Attached Link: http://www.jrdonohue.com/glow001i.mov

This little animation represents a quest for a stylized look. Animated and rendered in "realism" mode in Poser6, I took it out to After Effects and applied a few filters, especially "Glow."

I tried the Van Gogh plug-in effect for watercolor, not convincing at all IMO.

I don't know...the quest for hyper-realism and full motion for the small studio...I think it is impractical, even rendering in Max or Carrera on a farm. Yes, you can go "pretty far" given today's shaders and various GI, but to REALLY convince in realism, you have to push each frame so far, and be so meticulous, because every flaw tells, breaks the suspension of disbelief. Even at DVD frame size (don't even think about full film size) the render time is very high.

But more important is the time taken tweaking, experimenting, re-rendering, etc. to make every frame convincing, in full motion.

I am starting to think it is a non-starter. There have been threads on this forum and CGSociety, for instance, that make the point that even with maximum effort and expense, such as in Midnight Express and FF, you "can't get there from here" with animation posturing as live action. [And with a nod to Peter Jackson...verisimilitude with CG is MUCH easier in action scenes than in human character close-ups and conversational interaction.]

To put it another way...."Just go get actors and a SONY prosumer CAM and go shoot the damn thing for real!"

Leaving that aside for a moment, and not being ready to move into what I call many-stops animation, where you essentially have a static background over which a small amount of animated action takes place with many fully-stopped frames, I am beginning to play around, not with toon, but with stylized versions of realistic character action. IMO this clip does not make it; it is not stylized enough. It just looks like attempted h-realism with effects applied. I am going to keep working to find something genuine.

::::: Opera :::::

"verisimilitude with CG is MUCH easier in action scenes than in human character close-ups and conversational interaction." Gollum coversing with himself didn't do it for you? admittedly not human but an entirely believeable character also..stylized & realism dont exactly go hand in hand, the best 'realism' I've seen is in films where your not meant to notice it's cg.& it's usually after I've seen the film before finding out whats cg do your best to make your character believable(not realistic) in animation(an art in itself)..then try making it look pretty with all the textures & fancy lighting.having the most realistic skin you've seen is useless if the character animation isn't up to par. the best part of your video is the subtle movements in the hair,(not so much the extreme movements) it's hard but not impossible to be a one man band making cg films..didn't 'sky captain' start in somebodys bedroom? Cheers

I've seen some animations that almost had me fooled, but they weren't full-length movies. Gollum in LOTR had me totally convinced it was a real character as I watched the movie, even though I knew all along that he wasn't. They did such a good job with his motions and personality, it didn't even matter if a few shots here and there didn't look totally realistic. As for your sample animation above, I think it's an interesting effect, but it does look cartoony, especially with that drop shadow effect around the character. The glow looks good, but then there's a darker outline that makes it look a littele strange. Also, I think the hair moves around too easily. It seems even the slightest movement is making it wave about, even after she stops moving, which seems unrealistic. Hair may not be heavy, but it does have some weight to it. I wouldn't worry so much about making it look realistic, and concentrate mostly on a good story, with good animation.

As they say, sometimes the perfect can be the enemy of the good. I don't think anyone on any budget has yet created a regular, ordinary human being (as opposed to a fantasy character) that is indistinguishable from reality so if you are able to come anywhere within shouting distance, you're doing great? I don't think the computer horsepower or the software algorithms exist yet to make it practical or even achievable, close perhaps but not there yet. If what you want to do is create something that speaks to people then I agree that you should concentrate on story, character, personality etc. Remember, people still cry when they watch Bambi :-)

Attached Link: RENDERED IN CARRARA

Attached Link: http://www.panoscan.com/PanoPress/2005Press/Lemony/LemonySnicket.html

justpatrick, I was going for the very wispy hair, and may have gone a little too far. I'll try more weight. I am also looking into clothifying hair groups instead of going for full strand dynamics. Also, I discovered am issue with dynamic hair. If you have collision turned on, when the head is nearly still, but affected by very subtle "naturalizing" movements, the collision detection over-reacts and makes hair strands jump. Your comment about going for the story and animation is very good advice. My attitude is on that same track, but I am thinking of pulling back in some stylized way from "approaching realism" to "signal" to the watcher that I am not pretending to verisimilitude. That will remove the distraction of the audience "seeing how close he can get". As I said, though, not all the way back to "many-stop" partial animation. Thank you for your response. ::::: Opera :::::

"That will remove the distraction of the audience "seeing how close he can get".

