An anime 30 minutes chapter cost 150.000 USD to do.
I don’t think a FG needs more 2D sprites than that.
I know pixel art is pointless in 1080p and new ones should go the Skullgirls way.
http://www.crunchyroll.com/anime-news/2011/10/30-1/how-much-does-one-episode-of-anime-cost-to-make
Original work - 50,000 yen ($660)
Script - 200,000 yen ($2,640)
Episode Direction - 500,000 yen ($6,600)
Production - 2 million yen ($26,402)
Key Animation Supervision - 250,000 yen ($3,300)
Key Animation - 1.5 million yen ($19,801)
In-betweening - 1.1 million yen ($14,521)
Finishing - 1.2 million yen ($15,841)
Art (backgrounds) - 1.2 million yen ($15,841)
Photography - 700,000 yen ($9,240)
Sound - 1.2 million yen ($15,841)
Materials - 400,000 yen ($5,280)
Editing - 200,000 yen ($2,640)
Printing - 500,000 yen ($6,600)
You forget how crazy little RAM current consoles have (they are like what, 7 years old?). HD 2D takes a lot of space, more than 3D games. Skullgirls had to stream the frames from the harddrive, resulting in some graphical bugs on certain models, mostly on 360 I believe (the famous “my character looked like hitboxes for a while” bug).
To be fair, SG had to do that because it was a team game with tags/DHCs/etc. When a match is just 2 characters, I don’t think they stream the assets during gameplay.
And in general, this should be easily feasible next-gen since both Sony and Microsoft’s offerings both appear to have at least 2GB of RAM on board; even discounting whatever amount is going to be reserved for other applications, it’ll be several times what’s available to either console now.
I’m not too familiar with the world of 2D art and Japanese animation, but it seems like TV animation and sprite animation are wildly different worlds. Anime studios are using different techniques from video game studios to create their works. For one, anime studios usually air each frame three times to save money.
You need to take into account the number of man hours to make a frame of animation in a 2D fighting game. It’s WAY higher than a television program, because the expectation for consistency is much greater.
with a 3d model, your initial set is expensive in terms of console resources (RAM) but everything that model will do is extremely cheap (all their animations are just instructions)
with a sprite, each frame is it’s own image to load. so an idle animation is like loading lets say 30 images of the same thing, although they look very similar it doesn’t matter. then you have to do that for each other animation. it adds up very quickly.
Besides the fact even a lot of run of the mill PCs aren’t optimal for HD2D, and Skullgirls had problems streaming itself…
There’s too many variables like the studio’s transition to video game animation, the budget its having, and what the direction of the game is.
Skullgirls honestly looked real off at times. You had an entire neighborhood at the fight at times but just mildly spectating the fight. There’s stuff like them off perspective, some girl only two feet away being watercolored while the girls in front of her have bold cel shading. That Disney-esque mom with a live action rabbit. The city roadway at night where there’s a bunch of OCs that just twitch over and over despite getting the same detail as the main cast, which makes the depth of field really weird.
To be fair this isn’t 100% true you recycle animations within moves like Hurricane Kick is only like 5-6 unique frames. Idle animations are almost ALWAYS the most expensive animation because it’s so slow and tends to have a lot of unique frames but even then you normally have the character return to his original animation so your looking at a MAX of 30 or so but actual number is probably around 15-25. I mean a fireball even though it’s 40 something frames is only 8-10 frames of actual animation in 3S.
Most moves don’t have all unique frames for any moves. It has a lot to do with once the move is out you want it to look like that move long enough people know what’s going on. That’s another reason hit stop is used so people know what they blocked or got hit with. Skullgirls has a lot of animation and never seen their spritesheets o
Source: I wrote a 2D fighting game engine and tested it by animating 3S ryu sprites.
TV sprites use a LOT of cheater techniques to reuse assets. They have a computer program that does all the voice matching for them. Most rarely show legs which would have to be redrawn simply if they took a step onto some stairs. Instead they draw the character from the waist up and simply move him up and down like a paper puppet to simulate he is walking. They tend to have a lot of still moments where nothing is going on. Action scenes are more expensive though for sure.
