Single Image Model Extraction

purely magic !
that’s a future good tool for film 3d conversion.

Hi othoap,

First thing: it is simply marvelous what they did! From the technological aspect, i am amazed what kind of genius code they have created for this calculation – i don’t even have an idea where to start thinking to do such thing. There is no doubt that this technology will become in 10 years something like 3D Printing today.

Back to the question why it wont be published.

There are few reasons why many of this research projects never get to our PCs. (or never become a toy for $ 99,- for 3d artists)

1) Prof of concept.
The presented software works in the encapsulated environment, with predefined parameters. It is optimized for certain scientific cases. For example, in this video you see only orthogonal photos and lathe/primitives. It is marvelous what they did – but who in the world need it in current form? If they publish it as such, the first beta-tester idiot will prompt: “Hey, i tried to trace the Ironman and the shit comes out, LOL” (go to youtube and read what homo-sapiens write there).
Resume: it is great prof of concept. To complete it and make it idiot-resistant, it make take 10+ years of development.

2) Revenue is miserable.
If i take a software development company as an example, they would need about 3 programmers and 1-2 years to make some semi-professional version that may even trace Ironman head from the movie. Lets round it to 3*$5000,-*12 – about $200.000 plus $ 200.000 for the company costs (overhead, taxes, electricity, office, computers, you know already the stuff). We land by $ 500.000,- to 1 Million for some half-assed software.
And this software will NEVER earn $ 500.000,-. You can’t sell it for $ 500,-, If you can sell it for $99,-, once it is cracked you have anyway few random sales a week. You can’t seriously earn money under such conditions.
Beside – i don’t want only to cover the half million production, i need to earn additionally half million on top of it in order to please my investors.
The only way to see it on our computer is to wait if someone sit down and build it into Blender and then you have again a half-ass open source tool…
Resume: it is financially not attractive.

3) Usage is limited.
Who does need it? If it is a $ 99,- i may buy it for some random productions. It does not capture the surface details, so i can’t even use it for 3D printing. It is a small, camera projection tool for matte-painter and low-poly game artists. If you connect this to point 1 and point 2 above, you will see that there is no investor who will risk to invest money in such low profit business.

Don’t forget it is 3D world, the sales are low, and you don’t have so many customers to utilize this technology, as it may be the case in the 2D World (Photoshop etc), or Word etc.
Just my opinion based on my experiences… 🙂

The other thing is that a lot of cool papers have to be licensed from universities. Because universities have a low budget and researchers have to get funding by their own they very often work with large companies like Microsoft, Adobe, Nvidia and Pixar so these concepts get patented and very expensive to cut out competitors. With the fantastic comment of Igor you see this makes it very hard for start-up companies that don’t do “fancy” things like Mobile Apps to attract investors.

Sometimes, after many years, they find their way into software but since competition is decreasing into oligopolies/monopolies there is no need to bring these innovations to the public. Years ago sometimes these prototypes have been released as source code, but since this was “dirty” code it is impossible for people to simply recompile. Thanks to the funding structure this now happens rarely because these big companies need munition for the patent wars and keep the shareholders happy with patented research.

I’m always inspired but also very sad when I look at the fantastic papers, knowing this is dead knowledge for non-researchers, something that has been invented but that only stays in the research community until it is “important” enough for companies. Often only smaller developers like the Max plugin or Blender community are pushing these ideas, but they wouldn’t stand a patent claim if they do something that would hurt a big player…

Check this Siggraph Asia 2013 trailer to see what I mean with “being sad about this system”: http://www.youtube.com/watch?v=l81MqEWmR-g

Rune, I purchased Canoma years ago. I miss the innovation of Metacreations. I think Adobe bought it and killed it off.