera characteristics and sensitivity, etc.

Realtime Viewport

When a reference mesh is loaded, by default the lighting is cast directly from the camera. The default color applied to the object is a neutral grey and the overall appearance of the viewport is professional, with no irritatingly imposed style and feel of the software.

You can zoom and move very close to the reference mesh without any glitches or rendering errors - this is great compared to many software where you have all sorts of minor or major camera visualization problems like not-so-well defined near clipping planes or rendering artifacts.

Reference meshes with high polygon counts loaded fine. The performance is actually excellent, as all tested objects ranging from modest 300k to 21 million polygons loaded well and functioned without bogging down the system or the software. This thankfully leaves out the need to perform time-consuming polygon reduction on objects often before importing them.

The Options menu contains typical display options (wireframe, backface culling, show normals, etc). In addition to these you can also show vertex colors, use a see-through mode and colorize mesh groups. It's also possible to highlight border vertices and edges with different rendering styles.

The View menu in turn offers visibility options for framing selected or all objects to viewport and options to toggle on / off visibility of the polygonal geometry and the edges (net) and reference mesh. A floating group window is available for toggling visibility of different mesh groups. Floating logging and history windows are also accessed via the view menu.

Tools

Subdivision Surface Support

Topogun has a Subdivision tool that can be used to wrap a hi-res subdivided, retopologized mesh, around the original reference mesh in order to capture its details. The Subdivision extension can be used to generate intermediate subdivision levels for use in Mudbox or ZBrush, with compatible point order.

This subdivision feature is especially useful if your workflow includes the creation of a concept sculpt, which you need to retopologize before working on detailing.

Texture Baking

Topogun supports one pass map baking - instead of baking one map at a time, Topogun bakes all the selected map types in a single pass (similar to 3ds Max map baking). This is a good time-saver.

Topogun can bake diffuse, normal, displacement, hardware ambient occlusion (AO), software AO and cavity maps. Both 8-bit and 32-bit floating point displacement maps are supported. These are saved in .tiff format.

Hardware AO maps are generated with the help of the GPU. This is a much faster way than the traditional method of using Mental Ray or V-Ray.

One interesting bake feature is the AO dummy object. It allows you to load an occluding object for casting AO shadows on another object, but which won't be taken into account when generating other maps. This is especially useful if you want to have occlusion baked for objects such as a hat, or some other interchangeable item for a game model.

Another nice feature is “bake group colors”. It enables you to bake a map that uses randomly assigned colors to mark different grouped mesh elements, such as eyes, clothes or any separate parts you may have. These maps can then be used in Photoshop for color-based selections to easily isolate mesh parts on baked maps.

One of the caveats of Topogun is that it doesn't support creation of UV coordinates, so you have to go outside of Topogun to UV unwrap the retopologized mesh.

You can't perform polygon structure altering operations either (i.e. add or remove details after you have UV unwrapped the mesh), but you can modify the topology by moving the vertices, if you need to do last minute adjustments before baking.

SimpleEdit

SimpleEdit and SimpleCreate are the original Topogun geometry creation tools.

SimpleEdit can be used to manipulate scene geometry components in several different modes. Faces, edges and vertices can be manipulated one by one or in grouped selections, using the transform manipulator.

The tool enables user to quickly select edge rings and loops. You can also insert edge loops interactively with realtime feedback, remove edge loops and also collapse edge rings. Likewise, vertices can also be merged.

SimpleCreate

Simplecreate in turn can be used to create edges and vertices on top of the reference mesh. When Topogun detects a possible tri / quad polygon face, it will be created. Simplecreate can also be used to split existing edges.

One noteworthy fact is that Topogun works only with tris and quads, so you have to take this into consideration if you're used to creating many N-gons and then splitting them after you've decided where to have the edges.

Both SimpleEdit and SimpleCreate seem to have some slightly unintuitive aspects to their usage. It would be helpful for the user if the most important shortcut keys and functions available for these tools could be shown somewhere on the screen when active. Some of their "hidden" features are not so obvious, so one may only find out about them by reading the manual carefully.

Also, some operations require you to swap between these two tools, such as, for example, creating a detached edge. This may initially be very confusing, unless you've realized that the right mouse button (and C / T keys) can be used to swap these tools.

Bridge and Draw

The Bridge and Draw tools (which were available in Topogun 1.0) now support symmetry. Bridge is a fast way to fill holes between edges. The Draw tool allows you to use free-form sketching to produce geometry in a few different ways.

