Tuesday, October 31, 2006

The Manufactured Image: Successes

Just so you guys don’t think I’m a hater or anything, here are some scenes that I think work pretty well. They may not all be styles or color themes that everybody loves, but in general they’re quite sound and you can make compelling arguments that they have a structure that works.

cars-ok 021.jpg

cars-ok 01.jpg


IA-OKish 02.jpg


nemo ok2.jpg

nemo ok.jpg





openseason ok3.jpg

openseason ok.jpg

openseason ok4.jpg

robots ok.jpg

robots ok3.jpg




nemo ok3.jpg

See? I don’t think it’s all doom and gloom. Scenes can come out looking quite nice. Perhaps they were key scenes for the sequence and thus got a bit more attention? Maybe they’re just happy accidents? I do think some of them had a clear strong sense of design right from the start and thus the elements just worked. Some of them were finished relatively early in the production schedule to make them fit into trailers, thus they may have had the time and freedom to go over them and make some adjustments, a luxury that could not be afforded when the other 1200 scenes needed to get done under a crunch. Whatever the case, good art does happen under the current paradigm. It can’t be helped, there is a ton of talent and skill in every one of these studios. You can’t keep that from sneaking through.

Monday, October 30, 2006

The Manufactured Image: 25 Assembly Accidents

Continuing our discussion of the manufactured image, I’m not willing to go the easy road and say the artists, animators, technicians, designers or whomever just suck and that’s that. It’s not that simple, and more so, it’s not true. For me this is a discussion about what I think are some core problems with how stuff gets made in the manufacture/assemble paradigm. Personally if I’m going to be ballsy enough to call out a whole system then I need to do better than pick on one image. Thus in an effort to back up my ideas I want to show that the problem isn’t limited to just Shrek 3.
I’m sure on this point many will disagree with me, but here are some stills from various recent CG animated films that I think are good examples of what I call assembly accidents. I just grabbed these off the net after doing some quick searches. In doing this I am not trying to say that I think these films are bad, nor am I saying I could do anything in them any better. And it isn’t my intention to call out any individual artists or TD’s. For the most part the individual artists did very fine jobs on their specific task or element. In some examples I actually know some of the people involved and I know they did their level best they could and made some nice parts- they just weren’t responsible for the whole. I doubt anybody really was- thus the accidents.
No, these images are merely examples of the kind of visual accidents that are bound to happen under the current manufacture/assemble production paradigm that rules big budget Cg films. And this is not to say that these images aren’t good, either. Some of them still work pretty well. It’s just they have a few things that aren’t clear due to the accidents of assembly.

Check out the examples after the jump…

1) Where does the squirrel end and the rocks begin?


2) The bgrd directly behind the characters is unfortunately the most visually complex part of the bgrd. The lines all muddle the performance. Shift the characters to the screen left a bit on top of the quieter visual space there and this would be better.

3) Too many competing lines. Try the “Squint Test” (clear images work on a macro level even when viewed all blurry)


4) Color similarities between characters and background makes reading this a bit tricky.

5) Ditto, especially for the boy and the paper.

6) Top of image is ok, but the ant emerges from the boy’s gut like “Alien” due to tonal similarities of armor and ant color.


7) When saturation and tonal hegemony ruled the land


8) No comment.

9) Not bad, really. It’s a nice image. But how about if you put the yellow character over the blue bgrd and the blue character over the yellow bgrd to help them separate better? Plus the smaller character in front of the tire sculpture would be clearer. Right now it looks like there are tires stacked on top of the yellow car.

10) The crowd overpowers the scene, making it difficult to read the ball in the player’s hand. Tinting the whole crowd a similar hue to push them back and unify them into a single idea of a crowd would help.

11) Another case of a nice layout that could have been a little stronger. Put the darkly dressed boy on the left where the bgrd is light and the brightly dressed girl over the dark bgrd to the right. Just to add balance and separation from the bgrd a bit.

12) A little hard to read what’s what. Squint test.

13) Under the squint test Dory almost disappears.

14) I gots me the blues, baby. (foreground and background mush)

15) I generally love the look of the Ice Age films, but even they suffer their share of assembly accidents. In a lot of ways this shot works well, but the foreground character disappears into the mammoth. Silhouette on the armadillo character is nicely done, though.

16) Foreground works very well, but the darker shading lines in the background compete and disrupt the flow.


17) Trees behind E compete with her dark silhouette. Tree behind Bob crowds his face and breaks the staging space between the characters. Great pose on Bob, tho.

18) Another one so close. Mr. Incredible? Great! Syndrome? Lost.


19) Visual overload.

20) TMI.

21) Where does Rodney end and the machine begin?


22) The disappearance of RJ. AKA: Put the SUV in the garage.


23) Another close one. I like the facial expressions here- very appealing. But the brown squirrel over the brown log when you had a bright oppossum to put there instead? Assembly Accident.


