Toy Story 3

This is a new forum that can be used to discuss anything that's not directly related to JPatch.

Toy Story 3

Postby John VanSickle » Sat Jun 19, 2010 5:01 pm

Saw it today. Aside from making everything I do look like garbage, the story had a more poignant ending than the first two.
John VanSickle
 
Posts: 189
Joined: Sat Feb 16, 2008 2:17 am

Re: Toy Story 3

Postby dcuny » Sat Jun 19, 2010 6:16 pm

I probably won't be able to see it until it's out on video. :(

The renders I've seen look awesome, but the rendering times are commensurate. I've got the Pixar Short Films Collection, and I was watching it the other day to see what "ancient" Reyes rendering looks like. It's easy to forget that even the shorts had a large team behind them. For example, about 20 animators worked on Jiri's Game.

Anyway, I'll just note that from a technological perspective, Pixar's own films don't hold up that well, either.
dcuny
 
Posts: 2902
Joined: Fri May 21, 2004 6:07 am

Re: Toy Story 3

Postby John VanSickle » Wed Jun 23, 2010 12:51 pm

The difference between Toy Story and Toy Story 2 is quite vast from the viewpoint of lighting and texturing quality; but then Pixar certainly had the bucks to support longer rendering times on their farm.
John VanSickle
 
Posts: 189
Joined: Sat Feb 16, 2008 2:17 am

Re: Toy Story 3

Postby dcuny » Thu Jun 24, 2010 6:54 pm

I'm curious how all this ends up effecting the design of RenderMan. The primary failing of the renderer is that it didn't account for global illumination effects. Traditionally, GI effects are handled by ray tracing, but PRMan didn't initially support ray tracing, so for A Bug's Life they used BMRT for ray traced shadows.

Renderers like Arnold made animation with path tracing practical, but they could really only perform single-bounce illumination in outdoor scenes, and relied heavily on irradiance cache estimates to reduce calculations. (Most ray tracers are fast enough that TDs would rather do without the irradiance cache and have a bit of extra noise in the scene, instead of having to endlessly fiddle with the cache parameters and get odd frame-to-frame artifacts.)

The added ray tracing for Cars, but the calculation times for effects like approximate ambient occlusion were only practical for people with renderfarms at their disposal. Ray tracing is slow, and goes against the general design of REYES, which is designed to render geometry in bits, and then discard it. Ray tracing requires that geometry remain in memory. When there's too much geometry to hold in memory, you end up having to go to a caching scheme, which is a pain and makes things even slower. :(

Ratatouille used photon mapping for GI effects, storing the information in brick maps - a sort of 3D mip-map that could be cached to disk. The results were very nice, but it still didn't fit well into the REYES architecture.

Up changed over to point-based illumination, which is the state of the art for global illumination these days. Point-based lighting fits very well into the REYES architecture, because you can use the dicing algorithm to generate the point cloud. (It's actually a "disk" cloud, not points). Although caching of scenes may be necessary, often a scene which won't fit into memory for the raytracer will fit into memory for point-based occlusion.

There's a paper on color bleeding which talks about how this is implemented in PRMan. It's very slick - they've extended the algorithm in a number of ways. For example, to prevent disks which are occluded by other disks from contributing to a pixel's illumination, they rasterize the visible disks as seen by the point. The initial implementation used a hemicube (similar to early GI programs), but it turned out to be cheaper to use a cube map and avoid the trigonometry functions (the cube's resolution is 8x8 per side). This allows not only approximate ambient occlusion, but color bleeding effects, soft shadows, and even glossy reflections. 8)

Edit: Here's a link with a high level overview of point based occlusion. I found the following comment interesting:
The use of point-based rendering techniques is reserved for complicated scenes, scenes that have lots of geometry and displacement shaders.
But I suspect that they're referring to color bleeding effects, since approximate ambient occlusion is pretty much used everywhere now.
Last edited by dcuny on Thu Jun 24, 2010 8:18 pm, edited 1 time in total.
Reason: Added another link.
dcuny
 
Posts: 2902
Joined: Fri May 21, 2004 6:07 am

Re: Toy Story 3

Postby dcuny » Thu Jun 24, 2010 11:34 pm

The cool thing about point-based lighting is that it's independent of the renderer. For example, Pixar has implemented a stand-alone application called ptfilter which:
...performs a variety of point-based calculation on RenderMan's point cloud files. Ptfilter takes a point cloud file as an input and outputs a point cloud with new data. Ptfilter is most often used for calculating subsurface scattering diffusion, approximate occlusion and colorbleeding.

