CGI Anime

[Chiibi]

.............if all new anime were to become 3D CGI, I will cut the person responsible for such atrocity and then I will cut MYSELF.

There's a certain charm in 2-dimensional characters that feels so wrong when entering the third dimension.

Yes, yes,
YES.
THERE.
IS.

[Hiroi Sekai]

Originally Posted by creb

I must be missing something, because I can't help but feel that there is a mistaken belief that CGI has to be 3D.

As for my statement on there being no reason why CGI can't one day perfectly emulate "2D" animation, the key part was "one day". It just comes down to perfecting the math.

Okay, let's take down my extremities a little bit for this then. First, I'd like you to fully detail when you mean when you speak of CGI. If you're not speaking 3D, then they already do use CGI in anime for many things. They just don't make entire shows using it because it ruins the flow.

You can't put this down to mathematics. I'm wondering whether you mean to literally draw it from scratch on a computer through a tablet or something, because that's not exactly CGI. Trust me, I'm a current art major and there's a reason we learn fundamentals through drawing by hand and not by mouse/tablet. Tonality, pencil shadowing, contrasts, draw distance and many other things just need to be done through pencil and paper, as they have for years and years. If someone else saw as you did, we'd definitely toss out the papers and just work on the computers.

I'm not trying to sound obnoxious, I hope you understand. I'd just like to hear your reasoning as my years in university are being overturned at the moment.

[Triple_R]

Originally Posted by creb

I must be missing something, because I can't help but feel that there is a mistaken belief that CGI has to be 3D.

As for my statement on there being no reason why CGI can't one day perfectly emulate "2D" animation, the key part was "one day". It just comes down to perfecting the math.

I'm not sure why you're putting 2D in quotation marks.

Consider the difference between this...




and this...




Pixar movies tend to gravitate more towards the latter style.

Anime generally lacks that "computerized depth" feeling to it. Anime tends to look like drawings on paper, in motion. At least anime characters do.

That "drawings on paper, in motion" is generally what people mean by "2D".


I'm not saying one is inherently better than the other, as art is subjective. But they're definitely different. Somebody who likes that 2D style isn't necessarily going to like the "computerized depth" style (or vice versa).

[creb]

I'm no expert on CGI (or any form of animation), but articles such as the one on Wikipedia don't seem to preclude CGI being theoretically capable of emulating "hand drawn" art.

The common basis for CGI seems to be that it's algorithm based, and has nothing to do with whether it's "2D" or "3D".

I put it in quotes, because while 3D is the most prevalent form of CGI, there's absolutely no reason why algorithm based animation couldn't emulate that Simpson's picture. Why it isn't common right now, I don't know. Presumably the cost is not feasible, but one day, I'm sure it will be. A google search for why most CGI is 3d is very unhelpful.

[OceanBlue]

My biggest problem with 3D animation is that most of them don't really use distortion in movements. Bigger 3D production companies like Pixar will have their models squash and distort themselves to make more interesting movement, but for a lot of other companies that can't afford that or don't have the time due to a strict schedule, the models aren't really that interesting.

[Hiroi Sekai]

Originally Posted by creb

I'm no expert on CGI (or any form of animation), but articles such as the one on Wikipedia don't seem to preclude CGI being theoretically capable of emulating "hand drawn" art.

The common basis for CGI seems to be that it's algorithm based, and has nothing to do with whether it's "2D" or "3D".

I put it in quotes, because while 3D is the most prevalent form of CGI, there's absolutely no reason why algorithm based animation couldn't emulate that Simpson's picture. Why it isn't common right now, I don't know. Presumably the cost is not feasible, but one day, I'm sure it will be.

Well here's problem one, you're on Wikipedia.

That aside, I've already mentioned why it can never fully replicate hand drawings. Human hands aren't perfect, and even the most trained people won't be able to make 100% perfect lines with perfect colouring. You may not see it because they do a damned fine job, but if you zoomed about 10000 times into Okazaki Nagisa's hair, you'd start to see where lines diverge. Now punch every one of these little movements that happened every few pixels into an algorithm so they match. It can't be done.