24) Foreground is background part 2. The color tones flatten this one.

25) The Mother of All Accidents. There’s some really fun stuff going on here. You just have to find it.

I’ll stop there since (I hope) the point is made. Again, this isn’t a slam on anybody (although I am sure some will feel that it is). There’s some really great craftsmanship in just about every individual element shown here. It’s just that once all the elements were assembled they just don’t work together quite so well. Some images work as is, but could have been just a leeeeeetle bit better. So no death threats, please. I’m just trying to point out a flaw in the system.

The Manufactured Image: A Microcosm

For a primer, read my previous post.

Let me focus on one problem that will be a microcosm that mirrors the larger issues at hand. As I noted yesterday one of the things that big budget CG imagery lacks is a sense of connectedness. For years called “stiffness” or “coldness” by reviewers, CG has had a difficult time warming the eyes of it’s viewers. It’s not universally true of all CG projects. There are some smaller groups, short projects or individual artists who are doing some very fresh things artistically in CG, so the problem isn’t the medium. And I do think occasionally feature films do a fair job of creating pleasing imagery for many scenes (I’ll show some in later posts). But the klunky nature of the imagery in CG happens much more than it ought to. Why? As I said in my previous post, I think the system in which the images are made has flaws- it is a manufacture & assemble system. And even how we pose and animate characters in CG is a micro-mirror of this problem- the assembled image is made from characters whose poses are assemblies of many controls and parts.

For example, check out Ogre Fiona’s face, arguably the better of the bunch in the promo image from Shrek 3.


Yet it doesn’t completely work for some reason. Why? Well, it’s two emotions on different sides of her face. This side says one thing…

fiona R.jpg
And this says something else…

fiona L.jpg

Let’s examine how one creates poses and animation in CG. In CG every part of the facial control system (in fact every system) is isolated. That current paradigm has been necessary so that you can get the detailed control you need to get the shapes you need. But ultimately these things are just very complex articulated puppets with many restrictions to their ability to change shape. They have more variables to control than any other form of animated character. The most complex stop motion puppets have a miniscule fraction of the control of Cg puppets. Handdrawn characters are as direct as the tip of a pencil. Clay characters are totally malleable directly with your fingers. Animating Cg characters is a bit like handling uranium. You touch them only through thick gloves inside a box. One friend of mine, a 10+ year Disney veteran animator, described that his experience animating CG characters was about as easy and elegant as “drawing with boxing gloves on”. So we have these nifty digital toys- highly detailed, almost infitely subtle- yet burdened under their own complexity. Again bright minds are working to install a new paradigm of creating shape and form with these puppets, but for now the current system reigns.

The reality is that which is so easy to get in one clean line in a drawing is accomplished by an arrangement of multiple rigid elements in CG. So how we even pose the characters is an assembly of moved parts to create a sense of something alive and breathing. The best CG animators can rise above this limitation, many others lack the experience to do much more than get all the kinks out of their motion. With so many disparate and isolated pieces to move it’s a huge workflow challenge to make everything feel like it’s moving together, motivated by will or emotion and with purpose (this one fact alone accounts for perhaps 90% of the cost of cleaning up CG animation.) Many younger and less experienced animators working in film (and there are a lot) are doing all they can to just get the basic stuff working right. Never mind thinking about a higher plane of existence. If we could find ways to allow animators to work at the speed of thought on creating these performances and set them free from the assembly or parts paradigm then we can tap into a wide talent pool that currently is stifled by the technology.

Sunday, October 29, 2006

The Manufactured Image: Henry Ford would be proud

OK, kids. This is where I put on my crumudgeon hat and in the process no doubt make some people upset. Sorry, but it’s not my intent to make anybody feel bad, but instead try to point out some fundamental issues with the system. :)

OK, let’s have a look at this….

shrek3.jpg (click to enlarge)

I’m sorry, but I can’t get drunk enough to make this look good. Without a doubt this is a forest located deep down in Uncanny Valley. But this is merely a continuation of a problem that has been evident for some time. I’ve been saying for a while that big budget CG has been lumbering under the burden of it’s technological roots. It is difficult to deny that the current bent of big budget CG is toward over detailed and klunky visuals. There’s a lot to take issue with here. The lack of contrast, few quiet areas in the image, jangly poses, cliche’d layout, the haphazard accidental relationship of the background with the foreground, etc. It’s almost like nobody ever saw this all together until it was too late. The thing is, if it was made like 99% of the imagery in big budget CG then most likely nobody did see it until it was too late. The problem is not so much with any single artist. That’s because in all likelihood no single artist is responsible for this. It is assembly line imagery. The flaw is in the system under which this is made.