Sascha's never been keen on writing a "native" renderer for JPatch, but I'll note that JPatch is already a high-quality micro-polygon renderer, just like REYES. So (in theory) point-based rendering features (like approximate ambient occlusion) could be bolted on to it...

...assuming that someone wanted to write it. :P
dcuny
 
Posts: 2902
Joined: Fri May 21, 2004 6:07 am

Re: Toy Story 3

Postby dcuny » Thu Jul 15, 2010 6:13 pm

I still haven't seen Toy Story 3 (and I'm not likely to see it until it's out on video :(), but our family were at the California State Fair yesterday, and they've got an exhibit on the history of animation. There's some nice stuff, like one of the sets used in Ray Harryhausen's "The Tortoise and the Hare", maquettes from Shrek, lots of video clips from Cloudy With A Chance of Meatballs showing the development of scenes, and so on.
as well as a number of videos

The Pixar section was mostly promoting Toy Story 3, and they had a couple of videos playing. One of them had a number of clips from the "original" Toy Story, apparently before the massive rewrite. For example, they had a shot where Woody tricks Buzz into getting wedged behind the dresser. I hadn't seen the shots before, and the artwork was a lot closer to the original Bud Luckey designs - very cool to see. Maybe they'll be on the Toy Story 3 video or are on some other extended video.

In the "Making Of" book for Toy Story, they mention that when they put all the clips together for the first time, they realized the film was terribly broken. It makes me wonder if there'll ever be a "bootleg" version of Toy Story floating around from that version. :P (Sort of like one of the booths at the fair that was selling "official" version of Song of the South on DVD ;)).
dcuny
 
Posts: 2902
Joined: Fri May 21, 2004 6:07 am

Re: Toy Story 3

Postby pndragon » Mon Jul 26, 2010 8:02 pm

dcuny wrote:"official" version of Song of the South on DVD ;)).

What, exactly, is the official version? I haven't seen even seen a clip from that movie for over 20 years. Not Politically Correct.... I hate that term. Everything offends someone and to attempt to sanitize the older movies, like they are now and removing some really tremendous performances (I recently saw a Shirley Temple film that tried to eliminate most of Bill "Bojangles" Robinson's part because of his subservient role; It was a butchery) as sacrifices on the altar of PC.... well, to use another non-PC phrase, that's throwing the baby out with the bath water.

I'll climb off my soapbox in just a minute. It's just that my kids would probably enjoy "Song of the South" but they will also never see it, at least not the version I saw as a kid and that saddens me no end.
"We're so sorry, Uncle Albert,
But we haven't done a bloody thing all day."
--- Paul McCartney
pndragon
 
Posts: 591
Joined: Sun Dec 05, 2004 1:27 am
Location: North Carolina

Re: Toy Story 3

Postby dcuny » Tue Jul 27, 2010 5:03 am

pndragon wrote:What, exactly, is the official version? I haven't seen even seen a clip from that movie for over 20 years.
According to Wikipedia, the full version has been released outside of the US. Interestingly, from Snopes:
Perhaps lost in all the controversy over the film is the fact that James Baskett, a black man, was the very first live actor ever hired by Disney. Allegedly, though, Baskett was unable to attend the film's premiere in Atlanta because no hotel would give him a room.
dcuny
 
Posts: 2902
Joined: Fri May 21, 2004 6:07 am


Return to Off topic

Who is online

Users browsing this forum: No registered users and 1 guest

cron