The Making of Toy Story (1996) [pdf] | Hacker News Reader

The Making of Toy Story (1996) [pdf]

people.uncw.edu

66 pointspncnmnp15 comments

I laughed at this line from Isaacson's "bio" of Steve Jobs:

“As the Disney folks were showing him around, Jobs turned to Katzenberg and asked, “Is Disney happy with Pixar?” With great exuberance, Katzenberg answered yes. Then Jobs asked, “Do you think we at Pixar are happy with Disney?” Katzenberg said he assumed so. “No, we’re not,” Jobs said. “We want to do a film with you. That would make us happy.”

Silicon Graphics was such an emblematic company of the era. Too bad they are not around anymore.

Also super cool to read that idle Indigo2 became part of the RenderFarm!

During my time at PDI/DreamWorks, all idle desktop machines would be incorporated into the render farm as well. I don't know if this is the case anymore with so many features of the renderer and simulations software depending on the GPU.

I was surprised at how much CPU only rendering was still done. About 5 years ago, I was a TD at a post house/animation studio. Some of the guys were demoing GPU rendering to see if it was something they could start utilizing. The issue at hand was that they were so heavily invested in the existing data center (leased machines). They were doing the experiments of using multiple GPUs way exceeding the motherboards designs by using breakout cables so each PCIe slot connected multiple GPUs. It was very much Frankenputer with massive PSUs, and cards laying on a desktop not even in a chassis. Even then, the speed was impressive, just wasn't scalable at that time.

Most feature film work is still highly CPU only. This extends to farms being largely CPU only. In those events, idle workstations are actually more capable if the renderer can use the GPU, but that’s rare.

Pixar’s own Renderman is still primarily a CPU renderer for them even though they have XPU coming, and DreamWorks’ soon to be open sourced MoonRay is also primarily CPU only.

Most feature film scenes are still too large to fit in GPU memory comfortably. While some GPU renderers can handle out of core rendering (Redshift etc), there’s a lot of concessions to be made in terms of shader types (subsurface is/was much better on CPU renderers) and performance issues when shuttling data back and forth.

When we did some evaluation of using Redshift for some projects, our CPU renderer still handedly beat it for mid-large scenes, so it was rarely worth the effort to use it in our cases.

There’s also the issue of retrofitting infrastructure. The studios have hundreds of farm machines without GPUs. Most studios run on leases so would have a delay before GPUs would be next widely available in their farms.

Ironically, one of the best devices for XPU rendering would be the new Apple Silicon macs. Not having to shuttle data back and forth, plus having a relatively massive memory pool for the GPU is a really advantageous setup. In theory, they could actually out pace much higher end hardware for specific scenes if the renderers could take advantage of both CPU/GPU and unified memory.

It was inevitable. Computing power became cheap enough no one had to spend $200k on a workstation

Remarkable how 3D, Digital VFX and Animation are now common place from the biggest to the one person studio.

An interesting book to read is "A Biography of the Pixel" by Alvy Ray Smith[1], for those interested in the history of the sampling theorem, computer imaging and computer animation. It concludes with the steps leading up to the making of "Toy Story".

[1] https://direct.mit.edu/books/book/5145/A-Biography-of-the-Pi...

> Disk resources are equally vast, with a DiskFam of around 260 gigabytes of on-line storage containing approximately 140,000 digital image files.

The 140'000 image files would translate to an 80 minute runtime at 30 fps. So I assume that's all the storage they had.

Makes me think of the day when my friend brought a 256MB USB stick to school and we were blown away.

Films are 24fps.

Also, the first time I saw an SD card, I felt like I was in a Gibson novel.

You're right. That's what I found in the Wikipedia article[^1]:

> The film required 800,000 machine hours and 114,240 frames of animation in total, divided between 1,561 shots that totaled over 77 minutes.

This comes out to 24 FPS. The article talks of 140k frames, but it looks like they didn't use all of them for the final movie.

[^1]: https://en.wikipedia.org/wiki/Toy_Story

Many shots are left out by the time a film is finished and every shot has multiple iterations in each of many departments that touch it.

This post has an example diagram of a feature film pipeline.

https://www.gfx.dev/python-for-feature-film

Toy Story would have been more rudimentary and ad hoc, given how early it was, but this is generally the same pipeline that’s been in use even for old hand drawn animated films.

One interesting anecdote is of the character of Woody, the film's main protagonist. In early versions of the story, Woody was designed to be a villainous character who would bully the other toys. However, as the story developed, the filmmakers realized that this approach didn't fit with the overall tone of the film.

A VHS with some related history.

https://m.youtube.com/watch?v=1VdERcRJ3jI

I greatly enjoyed the read.

Here are my highlights.

https://nono.ma/the-making-of-toy-story

Reply on news.ycombinator.com