2013-11-25, 02:58 | Link #22 | |
Translator, Producer
Join Date: Nov 2003
Location: Tokyo, Japan
Age: 44
|
Quote:
You can't just take a studio which animates in the traditional way (even one that's all digital) and say "we're doing everything in vectors now!". That's almost as crazy as converting to all 3D CG, it's a different process entirely.
__________________
|
|
2013-11-25, 03:38 | Link #23 | |
Senior Member
Join Date: Jan 2006
|
Quote:
Drawn animation already IS basically vector, being composed of individual images and changes to them. It's just that instead of photographing the result (thus losing a lot of structural information) and letting compression algorithms (optimized for a completely different thing) determine the basic elements and their changes, you give specialized algorithms the data that has already been broken down into the best possible elements, allowing for much more efficient compression. I.e. if you are reusing background from frame 35 in frame 78923, all normal algorithms will have to store them at least twice (or more) - but optimized will only need that one copy. |
|
2013-11-25, 05:30 | Link #24 |
Banned
Join Date: Dec 2006
Age: 38
|
If you ever worked with vector graphics you'd know the result of converting to shapes can be really complex, so you have to smooth it out, which introduces loss early in process chain. It might be acceptable for lineart (then again in some cases not), but backgrounds?
|
2013-11-25, 22:36 | Link #25 | |
Translator, Producer
Join Date: Nov 2003
Location: Tokyo, Japan
Age: 44
|
Quote:
Drawn animation isn't vector because it's NOT composed of individual images and changes... It's _just_ individual images. In other words you don't create the "movement" data... This isn't 100% true of course, things like pans, zooms, walking movement, parallax are definitely done using motion data in editing software, but a large portion is good old fashion drawings. What you seem to be suggesting is taking like, the final composition software project and somehow converting that into Flash.... That's not possible with current software and even if it was it wouldn't increase resolution because the underlying data elements are already rasterized by that point.
__________________
|
|
2013-11-26, 20:54 | Link #26 |
Senior Member
Join Date: Jan 2006
|
You are seriously underestimating the amount of material reuse in modern anime.
I.e. take a look at latest Tokyo Ravens episode (8). Almost everything is zoom, pan, parallax, reuse at a different zoom level, etc. Outside that, often only the mouths and eyes are moving (and I bet those are heavily reused, too). A lot of that reuse is completely impossible for generic algorithms to store in an efficient way. Even the hand-drawn animations are mostly the same with little changes, so a specialized algorithm will be able to handle compression much better than a generic one. The fact that data is rasterized is much less important than all that (especially with the amount of zooms used - the resolution of scans is already much higher than the final result). |
2013-11-27, 03:45 | Link #27 | |
Translator, Producer
Join Date: Nov 2003
Location: Tokyo, Japan
Age: 44
|
Quote:
Vector animation is hard to render in real time on hardware beyond a certain complexity. The memory and buffer requirements are extremely complex for a transmission protocol, because like you said there might be a background in frame 6 that's re-used in frame 32000... so do you keep that in your TV/dvd player's memory the whole time? What about shapes and gradients and shaders and all the stuff that makes good graphics? Think about it too much and your realize that a "vector" format is going to be very limited to be hardware implementable, or else require essentially a computer graphics card and DirectDraw-like API in your display device. A brand new decoding and encoding (or well, exporting, I suppose) software ecosystem would need to be created, and it would be COMPLETELY INAPPROPRIATE for anything that wasn't 2D animation. Try compressing your local news or a sports broadcast with vectors... In theory, yes, it's very much possible to compress your usual anime FAR more than HEVC or h.264... and to make that compression nearly resolution independent. But like any compression scheme if it's only efficient for a small subset of use cases it becomes useless overall.
__________________
|
|
2013-11-27, 06:13 | Link #28 |
Senior Member
Join Date: Jan 2006
|
Good points - such a format might not be suited for streaming broadcast, or if it does, you might need to retransfer some data more than once (i.e. if you turned it on mid-way) , eliminating lots of the compression gains. OTOH, how suited is 4k for broadcast is a good question, too - it requires up to 16x the bandwidth.
Playback for stand-alone player is probably not going to be a problem, though - you need to replace it all for 4k anyways, and hardware cost goes down rapidly, so you'll be able to have all the memory and cache you need. Plus I have a feeling that most fansubs are downloaded on PCs, where it is all available already. The problem is the source, as I doubt companies are going to readily release their raw vector data for fansubbing community only If they do adopt vector anime format for 4k, it is going to be for Steam/Crunchyroll/XBox One/PS4/BD's, etc. |
2013-11-27, 14:03 | Link #29 | ||
Senior Member
Join Date: Dec 2008
|
Quote:
Though some things have to be taken into account: - bitrate does not grow linearly with resolution - people have trouble seeing the difference between 720p and 1080p already. And since bitrate can be chosen almost arbitrarily you can keep it fairly low while still marketing 4K/8K. - new compression schemes (i.e. HEVC discussed earlier in this thread) increase compression Quote:
Any BD 4K/8K successor will have general purpose compression, i.e. most likely HEVC. |
||
2013-12-03, 03:01 | Link #30 |
Senior Member
Author
Join Date: Oct 2007
Location: Philippines
Age: 47
|
Hmmmm... it'll take a huge leap for some studios to embrace the huge formats, yet all they'll do (and for the sake of economics and time management) is NOT to change much of the traditional process (including the inclusion of CG sequences), but to simply take every frame at 720p for regular broadcast, but use 1080p for movies and miniseries -- anything bigger could be overkill given the limited budgets they get from broadcasters and additional variable income from merchandising.
