I wonder what's different between this and, say, if the guys who make Maxwell decided to make it run on Amazon EC3? multi-optics sounds like another physical light renderer
There is already something like that, for some time, called Felixrender.
I've never heard of that one; my point was more about creating a new physical renderer on the cloud vs an existing, production proven one. Have someone knowledgable explain how the new renderer is differentiated.
Btw, Autodesk has Autodesk 360 which is a collaboration and rendering tool on the cloud. Mental Images had worked for years on something callled Incremental Image which was a cloud based mental ray for collaborative rendering and reviewing across the web (it didn't pan out.)
I'm friends with Thiago and I was fortunate to have him demo TeamUp MultiOptics in person about month ago. Having seen it in action first-hand, I dare say It's fantastic!
Here's what I know:
What does MultiOptics have over other renderers?
** Unbiased renderers like Maxwell and similar tend to be very pretty but very slow. MultiOptics is stupid crazy fast, losing perceptible sampling noise within a minute or less. I don't have hard benchmarks to lay down but just commenting on what I saw.
** MO also uses material shading models that have been actually measured from physical real materials. It has a growing library of physical materials that look incredible. -- Their subsurface, hair and and sand grain materials are particularly phenomenal. (MO renders the most realistic sand I ever saw, and sand is really *hard* to get right! Sony would have loved to render Spiderman's Sandman in this had it existed back then.)
The asynchronicity feature of TeamUp is also a lot of fun. An absurd number of people can simultaneously edit the same scene, bringing in assets, creating/placing lights, repositioning anything, changing camera angles, rendering alternative cameras, changing materials and any attributes, etc. It's all done in the browser with modern technologies like HTML5 and WebGL. No Flash or any other plugins required. It runs very snappy on Chrome and Firefox. Oh and when I say asynchronous I really mean every nuisance. If I'm slowly rotating an asset, it doesn't update when I let go... it updates constantly every few ms and everyone can see the changes happening in their own browser.
I was told there is a backlog of all user events so it should be theoretically possible to have the scene go back in time to any point. Also there is no Save button. Everything is saved constantly (much like modern web apps like Google Docs.) If your computer dies, your building loses power, or you spill coffee on the keyboard, just go to another computer, login and open the scene exactly as it was left off; no problem!
It's also cool to just watch people work. A supervisor could very easily open scenes from his computer and see how the shots are doing in realtime! I can't wait to see them implement volume rendering, too. I bet MO fire and smoke is going to look incredible!
As for the point about MO not being production-proven, well, it's early still. Give them a chance. You know what's production proven though? Thiago Costa himself!
I'm very excited to see where this goes. It's not hard to envision a potential future where we have simpler, lighter computers at work and can do our jobs the same from a netbook, a workstation or even an iPad.