Wednesday, February 3, 2010

#0030 The Best CG Production Pipeline

I’ve been fortunate   to have worked for a number of companies during my career, either as a 3d artist, a compositor or a supervisor.  I've observed that while many companies have similar CG production pipelines, the structure does vary.  Further, over time, pipeline structures change.

Today we are back to the subject of pipelines with a look at some of the forms a production pipeline may take.  But first a quick review of the nature of CG pipelines:

Previous articles focused mostly on helping my readers see pipelines from a broader point of view.  I shared an epiphany I had a few years back that there are in any given project three classes of pipelines: production, material and approval (posts #0018, Understanding Pipeline, and #0019, Form Follows Function).  These coexist on a project, touching here and there to exchange information, but functioning in parallel, mutually dependent yet separate from one another.

While some may think that the three are really one pipeline, I see them as separate yet vital, largely because when one examines the three dimensions of the pipeline: personnel, tools, and procedures (post #0020, The Technology Is Not the Pipeline), it is evident that in most cases the personnel, tools and procedures for each class of pipeline are unique and separate.  While personnel may use tools and procedures from more than one pipeline, the structure and management of each class of pipeline is independent from the others.

Flussdiagramm (Programmablaufplan)
I differentiate between work-flows, the steps in a task done by one or more people, and a pipeline, the organization of personnel, tools and procedures to facilitate a work-flow (post #0022, Workflows vs Pipelines). In the same post, I introduce a concept of a Tree of Specialization to show that a work-flow can be divided and redivided into ever greater degrees of specialization.
The first five articles in this series brought us to elaborate the Seven Point Functional Description of Pipelines (post #0023, CG Pipelines Defined).  I'll put the description here again for convenience, but if anything in it is unclear, I suggest you read posts #0018, #0019,  #0020, #0022, and #0023

Seven Point Functional Description 
of CG Pipelines
  1. A CG pipeline belongs to one of three classes: 
    1. production  (task)            primary
    2. material      (data)           secondary
    3. approval     (meta-data)   tertiary
  2. A CG pipeline is comprised of three structural dimensions: 
    1. personnel
    2. tools 
    3. procedure
  3. A CG pipeline utilizes technology but is not the technology
  4. A CG pipeline divides a work-flow into separate and meaningful tasks assigned to two or more persons
  5. A CG pipeline tasks divisions are determined using a tree of specialization across the three dimensions
  6. The CG pipeline structural form is dictated by 
    1. the functional mission,
    2. resources available and 
    3. company culture
  7.  A CG pipeline is malleable
Four Forms of Production Pipelines

Any time two or more people get together to do a task, some system of cooperation and division of labor occurs.  If a couple of kids get together to do their chores, this may be completely ad hoc, with the dominant personality often suggesting to the other what each one does.  A group may get together for a little basketball, and again, they will structure their team to either play zones or one-on-one.  In many sports players often specialize: a baseball pitcher, a quarterback, a goalie.  Likewise, as the person responsible to run the crew, the CG Supervisor must determine the best form for his team. 

Many people working in CG assume that organizing the team, the CG production pipeline, by functional specialty is the only way.  While some people will argue the best form divides the work among specialists, working alone or in teams, others may argue that a good pipeline is flat, with strong generalists who can insure contuinity of design vision through a project.  It's sort of like the difference between zone and one-on-one basketball: the zone player focuses on developing skills and habits for playing a position, a functional specialization, while the one-on-one player focuses on his opponent, developing a project orientation, while being able to generally play anywhere.

Others will say the best organization is to structure the team around the technology. Another model is to break up the pipeline according to artistic styles, for example, photo-real vs. 3d cartoon vs. stylized, etc.  

These four organizational forms: project generalists, functional specialization, technological specialization and style specialization may represent the range of possibilities, but other forms may exist outside my experience and imagination.