A move with 12 unique frames for a game will probably cost you around $150 to make. Now you need 400 frames or so for a character so your looking at probably $3000-4500 per character. You can get a Full HD 3D model, high-res game model, and have it textured and rigged for $500-1000 depending on what your making. Animation won’t add more then a grand or so if you have a mo-cap setup and most of your movements are fairly realistic. If your move is highly unrealistic it will need to be key framed and it will be more expensive to do the move then mo-cap but less then the 2D sprite move. So if you already have a Mo-cap setup animation is PISS easy so 3D models are cheaper cost and resource wise. That and the fact people think 3D looks better makes it a no-contest for big companies on which to choose. The inital entry to 3D is very expensive and most people don’t have experience with it so that’s the reason Indy games go with 2D but they typically don’t pay their artists much because they are doing it for profit shares and maybe a small upfront cost for each spritesheet unless you have a real publisher or take on some upfront debt since making a high quality game is an expensive effort that gets more expensive the more help you need (netcode, marketing, blah blah blah).
It’s not the modern way of doing them, it’s just another way of doing them. In ArcSys case it was a waste of time because BlazBlue doesn’t look much better than GG in terms of consistency/accuracy (which is what tracing from 3D really helps with).
comparing 2d animation in video games and anime is not recommended in this case.
Cheap Japanese televised animation has already an established tradition. But it started with Hanna Barbera actually.
While Japanese AAA video games moved already to 3D a long time ago. It also has to do with the fact that talented Japanese animators prefer the video games over the animation industry due to better pay and working conditions.
besides, Blazblue and GG are more famous for their style rather than rich animation.
Hm, that actually sounds pretty good. Would reduce the time and QC commitment for sure. Any idea of the detail involved? Seems like a HD model is overkill since fine detailed doesn’t appear to normally translate into the drawings so by HD I assume HD just simply means high polygon. Interesting thing is you could map a normal map to the sprites and do dynamic lighting in 2D … which is pretty cool.
Basically, the 3D models are SUPER barebones basic models that they just use for basic shape and form. They add shading and details when converting to 2D to get that super detailed look.
Apples and oranges. SG’s method (some pros/cons)
[LIST]
[]sprites, but not true pixel art
[]costs several times less per-character to create than any current 3d or 2d FG by a major publisher
[]uses base sprites that are twice the resolution of native for finer/softer edges
[]^ limits the max level of detail that can be conveyed at native res(though nobody else has gotten close to making this look like a weakness and probably won’t unless SNKP starts making 720p sprites at the same level of quality as their upscaled 480p)
[/LIST]
SG also use 3D model bases for animating specific portions i.e. Painwheel’s blade, so it isn’t even like the methods are completely exclusive to each other. I would bet money that if ASW knew the tech could be applied this way 10 years ago, they would have gladly adopted it over what they’re currently doing because it boosts the two weakest things about their sprite offerings; the detail(for something at 720p native, they are ass compared to something like KOF which does it right) and the feasibility of budgeting more inbetween frames per animation(since SG’s process is cheaper).
It’s an interesting idea, but you have to realize it’s inefficient and costly to have an outside company do it.
Yes, an animation studio has a team of trained animators, but in terms of logistics it’d be annoying. If moves need to be tweaked, new characters are to be added, or animation assets have to be changed in any way, you have to go back to the company and have them redo it. This not only takes more time, but is expensive, and there’s no telling how much back and forth you have to do before everything is done the way you want.
The other problem is that the animation studio already has tons of its OWN projects to do. In order for Capcom or SNK to convince them to drop what they’re doing and make game art, they’d have to pay those studios a ton of money. Probably more money than a fighting game company wants to pay.
Having an in-house team is of course the better option because everything is faster, streamlined, and internally done. But these days Capcom and SNK don’t really have 2D staff, so that’s why they do 3D now.
Could Pixar handle cutscenes for Uncharted? Sure they could. But is Naughty Dog going to pay mega bucks to Pixar for them to do it? Of course not - they’d do it themselves. Having an animation studio do game sprites is the same thing.
The fuck are you talking about, there are inconsistencies on some animations of GG because none of their sprites are done 3d -> 2d
Just look at Testament’s Badlans or Order Sol’s 6HS, or Chipp 2P
BB has better accuracy and consistency across the board because it use the 3d->2d the same way that KOFXIII does