You're too worried about that, John. The audience doesn't care, unless they too are aspiring CG artists. ;-)

The reason John Woo's 'Hulk' movie was a bomb wasn't because the CG character had failed in appearing to be real (although arguably it did); it bombed because the movie sucked overall. 'FinalFantasy:The Spirits Within' didn't tank at the box office solely because the characters looked "too real"; it tanked because the story bored most audiences to death.

"I am thinking of pulling back in some stylized way from "approaching realism" to "signal" to the watcher that I am not pretending to verisimilitude"

If you're not trying to pass the entire movie off as being real, then the audience isn't going to need a "signal". Just tell them up front that the entire movie is CG, and you can get away with any rendering style you like. As mentioned, what will capture and hold their interest is the plot and animation themselves. The "style" of rendering is merely supplimental to those elements. Message edited on: 12/07/2005 09:26

hmmmm.

Abso-frigging-lutely what maxx said. If the script tells a good story, that engages the audience (regrardless of length), then the issue of 'realism' becomes more one of 'styleism'. Probably the most important point in all of it is to -never- break the viewer's willing suspension of disbelief. That holds true with textual or visual media. Final Fantasy:TSW was more concerned with showing how incredibly cool Aki Ross's hairstyle was than the actual coherence of the tale (the cookie cutter characters didn't help either). Look at 'The Incredibles'. Pixar being the house that did the CG didn't make the film; it was the script, the characters, and the storyflow. They did it right, in that you forgot it was a CG film with SSS and strand based hair and displacement mapping and multiple armatures and you see my point.... ;) Why don't you take a look at the bookstore in the writing section? There's a Writer's Digest book series called 'The Elements of Fiction Writing', with book titles like Plot, Description, Setting, Dialogue, Scene and Structure, Revision and a few others. They are general concept and idea books by different authors, not 'how to' cookie cutter manuals that would lead you to serious screwing up. And the ideas apply to scripts as well as novels or short stories. You =have= to have a solid foundation, or all the slick CG tricks are just going to be a waste of time.

I've got the story. It is abso-friggin-lutely great! That part is done. Edited. In the can. Sorry I did not clarify that to begin with in this thread. But it is all character interaction, boy+girl with no fantasy, no war, no car chase, no supernatural, no big city scenery or fly thrus. I also have the music waiting to be fit into place. I have decided to produce the first scene in normal realism and simply take the advice to not worry. Really, my only concern is that the actual shortfalls of cg will in an of itself break the suspension of disbelief. An animated film about relationships? In the past 5 hours I have been going deeper and deeper into After Effects, and if I deem it necessary to "back off the realism", the tools there are very up to the task. ::::: Opera :::::

"An animated film about relationships?" yeah,why the hell not? "it's easy if you try" :) http://www.39poundsoflove.com/main.html

That film looks very cool. It seems like an interaction between his live journey and animation. If he can do it with one finger, that puts all of our issues in perspective. ::::: Opera :::::

The Snickets stuff is impressive. Arguably, you can achieve a high degree of realism for short sequences. I'm thinking more in terms of a sustained performance with the normal range of camera setups and scenes you'd find in a "regular" film, if that makes any sense. In artificial intelligence, they have the so called Turing Test: A person is connected to both another person and a computer via a terminal. The person tries to distinguish which is which soley by asking questions. If the computer can answer arbitraily posed questions without giving away its non-human nature, it is deemed to be "intelligent." Perhaps we can formulate a Turing Test for CGI.

"Really, my only concern is that the actual shortfalls of cg will in an of itself break the suspension of disbelief." I see what you're saying here, but I think you're looking at it far too intensely. While I do believe it's true that if one is rendering a sequence in a very realistic style (photorealism), then the animation should hold up to that realism as well. This isn't entirely about "suspending disbelief", it's more to do with CONSISTANCY, and flow. If the animation is done in an exaggerated, cartoony way, but the visuals are done in a photorealistic style, then most audiences will get confused, and it WOULD potentially become very distracting (then again, it's never really been done before, so who knows?) ;-) What I'm trying to say to you is this: Don't fear that if you render in a realistic style, to compliment your particular story's genre, that you will lose the audience's respect or interest if it's not absolutely believable in every single frame!! That's nonesense. If they are interested in the story, and the animation is good, and the dialogue is believable, then they will continue to follow it, and not be distracted. As long as your target audience isn't a bunch of animation geeks like some of us here (me included), then they won't be overly critical of the visuals, so long as the rest of it holds up. Trust me on that. What matters most is continuity and substance.