I'll give you a little example, because it's important. No credit to myself, these two following images are created by a couple of favourite DA users of mine:

Spoiler for Hand-drawn:

Spoiler for Computer-drawn:

In the hand-drawn image, I used a couple of arrows to point down at key points. Notice how lines aren't complete, and colouring has seeped past the lines a bit? Zoomed out, it's not noticeable, but it's an aesthetic charm and a way to speed up the process as well.

Now have a look at the computer image. The lines are perfectly clean because the algorithms work their magic. As a result however, the tone itself is lighter and the lines have less of an impact. Now both are fine, but in movement this would be much different. It's not a matter of time, it's just something a computer wouldn't be able to accurately recreate to a T.

[creb]

I see no reason why imperfections can't be randomly generated algorithmically.

I stick by my belief that one day CGI will be able to emulate hand drawn art, and double blind tests will show it.

[Triple_R]

Originally Posted by creb

I see no reason why imperfections can't be randomly generated algorithmically.

I don't think it's that easy. The reason why these "imperfections" work with hand drawn art is because good artists have a keen sense of where lines should be faded a bit and where lines should be made heavier (to have more impact). These "imperfections" aren't always random for human artists, as they can arise at specific points.

Specific intuitive points where an artist senses that more or less line depth would help.

I have real doubts that you can program a computer to be intuitive in that way.


This, by the way, relates to what some people mean by synthesized music being "too perfect". A musical note going a bit higher than normal, a note carrying a deeper sense of emotion due to the voice cracking slightly, the pitch being not perfect but very passionately delivered - That feels human, and carries a certain "Oomph!", that I can feel absolutely none of with a Hatsune Miku song.

[Hiroi Sekai]

Originally Posted by creb

I see no reason why imperfections can't be randomly generated algorithmically.

I stick by my belief that one day CGI will be able to emulate hand drawn art, and double blind tests will show it.

The reason I'm persistent here is because you haven't explained yourself. It's just "it can happen because it's mathematically possible". I've worked in the actual art myself and have seen/felt the differences.

You could do a ton of theoretical things if you think "mathematically" and logically. Art is not a science however, it's expressive. The human eye can detect falsehoods very easily, and no matter how real you make something look, you will always spot something that makes it look false.

Answer one thing for me if you've seen Clannad. The alternate world scenes with the little girl and the robot- is it CG, or is it hand-drawn?

[Nightbat®]

Originally Posted by papermario13689

The reason I'm persistent here is because you haven't explained yourself. It's just "it can happen because it's mathematically possible". I've worked in the actual art myself and have seen/felt the differences.

I work with guitaramps
here there's the old "Tube vs Solidstate" -now with an added "/Modeling" debate

think it's impossible to recreate the nonlinear/chaotic distortion tube amps provide?
20 years ago, I'd say 'yeah', a decade later things sounded like badly done solid state

...but the past decade things have improved at a considerable rate, where there are already enough examples where the ear won't be able to win a blind () test

It may be a while, but at a certain point a CG rendering program, like with DAW sequencers, will have an option called "humanize" or "swing"
and I'm convinced that -as long as you manage to cram in enough 'random number generators',
there will be more and more needed imperfections inserted in the object to make it give a more human/chaotic feel

You can argue sound isn't visual, but then I'd argue that computers always were more advanced in the visual than the soundprocessing department
(why else are we still on 30 year old PoS Soundblaster 16 chips?)

[Hiroi Sekai]

Originally Posted by Nightbat®

I work with guitaramps
here there's the old "Tube vs Solidstate" -now with an added "/Modeling" debate

think it's impossible to recreate the nonlinear/chaotic distortion tube amps provide?
20 years ago, I'd say 'yeah', a decade later things sounded like badly done solid state

...but the past decade things have improved at a considerable rate, where there are already enough examples where the ear won't be able to win a blind () test

It may be a while, but at a certain point a CG rendering program, like with DAW sequencers, will have an option called "humanize" or "swing"
and I'm convinced that -as long as you manage to cram in enough 'random number generators',
there will be more and more needed imperfections inserted in the object to make it give a more human/chaotic feel

I've experimented with several amps, guitars and DAWs myself to understand your point. However, are we solving the problem or turning a blind eye? Technically, it's just hiding the fact it's emulated, and with enough use under both or enough knowledge, you'll always be able to separate them. It's not exactly perfect emulation.

Examples in both departments again.

