Maybe this is one of those “when I was a lad” moments (ref. [20]), but I clearly remember the end of the movie short era.   Along with mid-movie intermissions, “shorts” vanished some time around 1980, to be replaced by the long succession of “coming soon” movie trailers.

The location of this particular memory was the Burrill Lake Open Air Theatre in New South Wales.   This was one of the last open air theatres in Australia, with canvas seats slung between metal poles, and an awning at the rear in case it rained during the movie.   Surprisingly, it was a very comfortable movie experience, as long as it didn’t rain and everyone ended up squashed under the awning.

On this non-rainy night, my Dad was disappointed that the short film he had hoped to see wasn’t being shown.   Indeed, no short was being shown!   He asked about it, but they sneered “theatres aren’t showing shorts anymore”, as if Dad had asked them to show a silent film.   “Got rid of the lot of them,” the projectionist added.

Now, I’ll be the first to admit that this was a mixed blessing.   There were some terrific short movies out there, but many others were guaranteed insomnia cures.

However, even the loss of those boring shorts imposed an opportunity cost.   A generation of young filmmakers were deprived of the main venue to display their works.   “Shorts” were a step up into the feature film industry for many aspiring filmmakers, providing a venue for works which could prove their capacity for bigger and better productions.   Sure, film festivals still existed, but you didn’t get the wide exposure a pre-feature short provided.   Indeed, these festivals were mostly packed with other filmmakers in the same position as you.

But you know, shorts did live on, just not in theatres and film festivals.   That’s where we get to Pixar.   How do Pixar and short films relate?   Well, read on…

Pixar’s birth was as a research project in the Computer Division of George Lucas’s Industrial Light and Magic.   The addition of computerised motion controlled rigs to cameras was the technology which facilitated the ground-breaking special effects in Star Wars (1977).   Ever the cinematic technocrat, and doubtless also looking at eight even more ambitious eight Star Wars sequels he had planned.   Lucas wondered just what other filmmaking tasks computers could be useful for.

Thus the Computer Division in ILM was formed.   Four main projects emerged from this division: EditDroid, SoundDroid, Pixar-2D and Pixar-3D.

EditDroid was a non-linear film editing system.   Film was digitized onto analog laserdiscs, and software running on a Sun-1 workstation sequenced the film together based on a cut-list compiled by the editor.   By now you should be thinking Nero Vision, which pretty much achieves the same result, albeit much more conveniently.

EditDroid was well received by the industry, but was never a commercial success. While it was used on some other motion pictures, Lucas himself never utilized EditDroid on his own movies.   Some suggest that this fact and the product’s failure are not unrelated!

SoundDroid was a computerized audio editing and mixing workstation.   If you’re thinking Nero WaveEditor, you’re on the right track.   SoundDroid was widely used, including by Lucas’s own movies.   Both SoundDroid and EditDroid were spun off into a company called DroidWorks, which was eventually acquired by Avid.

Pixar-2D was a replacement for optical film printers.   As already noted, those groundbreaking special effects in Star Wars were accomplished by using robotic cameras to film miniatures against blue screens.   The color stock was then printed onto high-contrast (“hi-con”) black and white film stock, creating mattes which could mask out the model, or mask out the background and model mount.  Think of layers in Photoshop, and you’ll have a rough idea, except this was all put together using huge optical compositing systems known as optical film printers.

The famous asteroid belt sequence in “The Empire Strikes Back” (1980) often had twenty or thirty elements, plus associated mattes, all fed through layers of optics.   As the light passed through multiple lenses, film and mattes, it was attenuated and lost definition.   Although the lenses used were all extremely high quality, too many layers would leave the image looking muddy and undefined.

Alignment was another big challenge, as the mattes and source images had to be perfectly aligned to each other, or objectionable color fringes would appear around the objects in the shot.   Go see early prints of “Star Trek: The Motion Picture” (1979) for a demonstration of misaligned mattes.

The Computer Divisions solution was to composite the images digitally, which now seems obvious, but then was a huge engineering challenge.   Thus was born the Pixar Image Computer, which was a external graphics co-processor for Sun or Silicon Graphics workstations.   It was optimized for handling large, high-resolution 2D imagery and processing it at speed.   Disney bought a few, and they also found their way into a number of medical and defence industry applications as well.