Each of these four forms has it’s advantages. Perhaps because it is the most common form, many people erroneously think that a pipeline must be based on specialization of labor or function. In a large company, with 30, 100, or hundreds of artists working on perhaps one project, specialization provides a convenient way to divide the work-flow. As time passes, this specialization increases. For example, “particle systems” (dynamic physics-based simulation) emerged as a specialization around 15-20 years ago. Today there are specialists in cloth, fur, hair, fluids, smoke, fire, explosions, destruction, etc. There are specialists who model hair, descrribing it’s shape, position and length,“combing” it into shape, and other specialists who animate the hair. Within the specialty of surface modeling, there are specialist in organic and inorganic modeling; other modelers may specialize in vehicles, architecture, environments, weapons, creatures, people, and so forth. Artists may specialize according to subject matter, techniques, style, skills and tools.
Specialization according to technology may sound foreign, but it’s quite old, and still quite common today. For example, by the mid 1980’s the CG specializations (let’s not get into non-digital visual effects production specialties here) were four: paint, 3d, real-time video, and compositing. Paint or “2d” involved use of tools where the artist “painted” directly on a digital canvas with limited or no animation capabilities beyond traditional cell techniques. “3d”, involved using 3d computers to indirectly model, animate and render shots. Real-time CG involved the use of tools that could move one or more video polygons in real-time, one system could shape the video texture into 3d objects and morph them. Compositing, as today, put it all together, but the best tools were dedicated compositing systems. A fifth specialty was essential to support these: the programmer. Today, this tradition of specialization by tool continues and is evidenced by job ads for artists who use some certain software, such as Maya, Houdini, Nuke, After Effects to name a few. (There are hundreds.)
Specialization by Style could appear less evident in a CG production pipeline, but it’s actually the first determinant for a studio in designing a pipeline: is the film a visual effects film, with more or less invisible (or believable) effects, like Harry Potter or Lethal Weapon? Or is the project fully animated, like Shrek? Is it 2D, like Ants, or 3D, like Avatar? Whether the project is a film, game, TV documentary or corporate video, the style of the project will ultimately drive what CG production company may get the job and will determine how the CG production pipeline will be structured.
Specialization by Project-Oriented Generalization. Huh? You may rightly ask how generalization can be a specialization, but it is. Suppose your designing a production pipeline that must meet rapidly changing demands, say for a weekly TV show. You may be able to anticipate some specialized needs, but you may need to have mostly generalists who can pretty much take on any kind of problem. Or suppose you have a small boutique with a low volume, you might like to crew with generalists because you don’t know what you need.

The Very Best Form of Production Pipeline
 In actual practice  these four forms of production pipelines are usually not exclusive. They often co-exist. For example, a film CG production pipeline may be comprised of artists capable of producing within a specific style or genre, using the technologies currently owned by the company, organized according to function. A video CG production pipeline may be comprised of a core team of project-oriented generalists, supported by a few specialists (who may be hired ad hoc as needs change).
The best form of CG production pipeline?  

The best form of CG production pipeline is the one that can best fulfill the production mission with the lowest investment in new personnel, tools and procedures.

Tuesday, February 2, 2010

#0029 A Very Busy Winter

Hi all!  
Image via Wikipedia

Universal Studios Uniglobe 
After a 10 week hiatus, I've begun writing again, and the topic is CG Production Pipelines.  During these weeks I've been hit with the cold-flu bug three times, forced to visit both Universal Studios Tour (Hollywood) and Disneyland, trampled by dozens of visiting in-laws, settling and unpacking from my forced migration in September, and between running kids around to school and this and that and the wife to work (and this and that), there's been looking for work.  

On all fronts except finding work its been a great 10 weeks --I recovered (three times), enjoyed Universal and Mickey Mouse, survived the in-laws (who generously made sure all the fun wasn't a financial burden) and burned a lot of oil (and got the second car fixed three times).  I'm sure your lives have been just as hectic, and there were several holidays along the way to boot.

I've reviewed my posts on pipelines and have dived into the matter, getting at last to begin talking about CG Production Pipelines.  That will be in the next post (#0030), but first, let's talk a bit about some fun and not so fun stuff.

    Visual Effects Society events
During December and January, I attended three events of the Visual Effects Society, of which I am a new member (it took me a long time to turn in the application, but well worth it).  First was the VES party in December.  This was held at the ZOIC Studios in Culver City, near Los Angeles.  It was a great event and I really enjoyed seeing so many friends in the business after a long time.  