5:34 onwards, I consider this to be one of the greatest tones ever for a guitar. I'm not even that big on Pink Floyd but this sound is just hypnotizing and it's something I've never ever been able to recreate, no matter how much I change amps and alter sounds through DAWs. It's also individually specific, which is why art forms are so unique.

Also, the reason I mentioned Clannad was that it's the perfect visual representative of what I mean. The regular portions are clearly hand-drawn and are quite nice looking for what they are too. The alternative world is CG- conspicuous CG, to be exact. You can tell when they've swapped because something feels different and off to you as you feel the transition. It's a great example because it happens as you sit and experience it.

Unless you were trained in both sounds for several years and could split each wave of sound down to its purest forms, and then play both at the same time, I don't think it's a fair assumption to say it's the same. Sure, it's close in emulation, but it just won't be the exact same, and there's a charm to that.

[Vexx]

I still use tube amps but there are some things digital amps are very cool for.

I PREFER hand drawn key frames on anime. If they fade away, I'd just cherish the ones I have. But I suspect the Japanese culture of cherishing old style art forms will prevent that from completely happening.

[totoum]

Originally Posted by papermario13689

The reason I'm persistent here is because you haven't explained yourself. It's just "it can happen because it's mathematically possible". I've worked in the actual art myself and have seen/felt the differences.

I'm sure working in it makes you better at seeing,but honestly in shots like this and this I honesly thought "wow,I wouldn't suspect the character to be CG unless I already knew it",of course that was far from always being the case but it did happen from time to time.

That being said,yes to me one of the charms of anime is that it's hand drawn and while I do think some animes can have great CG integration ( almost everything that's not the characters in Wolf children is CG) I will hope that characters stay hand drawn.

However I have enjoyed some CG movies and look forward to others,a couple examples:
-I enjoyed the movie After School Midnighters,it was kind of like a dreamwork movie on acid
-Looking forward to the movie 009 re:cyborg

[Jan-Poo]

Originally Posted by papermario13689

The reason I'm persistent here is because you haven't explained yourself. It's just "it can happen because it's mathematically possible". I've worked in the actual art myself and have seen/felt the differences.

You could do a ton of theoretical things if you think "mathematically" and logically. Art is not a science however, it's expressive. The human eye can detect falsehoods very easily, and no matter how real you make something look, you will always spot something that makes it look false.

Answer one thing for me if you've seen Clannad. The alternate world scenes with the little girl and the robot- is it CG, or is it hand-drawn?

I don't think you've been quite convincing yourself though.
Showing the difference between what you can do with CG with the current technology and hand drawn art doesn't prove it will never improve in the future.

You don't need to show that, I think we all know already that there are noticeable differences at the current state.

To prove that it will be impossible you should rather show me why it is mathematical impossibility for a computer to emulate that kind of human error. You may know art, but this is a question that only a mathematician or a computer expert can really answer.

At the end of the day the discussion is just the usual "A machine will never be able to emulate human behavior" VS "human like machines in the future will exist".

[Hiroi Sekai]

Originally Posted by Jan-Poo

I don't think you've been quite convincing yourself though.
Showing the difference between what you can do with CG with the current technology and hand drawn art doesn't prove it will never improve in the future.

You don't need to show that, I think we all know already that there are noticeable differences at the current state.

To prove that it will be impossible you should rather show me why it is mathematical impossibility for a computer to emulate that kind of human error. You may know art, but this is a question that only a mathematician or a computer expert can really answer.

At the end of the day the discussion is just the usual "A machine will never be able to emulate human behavior" VS "human like machines in the future will exist".

I suppose you're right. I guess I was trying to make the "traditional" VS "modern" argument, realistically speaking. You say it's a matter of a mathematician or a computer expert to solve, but I don't think even they could solve this. It's really down to getting down and doing it, and then it's easier to understand why animators to this day are still on pencil/paper rather than pure CGI, and will continue to do so.

There are just some things a computer cannot emulate in this matter, and if it ever did, it would have so many advancements that it would be crazy. Picture for example, the weight a pencil falls on the paper. As you draw, you cannot hold a perfect weight against the pencil as you draw a circle. Somewhere along the way, it will darken, or it'll grow lighter in any way. This is why artists draw several light lines as a sketching process. Now picture an algorithm on the computer that gives a certain weight to the pencil. What algorithm can you come up with that reacts as your hand does against the digital pen? Do you use a constant variable that adds +1 weight to it at certain intervals? What if the drawing differs? What if you wanted to do a drawing that has no limits, like a Van Gogh painting? See, computers have variables (or else it wouldn't be a computer), and humans are past that. That's why the computers don't own us, and that's why so many jobs are still done by hand today instead of by machine (which some have done).