Imagine taking 10 talented solo singers and asking them to sing the US national anthem to the same instrumental track. But due to scheduling conflicts they have to each perform in solo, not as a group. Oh, and gee, we don’t have everybody’s performance here yet so you’ll need to just do your part the best you know how without hearing the others. Naturally these singers are to going to make it the best national anthem they know how. So they sing and sing, beautiful notes that rise and fall- all creating fabulous solo performances. Now take these 10 solo artist’s performances and mix them together in editing. The overall result would be hideous. There are no background singers, nobody is doing harmony, nobody takes the lead because all take the lead. It’d be like some kind of gladiator battle of voices. The jumble of notes flooding forth would cause ears to bleed.

This is the musical equivalent of CG feature film production. Read why I think so after the jump.

It is apparent to me that the production pipeline itself contributes to this sense of disjointedness we see in many scenes of CG feature films. In my experience as a Cg Supervisor on several long form projects it is clear that large CG animated projects are run more like a factory than anything else. Before becoming an animator I worked for almost 7 years in a wide variety of factories. I know factories. CG film production is a factory paradigm. These various CG assets are built in individual tasks, much like manufacturing widgets. The idea is that in the end all the various widgets are assembled to make a single image. So in many ways it’s no surprise at all that the end result feels haphazard and disjointed. It’s not the fault of the individual task-doers. They’re doing their best within their small isolated task (and small window of information about the whole) to make it the best they know how. Like the solo singers in the above example we do our best to make it a kick-butt (insert asset type here). And increasingly in the world of CG artists the measure of success or ability is in the artists/technicians ability to master details. It’s a micro focused paradigm. Left to our own devices individual Cg artists will over-detail things. It’s what we know and it’s what the CG world celebrates as having higher quality. I know because I’ve done it in the past myself, so you’re hearing it from one of your own. Thus the disease of detail-itis is built into the entire economy of CG production almost by default. There is little macro focus on the relationship of all assets in the creation of the image. Nobody can see the whole elephant at once, so we make our pieces based on our belief about what kind of object we think it ought to be (snake, tree, rope, etc.). But let me be clear on this- Detail-itis isn’t an unbalanced love for details over macro view of art. It’s merely the result of an inability to even be allowed to think on a macro level. Thus we focus our energy where we are given permission to put our mark on the art- in the details too small to be managed.

In a CG film production you usually don’t see everything all together until it’s almost done. More often than not the asset tasks are assigned with little information about how these individual elements fit into the larger collective whole of the imagery for which they are being created. When a modeler makes a lamp he is not shown all the scenes in which that lamp will be seen, where it appears in screen each time, how close we are to it, etc. Similarly animators have no information about the tonal qualities of the background to work with and zero info about the lighting or FX usually. Texture artists are given modeled props to paint, again in a vacuum. And so it goes. Lacking an understanding of context it is impossible for the individual task artist to make well informed priority judgments. The approval system isn’t any better. The director (via an endless string of departmental meetings) is shown models, environments, textures, materials, lighting, animation, cameras and FX all in different meetings. There is a significant gap of time where a director (or art director) on a CG project never sees the image with its elements all together. Once he approves the concept image or the color key (which are painted traditionally and usually not accurate representations of the true final look of CG) it is all broken apart. He or she just won’t see everything together again in the context of a given scene in the film until months later when the first lit and rendered scenes roll off the render servers. But on just about every production by the time they get to see everything together (usually lighting dailies) the production schedule is in crunch mode. Thus all they’re usually able to do is fix the technical problems (hopefully) and get the frames rendered without mistakes and with some measure of consistency. There’s no time to change sets or camera angles or animation or models or textures or anything to make the scene work as a piece of art. It’s all most crews can do to get the darn thing rendered, assembled, comped and done on time. Because of the huge cost involved lighting is not the place to be making changes, but ironically it’s the first place where you have all of the pieces ready put together. It’s only here that one can begin to arrange these elements to create pleasing singular works of artistic imagery, but the ability to do so is a luxury that cannot be afforded. The paradox is staggering.

Artistry isn’t assembled. Cars, disposable razors, radios & lawn mowers are. The primary benefit of manufacturing is consistency of product combined with cost control. Is it any wonder then that many CG films feel the same? To combat this somebody needs to be allowed to think about the film from the standpoint of appealing artistic imagery. And they need to be able to do this before the money and time runs out. This is supposed to be the director’s or art director’s job, but in CG a director is often little more than parts inspector- whether by chance or by choice.

No doubt this film has a ton of talented & skilled people working on it. All CG films do. Great modelers, texture artists, lighters, animators, etc. All people who’s skills put them at or near the top of the heap. I have close personal friends working on this and I know for certain that they’re way better artists than this- when they are responsible for the whole image, that is. But in the current CG film production paradigm they’re not allowed to be. It would seem no one is.

So besides complaining about this, what do I have to offer? Well, I’ve got some ideas. I’ll share those as I go along. Stay tuned.