This is a great thread! Maxxx, I've been watching your animations in the animation outlet and the animation forum, and find your work very inspiring!

Thrashing this out is helping me, thanks. I was making myself
stuck and knew I had to throw it out for reality check (forgive the pun.)

I'm leaving on the 21st to visit family and hope
to have a finished sequence to show them, so
I'll post more in the next few weeks.

Here are four stills. Click on thumbnail for full frame.

The rear shot with the furniture has some very odd lighting effects the back of her head seems to have a luminescent glow and there are no ground shadows at all or perhaps this is an intentional "stylized" look ? and her feet positioning seems rather strange and would be difficult to maintin balance for very long. also was that purple dress/shirt item draped with a cloth simulation? if so you may want to adjust some parameters and run it again becuase it looks very rigid and is not laying flat. the lighting on the 2 bottum frames seems very flat a little "washed out" and has very little contrast what are your minutes per frame times?? i imagine you are likely having to make serious lighting/texture compromises to get them reasonabl***(less than 10 min per frame)*** using poser a render engine but as already stated good story telling will prevail over gratuitous "Dynamic" hair simulation good luck

Looks interesting so far, John. Are these mockups, or finals? Will you be compositing shadows in post?

Greg, not finals, backdrops and shadows on separate run, will be compositing. Still working out relationship of difuse IBL/AO and shadows cast with secondary lights. This scene has extremely bright low angle direct sunlight in one corner while quite dark in the other, but with some rays finding their way across (there's a mirror), such as the one striking the back of her head. I am rather fond of unusual/extreme lighting. ::::: Opera :::::

"what are your minutes per frame times?? i imagine you are likely having to make serious lighting/texture compromises to get them reasonabl(less than 10 min per frame) using poser a render engine" I'm interested in this one too. "I am rather fond of unusual/extreme lighting." As am I. I have to touch on something you stated earlier: You said, "I don't know...the quest for hyper-realism and full motion for the small studio...I think it is impractical, even rendering in Max or Carrera on a farm." What is it about realism (or hyper-realism, as you put it) that seems impractical? Is it the lighting? Rendering time? The laborious animation that is required? In my opinion, the first two problems of realism in 3D (lighting and rendering) are solved, or at least close enough to fool most people. We can absolutely achieve a photorealistic look to our animations these days that matches live action, using certain proven techniques, and rendertime isn't much of an issue anymore either. In fact, I'm rendering out an animated sequence right now using HDRI lighting and subsurface scattering, and I'm seeing rendertimes of a mere 9 minutes per frame at high resolutions, with no worries about flicker or artifacts. The render speed is aided by a technique known as "texture baking", where you run a GI pass only once, and it literally bakes the result to a texture map (or light map) that you then apply to the model or scene. I'm doing this for the environment, and then rendering the character pass seperately, using a fast hybrid GI technique. This provides a very realistic lighting solution that can be used reasonably in animation. I should be able to post a sample of the result by tomorrow. On the other hand, in order to get TRULY realistic results, you are correct in assuming we must be meticulous about the task of animating itself. Every little muscle flex, or ambient movement on our characters add to the illusion. That's the part where I think CG is still evolving, and hasn't yet reached the point where it's absolutely indistinguishable from reality. However, if you take care in doing the best job you can do while animating your characters, then the illusion is definitely enough to suspend disbelief in the general audience.

1 IBL light, one spot, some AO on skin, min shading.084 pixel samples 5. Kate Hair (about 22,000 strands) thickened. Hair "not visible in raytrace". average render time 2.89 min per frame, plus perhaps 35 min for hair simulation. The issue is not render time...it is exactly as you say, "we must be meticulous about the task of animating itself. Every little muscle flex, or ambient movement on our characters add to the illusion" and I often want to desaturate to make it B/W all the way, or to go toon, just to get the distraction of shade and color out of the way, to go minimal, the better to concentrate on the animation and making it believable. This would bring the story and music even more into the foreground. By "impractical" I mean that there are too many areas to worry about getting right, it might be a relief to abandon photo realism. However, right now I am trying on the idea you propose in your last sentence. Dial up the realism best possible then "let go", don't fiddle, concntrate on animation, won't matter how "hyper" real the render is. Thanks. ::::: Opera :::::

looking at the posted animation itself i see a lack of ambient secondary motion. when someone lowers thier head and shakes it like that there is always torso/shoulder movement that occurs with it.right now her segregated head movement looks rather robotic like a disney animatronic rig.

