CGarchitect’s 2020 archviz render engine survey
Sep 22, 2020 by CGPress Staff
18
|
CGarchitect has published the latest results of its illuminating annual render engine survey focused on trends in the archviz industry. Read it on CGarchitect.
I know it’s a poll and not actual market share. But if it is an indicator, no wonder Chaos Group feels it can jack up prices and shorten the time span between major version releases as much as it has without fear of much repercussion. More revenue with just a little loss of customers. They, or rather we, really need better competition that can compete with them on the same level. In this survey, their biggest competitor being tested is a game engine. Weird times.
Well, Chaos owns also Corona, and the Corona price has been same for years now, so technically speaking, they are also competing against themselves. I am not sure how long will this last, but as long as it does, the situation isn’t as grim as it seems.
Of course, game engines, especially Unreal are bringing crazy innovations to the table, and pouring increasingly more resources into attracting visualization market, so even if the above wasn’t the case, they still have a lot of competing to do in following years.
I’d not call these weird times, I’d actually dare to call them good times 🙂
I wouldn’t say necessarily competing against themselves. We were looking at Corona as an option to move away from VRay’s rising cost, since it can load VRay materials and some Phoenix stuff. So instead of losing a customer altogether, they are just sending them to a product that isn’t quite as feature rich. Almost feels like they don’t care if smaller studios leave VRay anyways, but have Corona as an option for those that do.
They are also pushing subscription more, how long before they don’t even offer perpetual?
I have used many render engine for years(including V-ray, Corona, mentalray, redshift) never find better than Arnold for Arch-Viz and other projects
How do you get Caustics with Arnold?
There are always workarounds: https://docs.arnoldrenderer.com/display/A5AFCUG/Caustics
I know in some specific cases are needed. But… Do you use caustics much? In archviz we used to have it deactivated just to save time and fake them in one way or another. (10 years ago).
You know those are not caustics… those are pathtracer denoised splotches (or not denoised) hahaha
And yes, they are quite important as of today standards for Arch/Viz, Corona solved them pretty well, and LuxCoreRender too, keep in mind that in the old days performance was not the same, and quality was also not the same.
In the case you have a pool, you need caustics, if you have a big metallic hero object in your scene, you need caustics, if you have a glass table, or some glass dishes, or some drink, you need caustics by today’s standards, and in general, as soon as you have a highly reflective floor, you need proper reflective caustics, it’s quite basic for ArchViz nowadays 🙂
Arnold has exactly the same caustics capabilities of Cycles, and they are exactly equally useless hehehe
Back then we used to use spotlight with a texture of baked caustics for swiming pools, was looking as good as real caustics, no noise, and full artistic control + 10 times less render times. Now for sure render times decreased a lot, but I dont know if its not still better to use some of this tricks for better artistic control over full “real” caustics.
I just opened over 100 images from cgarchitect trying to found any caustic…. didnt found a single one.
Not saying we dont need caustics. but I dont know if not having caustics its the main factor arnold is not as used in archviz.
Not at all, the main factor is that it’s slow hehehe
Look for a pool, you will see caustics XD
Compared to vray(brute force) or Corona(Path tracing), Arnold is just faster, much more responsive, and supports heavier scenes much more efficiently.
Are you joking?
Arnold is one of the slower render engines I faced in my life, specially for interior scenes.
Since you seems to have it, please do some tests with those engines and show the render times, or show some kind of benchmark, because even a simple scene with 7 animated Lego characters was taking around 7 minutes when in the same computer Cycles would take 2 minutes as much, and we are speaking of an open space-only-characters scene.
BTW you mention Corona (pathtracing), why dońt you mention Corona with UHD, which is the actual system used because there is no actual real difference in final result, but there is a difference in speed, or VRay (brute force), when brute force should not be used so…
… yes, any truck can be faster than a Ferrari if we have the brake on the Ferrari.
People on average will always pick the best tool for the job. If Arnold is on 12th position in a survey specifically aimed at ArchViz, just 2 posts above MentalRay, which is now dead for several years, that pretty much speaks for itself.
ESPECIALLY since it comes as an out of the box default renderer in a software most widely used for ArchViz around the world, yet it’s superseded by standalone packages or renderers in other, way less ArchViz-popular 3D packages.