The Nero suite by itself doesn’t provide support for chroma-keying or color-based matting.   But a number of other well-known packages do, and all modern film compositing is done on SGI’s, PC’s or Macintoshes using high-end compositing software.   Nowadays, the optical film printer is an impressive museum exhibit.

So finally we come to Pixar-3D.   Pixar-3D was initially envisaged as a full hardware implementation of the REYES (“Renders Everything You Ever Saw”) 3D pipeline.   It was planned as a collection of boards, each of which would implement one stage of REYES.   The first to be completed was the filtering/pixel-reconstruction board, called “RM-1”, which was designed using a parallel array of INMOS Transputer CPUs.   However, the designers soon realized that the board was so elegant they could implement the whole pipeline on this “RM-1” board.   Not as fast, but much more flexibly.

By this time Lucas has spun-off Pixar as a separate company, modelling the approach taken with DroidWorks.   It is well known that Steve Jobs provided a significant amount of funding to keep Pixar working, without which the company would certainly have folded.   The RM-1 computer was only ever used to render a single short, the Academy Award-winning “Tin Toy”, and never found a customer outside of Pixar and ILM.   Reference [14] adds that Moore’s law doomed the RM-1 as a long-term solution, and that custom hardware like the RM-1 is very challenging to support.

But Pixar realized that the value they owned wasn’t in the RM-1 hardware, but in the software interface which powered it.   This software was known as the Renderman API, and it exists to this day as the foremost “production quality” API for film quality, non-realtime graphics.   Pixar themselves sell the PRman (Photo-Realistic Renderman) implementation of the interface, but there are a handful of others also.   The specification is freely available for download (Reference [22]).

But how do you show off your incredible new technology?   Well, that’s where short films come in.

Every year, the Association for Computing Machinery (ACM) hosts the SIGGRAPH conference.   SIGGRAPH stands for “Special Interest Group on Graphics and Interactive Techniques”, and almost every single significant advance in computer generated imagery has been announced there over the years.   It’s the “must go” event for researchers, companies and associations interested in computer graphics.

Film clips were already fairly commonplace at SIGGRAPH, usually displaying a new lighting algorithm or rendering technique.   Most were purely technical demonstrations, with little or no artistic merit, as impressive though they were.

“Luxo Jr” was a short film Pixar presented at SIGGRAPH 1986.   All accounts describe the production as a mad rush to make the conference deadlines, with animators bringing sleeping bags into Pixar and napping under desks.   Anyone who’s ever seen a Pixar movie will know the main character of the short: a table lamp.   It’s now part of Pixar’s logo, which is shown at the start of every Pixar movie.   Actually, the lamp was the modelled on inexpensive Luxo-brand table lamp which Pixar had put on all of their animator’s desks.

At the time, most traditional (cell) animators firmly believed that CGI could never portray a character.   It seems like an ignorant statement now, but passions ran hot, fuelled by fears of being made redundant.   “A computer will never do Snow White” was mid-1980’s quote from one traditional animator.

“Luxo Jr” proved them wrong.   Here was a realistically-rendered table lamp, who was a parent.   Here was another, which was a child.   Without any dialog, and using very basic rendering techniques, Luxo Jr. portrayed a parental relationship, character, emotion, intent and humor.   It was a revelation.   CGI could do warm, human stories, even if the characters were table lamps!

Historical note : Luxo Jr was the second movie from the folks who founded “Pixar”, but there was one which preceded it.    While at ILM, a short called “The Adventures of Andre and Wally B” was rendered on a Cray X-MP/48 supercomputer using very early software.   While technically interesting for many reasons, “The Adventures of Andre and Wally B” was much closer to a technical demo reel than the later character-driven Pixar shorts.   It is generally not considered to be a Pixar short per se.

In the years which followed, the SIGGRAPH Pixar short film became something of an institution.   Most of these shorts can now be purchased through iTunes, or obtained on the Pixar shorts DVD collection.   Pixar’s watershed first movie, Toy Story, was partly inspired by the 1988 RM-1 rendered short “Tin Toy”.