Only the most ambitious, risk-taking studio will go for 4k from start to finish -- I'll not be surprised if anime industry king Studio Ghibli decides to become the first to create a 4k production using traditional animation processes. I feel that the only genres that'll regularly use 4K are high-budget movies and major sports events. Also, recall that Japan has some of the cheapest 100mbps broadband services on earth, and this means they can use HD and possibly 4k streaming on a regular basis.
__________________
|
2013-12-03, 07:36 | Link #31 |
Senior Member
Join Date: Sep 2007
|
I think for anime companies it would be a setback. I was in the tv business until 2010, I saw what it was to jump from Standard resolution to HD and the time and cost increase that it carried (the whole rendering process doubled, to name what was most critical at the time). The whole process of 4k may create a serious burden in the production area of any company, and we know how budgets are reduced because something didn´t go as expected and so on.
We had to eventually replace all our rendering machines and workstation to keep the times schedules, was a mess. |
2013-12-04, 02:41 | Link #32 |
Senior Member
Author
Join Date: Oct 2007
Location: Philippines
Age: 47
|
The other day I was thinking about the eventual decline of traditional television, and even cable TV, as the viewership goes mobile through streaming video on smartphones. I'm no longer using a TV unless I want to watch some news, as much of my entertainment options are on the computer. Almost everyone else are doing almost everything on their smartphones and tablets these days, especially social networking, games and streaming video.
If that would be the case, who's going to use those 4k sets? The more likely target customer for these are the high-end enthusiasts and the early adapters who could risk a few more Cs to be the first to have the sets on their block. Everyone else wishing for a budget 4k may have to settle first for a decoder, hoping that the price per unit comes down. @dmaxzero: Of course, trying to move up to 4k would be costly, in terms of technology required, the equipment and the gargantuan amount of video storage for production.
__________________
|
2013-12-06, 13:37 | Link #33 |
Member
Join Date: Jul 2007
|
Yes it will definitely have an impact: More upscales from attention whores for leach-counts.
More discussions for Encoders with "Group-Leaders" who go: "Meh why you don't Encode in Super-HD. The other Group released in this resolution to ...." Good that i am to old for this shit. I go and do something constructive like crashing in BugField 4.
__________________
|
2013-12-11, 10:52 | Link #35 |
Translator, Producer
Join Date: Nov 2003
Location: Tokyo, Japan
Age: 44
|
So the other day I was inside Studio Pierrot and actually learned specifics about their animation and compositing process.
Simplified, this is how they animate Naruto, e.g.: 1. Draw backgrounds, character keyframes (mostly done by hand still. Some directors draw on tablets, though). 2. Scan both to targa files. 3. Send keyframes to in betweeners, who produce short animated targa files for each moving object. 4. These final files are colored using Paintman. 5. Background and object source targa files are assembled into cuts using Adobe After Effects. Camera work, digital effects, and the animation itself (i.e. timing of the underlying object frames) are done in Adobe after effects. 6. Each finalized cut is rendered to another targa video. 7. These cuts are imported and assembled in nonlinear editing software (Final Cut Pro, I suspect), and output to HDCAM masters. The raw underlying targa files are about 80 GB per episode. All in all it was a lot more straightforward than I thought. Just, you know, ya gotta draw a ton of pictures .
__________________
|
2013-12-12, 13:51 | Link #37 | |
Translator, Producer
Join Date: Nov 2003
Location: Tokyo, Japan
Age: 44
|
Quote:
A better question is what dpi things are scanned in. Since the character art is done on pretty standard size A4 for the most part, if we assume they scan at 300dpi, that's ~3500x2000 for a full page. And that animation of Pain is just the style of the animation director they used for that episode...
__________________
|
|
2013-12-12, 18:09 | Link #38 |
Junior Member
Join Date: Dec 2009
|
Well I know Matsumoto Norio has a peculiar style, but compared to the animation he was involved with in sword of the stranger
Anyway, it's highly interesting that you share your real-life experiences with the process of producing anime, and thanks for that Last edited by mastrboy; 2017-05-02 at 12:46. Reason: follow up question |
2014-01-04, 19:14 | Link #39 |
Senior Member
Join Date: Dec 2008
|
Looks like 10 Bit H.265 has a good chance:
http://www.hdtvtest.co.uk/news/lg-hevc-201401033549.htm |
2014-01-08, 18:49 | Link #40 | |
Former Triad Typesetter
Join Date: Dec 2003
Location: Washington, DC
Age: 40
|
Quote:
If you're using a 24" screen as a 4K TV, well, then I agree—that doesn't make much sense to me either.
__________________
|
|
|
|