Good observation. I removed that motion when I was struggling with dynamic hair bounce with collision on. I forgot to restore the movement of her shoulders/torso. It has been restored now. I am fine tuning animation, will post update soon. ::::: Opera :::::

"However, right now I am trying on the idea you propose in your last sentence. Dial up the realism best possible then "let go", don't fiddle, concntrate on animation, won't matter how "hyper" real the render is. Thanks." FINALLY! LOL. Just remember, John... your audience will know it's CG. You're not trying to fool them into thinking it's live action footage they're seeing, so their expectations will have already been conditioned not to expect "reality", and their perception of noticing things that aren't "spot-on" won't be as high. In other words, if they know beforehand they are watching CG, they won't be too critical of things that are a little off, but they will appreciate something that's well done! What you are doing here is much different than trying to incorporate CG into live action. There's NO live action to compare it against! See my point? Also, I'd strongly suggest posting any animations you have to the Animation forum on this site, rather than the Poser forum. I've been getting some very valuable feedback over there, outside the realm of "that's excellent" or "great job". Some really good critical viewpoints are being exchanged lately. That goes for you too, Wolf!! ;-D

Attached Link: http://www.renderosity.com/messages.ez?ForumID=12351&Form.ShowMessage=2498484

Anyone interested in commenting, please continue there. Thanks.

::::: Opera ::::

Message edited on: 12/08/2005 20:33

Half of that "believabe animation" you see is mocap. As for the rest of my response, I will do it in the Animation Forum.

when I say believable I'm thinking of Pixar,as far as I know they don't use much mocap it's the subtle movements that make an animation believable to me..get yourself a mirror & act out your movements I suck as an animator so shall bow out here good luck Opera :)

I am thinking ILM, Weta and SP Imageworks ;)

I saw the Polar express this week,imageworks obviously extensively used mocap on that,& it shows,the facial animation is horrible..saved only by a good story.

"I saw the Polar express this week,imageworks obviously extensively used mocap on that,& it shows,the facial animation is horrible..saved only by a good story" The irony there is that they actually used an advanced facial mocap rig to capture Tom Hank's facial movement and all the subtle expressions he did as he spoke his parts. Each actor was suited with dozens of small spherical targets on a leotard and dozens more on his or her face. I'd venture to guess that Tom wasn't very comfortable with all that stuff attached to his face, and probably gave a poor performance because of it.

I was reading something about that in a Cinefex a while ago. Basically you have to remember that you are not capturing underlying movement, but rather surface movement. If you capture Tom's performance that movement will only look ok on Tom's face because his performance is based on his skeletal and muscular anatomy. That's why nothing can beat some good keyframes.

actually..Toms characters would have been the best out of the lot,I loved the ghost character riding on top of the train..in fact all the train-top scenes were a treat the kids though just looked so wooden. mocap for body movement is great & surely indispensible but I'm thinking facial animation still needs that human touch of keyframing

"That's why nothing can beat some good keyframes." "mocap for body movement is great & surely indispensible but I'm thinking facial animation still needs that human touch of keyframing." I agree 100% with both of these statements. Hand keyframing is still the best way to go about facial animation, and even with motion-captured body animation, there's almost always the need for "hands on" keyframe input.

I've been watching the situation at SoftImage with their product "FaceRobot." I found out yesterday it is indeed coming to market first quarter 2006. However, XSI decided not to give it a "prosumer" price, instead perhaps over $20,000. For that price I was told it was "intended to replace full-time facial animators." Supposedly it animates the deep tissues in the face, not just the "surface"

Hmm....

I am sticking with hand keyframes for facial.

::::: Opera :::::

Message edited on: 12/09/2005 02:34

Not that I want to put down FaceRobot, but keep in mind that all these "new cool tools for 3D apps" could not-so-very-easily be done in any other app, even the older ones. You could make your own FaceRobot-like in an older softimage if you just spent that countless amount of hours setting up a facial rig. For example. Maya 7 allows you to Paint the morphs and animate that painted maps to make "dynamic morphs" but you can even do that in version 3 if you have enough MEL knowledge... My point is that you don't need the "top of the line" to get the job done ;)