But Moore’s law changes everything: most people’s laptops today far exceed the capabilities of what was available to the pre-“Toy Story” Pixar.   The DEC VAX 11/780 minicomputer on which Renderman was originally developed had a 32bit CPU clocked at 5MHz, with 8K of cache, and could be expanded to an awesome 16MB of RAM.   That’s much less powerful than the microcontroller which is built into the disk drive in a modern PC, let alone the whole PC.

There are low-cost and even free tools which far exceed the capabilities of the toolset available to these pioneers of feature CGI.   There is also a thriving online marketplace for 3D models, sets and props, which significantly decreases the work involved in set dressing.

From its roots as a CD authoring application, Nero has become a powerful multimedia toolset capable of producing and mastering standard- and high-definition multimedia content.    In owning it, or comparable tools, you’ve got a capable production facility in your PC.   Decent SD 1-CCD digital movie cameras can be purchased for a couple of hundred dollars, and a good SD 3-CCD camera is around the USD$1000 mark.   A high-definition camera might set you back a couple of thousand dollars, but is still cheaper than the second hand 16mm camera of a decade ago.   That’s enough for a basic live action production.

As for CGI, the tools are already available, and in an upcoming post, I will list a selection of CGI packages and both free and commercial model sources which would allow a talented individual to produce something comparable in quality to the early Pixar shorts.   DAZ Studio, one of the free tools which will be discussed in the next post, even incorporates 3Delight, a Renderman-compliant photorealistic renderer.

 

Image: DAZ Studio v1.7 displaying a 3Delight-rendered sample image.   Render time was 2 minutes and 19 seconds on a Athlon X2 6000+ with 2G of DDR2 DRAM.

At the start of this post, I bemoaned the death of the short film as a “step up” into the professional film industry.   Well, that too is a solved problem.   We have My Nero, and YouTube, and many other video sharing and social networking sites.   The barriers for entry are very low, and compelling content does attract widespread attention.

So why isn’t everyone doing this?   Well, it’s because tools are only a part of the problem.   You still need talent, perseverance, and vision.   All of these need to be combined to produce interesting, compelling content which people want to watch.

In the CGI field, many of us remember Tron (1982), which was widely touted as one of the first movies to use CGI (actually, it wasn’t – see [19]).   Fewer still would remember “The Last Starfighter” (1984), which although notable for it’s production design, is entirely forgettable for it’s plot and character.

It’s not that you are able to “do” CGI and put it together, it’s how you do it, and what you show.   It’s the story, the emotion, and the characters.   It’s portraying a lifeless desk lamp, yet imbuing it with personality, emotion and intent.    It’s the difference between the SIGGRAPH demo reel showing an office set rendered using a radiosity algorithm, and Luxo Jr.

So you have the 2D tools in Nero, and in the next post I’ll cover the 3D tools.   My aim is simple: a couple of hundred bucks + talent + time = a short which would rival Pixar’s early films.

Possible?   Yes.   Easy?   That’s up to you.

Sadly, Burrill Lake Open Air Theatre was demolished in 1999.   The land was subdivided, and sold off for housing development.   All you’ll find there now is a block of cookie-cutter project homes, with little evidence of that beautiful old theatre.

But in closing this trip through CGI’s yesteryear, let’s pause and ponder how cool the gals and guys at Pixar are, because they’ve brought back the movie short!   Each and every Pixar feature, and DVD release, is accompanied by a new short.   Indeed, many of us look forward to the short even more than the movie it accompanies.

 

Afterword:

This article is intended to be a call to arms to artists out there, using recent histories to remind everyone how far CGI has come in a very short time, and how accessible it has become to everyone.   Almost all of the CGI pioneers are still alive today, and actively working in various companies.

I am very aware of the risks in putting together a history as I have above.   In doing so, I’ve undertaken quite a bit of research, and have tried to collaborate multiple sources to confirm dates and facts.   Most of those sources are listed below.

However, if I’ve made a mistake, please either email me, or post a correction in the comments below.

Oh, and “a computer will never do Snow White” guy?   Apparently now works as a character designer and lead animator on CGI features.   I am told he was one of the senior animators on “Happy Feet”.

References:

1. Pixar Corporation (now part of Disney)

www.pixar.com

2. Pixar (Wikipedia)

http://en.wikipedia.org/wiki/Pixar

3. Pixar’s Short Films Page:

http://www.pixar.com/shorts/index.html

4. The Adventures of Andre and Wally B (Wikipedia)

http://en.wikipedia.org/wiki/The_Adventures_of_Andr%C3%A9_and_Wally_B .