Question mark in Esbjerg 
The second event, in mid-January, was the the judging of 2009 VES Awards nominees.  This was an awesome event, held in Sydney, London, Vancouver, San Francisco, and Los Angeles.  At each location, select groups of VES members screened and voted on the submitted work.  Each group was comprised of members who had no connection with the works in that category.  It was a long day for us in Los Angeles --from 8AM until around 6 PM, but well worth it.  Our group reviewed six categories, and it was a real treat to see all the works submitted.  The state of visual effects ART is very good indeed.

A week later was the Los Angeles VES Minifest. This was three separate events in one location, one after another. In the morning, we were treated to a panel entitled, Prosthetic Character Work in the Digital Age. This panel looked like a bust 10 minuted before it started, but suddenly about a zillion people crammed into the theater and we were treated to an informative and entertaining discussion.  We saw how important prosthetic work is today --perhaps more so as film audiences continue to demand more and more exciting and incredible effects. (For those unfamiliar with the term "prosthetic work", this does not only mean artificial limbs, but includes make-up, other body parts, and puppet-controlled characters, like the shark from Jaws.)   The compositing of prosthetic work using vfx rotoscope, 3d models and other digital techniques likewise continues to be crucial to the process.

The second session of the minifest was a screening of selected VES award categories.  The VES Awards committee selected six categories (out of some 20 or so) and screened the nominees.  I was a judge for three of the categories, and it was really interesting to see which of the many potential nominees our balloting selected.  It was also great fun to see the other three categories.  I can only say that the caliber of work was stunning.  If you can go to the VES Awards dinner next month, do so.

The Ghostbusters, (from left) Spengler, Stantz... 
The highlight of the evening was Who You Gonna Call? Special 25½ Year Anniversary Ghostbusters
Screening and Panel Discussion.
  I was tempted to skip this, it sounded like it might be a bust, but I grabbed a fish taco during our dinner break and wandered back to the theater.  I was really glad I did.  The panel, with some 12 or so of the visual effects auters, including Richard Edlund, was quite entertaining.  Edlund told us how BOSS FILMS was founded in the process, and how ad hoc everything was --for example, the camera they needed to film their work didn't exist, so they had to "invent" it and build it during production.  It was great fun hearing all the stories, which took me back to my own  professional CG beginnings, which I realized as I sat there were 27 years ago, to the day.   Afterwards they screened the film Ghostbusters,  which I always enjoy.

     3D using "2D" tools

The other event I attended in December was the L.A. Siggraph Meeting, which included several presenters, ending with my friend Steve Wright , a terrific compositor and compositing instructor, who spoke on using Nuke's 3d tools. Steve's website:

This was quite interesting to me, it's what caused me to trek the 70 miles down the mountain to UCLA, because I've been using 2D tools to do 3D for many years, and because I'm in the process of cross-training from After Effects to Nuke.  

Aside from the ability to displace a polygonal mesh using an image source, all the functions demonstrated as great Nuke features were old hat to me with After Effects.  However, the ease of use and 3D display seemed more powerful (although it may have been the hardware).  Steve's presentation was really good, he made it really simple for the audience.  

I was mostly amazed at how "new" this was for so many people, but then I realized that this is something I've been seeing since my video days, when we did this sort of stuff using real-time video manipulators like Ampex's ADO and Quantel's Mirage.  Flash forward 27 years and it's not real time, but it's also not video.  It looks and works great.  I look forward to sitting down to this tutorial later this week to compare for myself how NUKE and After Effects each handle 3D.

A Few Awesome Films

I finally saw 2012, Nine, and Coraline on my VES screener DVD's.  At the VES Awards events I attended I saw just enough of these films to want to see the visual effects.  All three are truly awesome, and each represents a totally different visual effects genre.  I look forward to seeing them again.

As soon as the wife gets a free minute, but no later than this week, I'm off to see Avatar.   I saw a few clips of this at the VES events, and it's a must see in 3D.

That's been my 10 weeks.  I hope yours was productive and stimulating.  Mine was great, except for that job thing.  If you know anyone who needs a 3D generalist powerhouse, a master compositor, or perhaps a CG supervisor (I know a little about that), please send them my way.  I'm really tired of Facebook games. :)

Have a great day, and I hope you enjoy my next post, #0030, The Best CG Pipeline.