Draw offers four (4) different drawing modes - Line, Loft, Eraser and Sweep mode. With the new Sweep mode, you sketch a line along the length of a part of your reference mesh, like a leg or an arm, and then just define the number of steps. You get a swept geometry wrapped around the specific part of your reference mesh.

New Tools in Topogun 2.0

Brush

The new brush tool lets you modify the mesh using a soft selection, and it also works with symmetry. You can modify the size, strength and falloff of the brush. You can limit a brush to selected vertices and also restrict its effect on border zones. This makes smooth adjustments inside a garment, for example, very easy.

The Brush tool has seven (7) different working modes - Move, Relax, Inflate, Standard, SProject, NProject and Select. These offer various translation, fine-tuning and projection methods.

The most conventional ones are the Relax and Inflate modes. The Relax mode relaxes vertices and does reprojection. The Shift key can be used to quickly access relax in a similar way as smoothing in ZBrush and Mudbox. The Inflate mode helps you fix the intersection of low resolution geometry created on top of the hi-res reference mesh.

The Brush tool has built-in support for pressure sensitive tablets. You can adjust its settings and you can also switch it off completely if needed, which is great.

After using this tool for a while, the brush feels pretty fluid and has an overall nice feel to it, as the refresh rate of the brush tip is good and movement feels soft.

Morpher

The Morpher extension can be used to either make last moment fine-tuning changes to a mesh, or create major shape changes to the low polygon retopologized geometry. Thus, it can be used to correct the pose of a character relative to a reference mesh or it can be used to create low resolution morph targets that match the changed shape of a hi-res reference mesh.

Symmetry Planes

Topogun 2.0 offers quite comprehensive symmetry support - typical standardized symmetry planes exist as default and you can define custom symmetry planes.

Since this tool is clearly supposed to speed up workflow, it would have been nice to be able to manually adjust the symmetry plane using a transform gizmo instead of using numeric inputs or the align function.

Symmetry Warps

The Symmetry Warp tool can be used to retopologize slightly asymmetrical objects. This can be achieved with local space deformers of sorts that you can visually place on the surface of the reference mesh. You start by placing a red manipulator circle on one side of mesh, and then you adjust a blue manipulator on the other side to define the asymmetry. You can create multiple symmetry warps, which will be saved along with the scene.

You can place the symmetry warp on either side of the symmetry plane, which gives you more flexibility. Symmetry Warp tool works pretty neatly, setup is quite straightforward and the end result is predictable.

Tube

Another tool added to Topogun is the new Tube tool, it can be used when you have clearly tubular shapes in your reference mesh. It allows you to create cross sections and then produce a mesh.

The Tube tool works pretty well and is easy to use. The tube creation happens with OK speed and accuracy, and somewhat random and lazy placement of the cross sections on complex detailed surfaces doesn't cause problems and clean geometry is produced.

Again, a contextual summary of the tool's usage placed in the bottom bar could have helped, as initially without reading the manual it may not be obvious how to finish the tube creation process - there's no "create" button on the screen, you have to click the right mouse button.

Extrude

The new Extrude tool has similarities with SimpleCreate. It has three (3) different working modes - Crawl, Project Back and Free. In short, it's a tool designed to extend open edges, to quickly cover the reference mesh surface with a uniform mesh structure.

It's now much more easy to retopologize cylindrical shapes like arms, legs and fingers, when compared to Topogun 1.0.

The Free mode extrudes the newly created polygons without snapping them on to the reference mesh. If you want to project the new vertices directly onto the surface, you have to use the Crawl mode.

The Crawl mode lays edges you extrude on top of the reference mesh's surface and it also works with symmetry.

The Project Back mode is a blend between the Crawl and Free modes - it shows the extrusion mesh growing in free air and when you let the extrusion off, it projects back to the surface.

Conclusions

Topogun 2.0 is a professional grade tool for retopologization and map baking. It is very easy to use and adapt to. Artists who know what they are looking for will find many useful tweaks and features hidden under the simple and clean UI.

The software offers quite a few good tools for working with organic and mechanical subjects and it can handle very high resolution reference meshes. The new mesh creation tools and improved symmetry support boost Topogun 2.0's workflow a lot. Baking also works very fast and has enough options for fine tuning the end result.

Minor UI improvements would make Topogun 2.0 even better - button customization and generally the lack of an aid for learning keyboard shortcuts in Topogun 2.0 is probably something that could be addressed easily.

Considering the current feature set, automatic retopologization (something similar to 3D-Coat's Autopo or ZBrush's ReMesh All / Unified Skin) and UV unwrapping features would be a great addition to the next installment of the software.