If a renderer starts in such an advantageous position as being a default, out of the box renderer in the most popular ArchViz 3D software and still manages to end up on 12th place in such a survey, it’s actually more of an indicator of how exceptionally hated it is. 😉
This statement… it makes me wonder if we are using the same Max/Arnold. Honestly, the last time I wanted to use Arnold in Max 2020 for a quick test the AS would just paint a black area in 1/3 of the viewport, the lights would produce no lighting at all due to the really cumbersome and useless exposure setup limbo and everything would just be crazy laggy even with 1 box and a light.
Arnold is – to this day – not integrated in Max to the same level 3rd parties have managed to commit to. The modifier based renderer properties are a bad idea driven by what can be deemed only as general lack of production experience, there’s no way you can manage these and get an idea how a scene is set up.
Sure, Arnold has some really cool and production proven features, it’s definitely a solid world-class renderer, but the integration in Max is not where it would have been had 3rd parties integrated it.
It is annoying to see our money spent on Arnold and related development/integration. The numbers in the poll speak for themselves, I guess.
Also, dismissing a piece of software by picking out one missing feature (caustics) doesn’t make sense in any serious discussion, especially with a feature that is negligible most of the time. There are many other issues worth pointing out, and most of them are on the Max integration side.
But maybe, Autodesk just listens to the wrong people… yes, they definitely do.
Yes, this pretty sums up the problems I found in Arnold currently. Some of them has been fixed (lights, improvements on abc importers and others). Some others remain, Arnold is doing things on the “Arnold way” to keep it unify with multiple applications. Thats the reason there are modifier based render properties, instead of adopting a more 3dsmax mentality, and keeping the renderer more tighlty integrated. But saying that Arnold improved a lot in max during last 2 years, you have as well Arnold operators, to override any parameter at render time procedurally that is quite nice for huge scene management actually. We even get Tyflow compatibility recently!
But most people dont know some of this newer tools, and needs time to get noticed.
Saying all of that, its an archviz survey where there is a clear king that almost everybody know on this sector. I have seen a bigger interest in arnold in other areas for sure during last year (still not sure if will change much Arnold ranking position). For sure not having a frame buffer in conditions in max, no LUTS, no ACES, no color management I think its a bigger factor to dont use Arnold than “not having caustics”, that yes its another small feature missing, but not think has as much weight as all the others.
I want to clarify one thing.
In no way I’m saying that Arnold is a bad render engine, Arnold is an amazing and powerful render engine, but it’s target it’s what it is, and it’s unable to render a full interior scene at 1080p in under 5 to 7 minutes, even in powerful machines, which is something basic for Arch/Viz unless you are able to absorb render costs and/or you are able to increase prices.
That’s the main reason for arnold not being used in ArchViz, it could be used in the same way Maxwell is used, but see where Maxwell is in that survey, and supposedly it was mainly for photorealism and archviz.
There are three numbers there that feel quite a bit surprising, the first one is the low penetration of redshift in the archviz realm, I would have expected a bigger number, it’s even below to good old mental ray!.
The second one is the growth of Cycles, it went from position 14 in the last to years, to position 9 this year, of course nothing comparable with any of the other engines, but quite remarcable I think.
Also another remarcable thing is the usage of Blender that can be extrapolated, I’m not sure about the survey methodology, but it’s interesting that if we add the two blender native render engines percentages, Cycles and Eevee, 10.4% of users using Blender, of course I could imagine that the result is not as linear and as simple as doing the addition, but nevertheless it’s interesting that there has been a growth in the Blender native render engines usage inside the ArchViz realm in one year.
Also I always thought that this survey is kind of skewed, since the majority of CGArchitect users seem to be max users, and I know several ArchViz artists that don’t like to be there just because it feels a bit like a “Max only” community, but reality is that max is king in the ArchViz industry, no doubt, but anyways, it’s a tiny bit skewed towards max users, I don’t think there is nothing they can do to solve this, since they work with their community, nothing more, nothing less.
Interesting information anyways.
I don’t think you can add the percentages like that. Combining the values in that way would result in an overall value exeeding 100%.
Maybe the survey allowed for multiple choices?
What I think it says is that 6% use cycles for some stuff. Seeing as its free, maybe not that surprising.
Of course, I doubt it’s a linear addition 🙂
However the increase in usage from 2018/2019 to 2020 is there, that’s what I find interesting 🙂