(image credit: ScreenDaily.com)
Two noteworthy industry publications were covering the AbuDhabi Film Festival, and had the foresight to attend the Masterclass on 3D Film Production in the 21st Century. They each came out with different headlines for their reports.
Both articles were interesting to me, as it was an example of what the audience may have taken away from the two and half hour masterclass. Designing the masterclass in itself was a challenge, as I did not know the composition of the audience. I had to take into account the possibility of Directors, Producers, Film Students, animators and as I learned later from a few people; They were IT technicians looking at a switch in career.
I’ll try to use some lines from the two reviews and explain in a bit more detail from my point of view what was happening on-stage. There were many topics covered and I took 3D to mean both, regular 3D CG and of course Stereoscopic 3D.
This article will be in parts as the title shows. We start off with the enormous strides that are being made in bringing Digital Actors to life. The clip below was one of my opening slides:
The video, no… “creation” is the work of the extremely talented Jorge Jimenez. Here’s the best part, for the masterclass, I had downloaded the EXE file from his site. So the exe starts by playing the video (actually realtime render with camera path), until the stunned audience saw me take the mouse and move the head around in all it’s SSSS glory in realtime. All this on my HP HDX laptop that has a three year old Nvidia 9600 GT chip!
The holy grail of CG human rendering is to achieve “realism”, but read the second paragraph from Jorge’s site, and you will see that for cinematic realism, it could actually be a lot easier as we “dirty down” the otherwise pristine CG rendering, and yet use methods such as Jorge’s separable SSS, to speed up render time with today’s GPUs.
“Remember”, I said to the audience, “We are not looking for real-time realism as Games developers need to. For Cinema you could offline the process, but use the giant strides that Pixel Shaders, and other real time graphics technologies now allow us”.
For the more technically inclined, here’s a paper from Nvidia on the subject of rendering more realistic Human skin in CG)
The main takeaways from the first 30 minutes of my talk were:
- Game Engines and CG technology are going to drive Cinematic production henceforth.
- Hybrid Cinema Production is what 3D film making will be all about. The mixing of Live action and CG, both for Actors and the environment or “Stage”
- I made it a point to repeatedly explain the difference again between vanilla 3D and Stereoscopic 3D
- The person (actor) undergoes a whole body scan.
- Even more detailed scan of the skin face texture, and maybe hands.
- This skin texture is “baked” let’s say on Oct 15th 2012 and the person (actor) is 35 years old.
- Facial expressions are captured based on algorithms that are unique to a facial capture software. Through CG morph targets, and CG Face rigs, almost any expression could then potentially be synthesized. See the FACS explanation for more technical details on Facial Action coding.
- Facial expressions unique to an actor can also be captured as a “macro” rather than re-creating it via FACS.
- Finally, motion capture. Though not needed, as it can be re-targetted from any actor, in certain cases, the best mo-cap performance for a certain “Style” is best acquired from the orginal “Human”.