Bottom line, until the day that we humans are on the same level as computers, there will be differences. Once that day hits, we'll have a hell of lot more to worry about than simple artistic comparisons. Run, the computers are taking over.

P.S. I'd still like a hypothesis on how it will come about that computers can fully emulate it to a T. I've kinda just been rambling and would like to hear why something like that may happen.

[Jan-Poo]

I think emulating the behavior of painting tools is a lot more accessible, even though we still have a long way. In the end that's just a matter of physics.

I think it would be an interesting challenge to create a sort of tablet that react in a hyperrealistic manner to an artist hand. It should be able to discern even the slightest change in pressure while emulating the kind of lines that would be traced by the selected tool on paper or other surfaces, including the occasional errors (if so desired).

Normal tablet cannot do that and the surface should be changed so to give the right feedback (a flat glass screen wouldn't do).

But if such thing could be done, and I don't think we're that far off, it would be quite a revolution I think. While classic hand drawn art has its charm, computer graphic opens the path to a lot more possibilities. If you could merge the two things, it would be great.

[Hiroi Sekai]

Originally Posted by Jan-Poo

I think emulating the behavior of painting tools is a lot more accessible, even though we still have a long way. In the end that's just a matter of physics.

I think it would be an interesting challenge to create a sort of tablet that react in a hyperrealistic manner to an artist hand. It should be able to discern even the slightest change in pressure while emulating the kind of lines that would be traced by the selected tool on paper or other surfaces, including the occasional errors (if so desired).

Normal tablet cannot do that and the surface should be changed so to give the right feedback (a flat glass screen wouldn't do).

But if such thing could be done, and I don't think we're that far off, it would be quite a revolution I think. While classic hand drawn art has its charm, computer graphic opens the path to a lot more possibilities. If you could merge the two things, it would be great.

Do you think such a thing would replicate traditional art, overtake it, or combine with it to make something entirely new? It's an interesting idea for sure. Also, I know it's science and all, but can we seriously replicate exactly how paints mix with other surfaces? It's not just a matter of physics, this hyperrealistic tablet idea. The computer needs to interpret all of this as data, process it and correctly display it in real time. It also has to be more than a standard tablet, as for it to properly replicate the idea, it'll have to display on the tablet itself as well so you can actually take your hand off of it once you've started and start again where you need to on the exact pixel. Is it a serious possibility?

[RRW]

thought i always wondering if there is a software where it turn 3D movie then "compress" it to 2D movie.

that would be interesting to see since it easier to make movie in 3D then in order to make it 2D quality. you can just convert it

[Hiroi Sekai]

Originally Posted by RRW

thought i always wondering if there is a software where it turn 3D movie then "compress" it to 2D movie.

that would be interesting to see since it easier to make movie in 3D then in order to make it 2D quality. you can just convert it

I'm not exactly sure what you mean. Do you mean 3D character models compressed to 2D, or do you mean 2D characters moving about in a 3D world? The latter has been attempted in some series before, but the former is something different.

You'd have to pull a full dimension out from everything that you've ever done with it: keyframes, backgrounds, characters, etc., and then, why wouldn't you just make it 2D to begin with? Sorry if I'm misunderstanding.

[RRW]

Originally Posted by papermario13689

I'm not exactly sure what you mean. Do you mean 3D character models compressed to 2D, or do you mean 2D characters moving about in a 3D world? The latter has been attempted in some series before, but the former is something different.

You'd have to pull a full dimension out from everything that you've ever done with it: keyframes, backgrounds, characters, etc., and then, why wouldn't you just make it 2D to begin with? Sorry if I'm misunderstanding.

3D->2D, they idea is you make a movie in 3D first along with animation and the lighting/shadowing. then you take "snapshot" of each second of the movie. do some of editing to "2D-fied" it.

the advantage of this is that you can get high quality animation and lighting/shadowing easier and possibly consistent quality for whole movie.

then again i little experienced in animation so i dunno will this theory will work IRL