5. Luxo Jr (Wikipedia)

http://en.wikipedia.org/wiki/Luxo_Jr .

6. Tin Toy (Wikipedia)

http://en.wikipedia.org/wiki/Tin_Toy

7. SIGGRAPH

http://www.siggraph.org/

8. A Critical History of Computer Graphics and Animation

http://design.osu.edu/carlson/history/ID797.html

Section 11 is most relevant, but there is a huge amount of background plus a good collection of historical CGI imagery.   I’d recommend finding a few hours, and settling down to read the lot.

9. Renderman (CG Society Information Page)

http://wiki.cgsociety.org/index.php/Renderman

10. Renderman (Wikipedia Disambiguation Page)

http://en.wikipedia.org/wiki/Renderman

Links to the various instances of what is generically referred to as Renderman.   Of special interest is the list of the numerous implementations of the Renderman specification, of which PRman (Pixar’s product) is just one.

11. The REYES Image Rendering Architecture

http://graphics.pixar.com/Reyes/

12. Pixar’s Online Papers Archive

http://graphics.pixar.com/

13. “The Road to Point REYES” – A Seminal CGI Image from the ILM CG Era

http://wiki.jpatch.com/doku.php?id=images:images#1983

Rumored to be what REYES really refers to, with “Renders Everything You Ever Saw” just a convenient fit for a pre-chosen acronym.

14. Advanced Renderman: Creating CGI for Motion Pictures (Book)

    by Anthony A. Apodaca and Larry Gritz

    Morgan Kaufmann; 1st edition (December 8, 1999)

    ISBN 1-55860-618-1  

15. A Historical Timeline of Computer Graphics and Animation

http://design.osu.edu/carlson/history/timeline.html

16. ILM Tour

http://insideskywalkerranch.com/ilm-tour.htm

This is one of the few sites online that give some detail about the SoundDroid.   Most sites just reference it, with little further detail.

17. Lights, Cameras… Computers (Article, August 1994)

http://www.digitalzoo.com.au/lunchtime/lunch_docs/books_05_vintage_01.htm

18. EditDroid (Wikipedia)

http://en.wikipedia.org/wiki/EditDroid

19. Timeline of CGI in Film and Television

http://www.emalta.com/mediawiki/index.php/Timeline_of_CGI_in_film_and_television

20. “When I was a Lad” moments….

http://en.wikipedia.org/wiki/Four_Yorkshiremen_sketch

21. Ram Meenakshisundaram's Transputer Home Page

http://www.classiccmp.org/transputer/

Most people won’t have heard of the INMOS Transputer, but in the 1980’s that CPU was synonymous with high-speed floating point and multiprocessing.   INMOS was eventually sold, and acquired by ST Microelectronics.   Although generally considered a dead architecture, there are actually more Transputer chips being manufactured today than ever.   When ST Micro acquired INMOS, they renamed the Transputer architecture the ST20.   So if you’ve got a set-top box with the STi5000-series CPU, or a GPS with a ST20-GP6 chipset, you actually own a system powered by a Transputer.   There are literally millions of them out there now.

22. The Renderman Interface Specification

http://renderman.pixar.com/products/rispec/index.htm

23. Pixar Image Computer (Wikipedia)

http://en.wikipedia.org/wiki/Pixar_Image_Computer

24. Pixar Image Computer

http://www.specktech.com/PixarImageComputer.html

This reference shows some details of the custom SIMD CPU called CHAPs. Interestingly, the PIC ran NeWS (the Network Extensible Windowing System), a technically superior but unsuccessful competitor to the god-awful MIT X11 windowing system.

I’d so love Pixar to release the engineering documentation for this design.   It might not have been a commercial success, but as a significant piece of computing history, it deserves to be remembered.   See my blog entry [XXX] here on a place which could host this documentation.

25. A Night under the Stars (Short Documentary)

http://users.bigpond.net.au/alanrich/doco.html

At the start of this blog post, I bemoaned the disappearance of the short at Burrill Lake Open Air Theatre, then the theatre itself.   Just before posting this article, I was therefore deeply amusing to discover that someone has made a short film about the old theatre.   Yep, they’re selling it via the Internet.