2D (sprite based/pixel art) or 3D for fighting games (art talk)


#1

If you had the choice for an fighting games look, would it be 2D or 3D? If someone would make a fighter with pixel art like in the old days, would it sell? Please discuss


#2

#1 hey dude. @dawidstrauss welcome to SRK TT

#2 no, see the beauty that is Yatagerasu: Attack on Cataclysm. It sold just barely enough to exist, but not enough thrive, regardless of its amazing gameplay and devs pedigree :frowning:


#3

You’re clearly new here, there’s a whole bunch of threads about independent fighting game projects that people here are working on. Some use 2D art, some use 3D art. I’m even making one that uses 3D. This isn’t exactly a new or interesting topic, as people here are actually making fighting games that involve this decision. And even if it was, I don’t think it belongs in Tech Talk.

Maybe you should try reading the forum? Particularly “Fighting Game Discussion”.


#4

Thanks guys, I’ll check out the forum more before posting further


#5

People use 3D now because its cheaper. The game engine and tools already exist so you don’t have to develop all the tools to build your game.
It is why Street Fighter IV and V and KOF 14 exist in the formats they do. Game is made in the unreal engine and 3D models lead to assisting in ways animation 2D can’t.
2D animation, the smoother animation you want the more frames you got to do. In old school NES days a sprite might have 2 or 3 frames for walking and the reverse direction is just that sprite flipped.
In hand drawn animation, you got 23-26 frames per second (Disney goes for 26). You got games like Street Fighter 3rd Strike that has hundreds of not thousands of frames, which takes up alot of memory (storage and ram) and alot of time to do. I think so far only Skull girls still does animation by hand, and they built the engine and tools to do exactly that. They are no way as smooth as traditional hand done animation just in the nuances how games work. Make you animation too well in effects game play (we started to see this in later SNES and Genesis games).

As for Pixle art? Its already a niche area with a niche market. Are you going for 8-bit or closer to 16/32 bit pixle art?
What makes a retro looking game successful is is playing off nostalgia. Looks at Shovel Knight, there is many tropes and parodies of other 8 bit games in Shovel Knight.
Shovel Knight makes indirect references to Super Mario Bros, Mega Man, Castlevania, Zelda II the Adventures of Link, Duck Tales (the game) and possibly a few others.

For a Sprite based pixel art Fighting game to be popular you have to look at the Golden age of fighters the 90s, look at all these SNK and Capcom Fighters (maybe a peak at early Mortal Kombat) as see whats so iconic with those games. I would also study some of the Beat em Ups of the era Golden Axe, Streets of Rage, Final Fight, Teenage Mountain Ninja Turtles the Arcade game, the Simpsons Fighting game, X-men Arcade, the Avengers Arcade the 2 D&D Beat em Ups if nothing else just for some variety and a study how characters are expressed in animation and appearance.


#6

@darksakul I have an extensive history with pixel art and animation, and it doesn’t hurt that I love the old schooly of it. I have a passion for the 3rd strikes and the kof 98’s. I just feel like I want to make something awesome, and what people like the most, especially in this community.


#7

To clarify this: there are 24 fps in film, so for high budget stuff, most animators work on ‘the twos’, so 12 drawings with each frame shown twice to get to 24. Not sure where you got Disney going for 26.

@dawidstrauss If your passion is in 2D sprite animation, then do that. Best advice: don’t draw at 1080p or greater. Keep your sprite size small as it will allow you to work faster.


#8

Didn’t get a single challenger during my entire lab session today…I’m not ready to let go of Yatagarasu!


#9

Honestly, been started porting my fighting game engine to unreal recently. So i’ll chime in.

3D is the standard and future. 2D has place in the pipeline and for a single passionate artist wanting to make a small fun project you can do low keyframe 2D (look at Vanguard Princess which was only 1 guy for the vast majority of its development), but when you get into larger projects with a time table 3D is cheaper and easier. Also look at Xrd where it is better looking than most 2D or 3D and still maintains that 2D feel, though that artist workflow is very difficult to emulate unless you are really skilled and requires more time (aka money) than standard 3D workflow. 2D still has a big place in the development enviroment for concept art, UI, potential character portraits, etc. It just is too costly to do sprite based games that aren’t animated in that cheap mobile game looking format.

Personally I’d go for stylized 3D over 2D if I wanted to make a project (currently considering trying again as my last project had to be put down because of a work promotion). However, you also have to consider stuff like what audience are you selling too and whatnot. I really like the stylized stylings but the fact is that realistic hi-def sells better and has a bigger market. I’m convinced if you have cool enough looking characters with enough charm your style really just has to look good, but that is a whole other discussion.

Also the high profit item for fighting games is alternate outfits which aren’t even possible in 2D. You’d be cutting off a major revenue stream by going 2D.


#10

@sephiroth73003 Thanks :slight_smile: but this is less about time and money and more about a passion project I’m probably going to do for free. I have about 9 years of industry experience in art for both AAA and indie, and I have learned that when you tackle a game for the first time or to experiment etc. Don’t try to sell it. I want to make something rad, hell even if it fails.


#11

It’s actually more expensive, however it’s easier and gets results faster. Depending on the project, give or take, sometimes it may cost more. But overall, the end result is that obtaining 3D is a lot easier, and not cheaper. Majority of the time, it’s actually more expensive.

Also, 3D is poop and will always be poop.
2D FOR LYFE!!!


#12

The fuck is this doing here…


#13

No problem man there are several 2D engines though none support netcode out of the box to my knowledge. My plan for this winter break is to update my 2D fighting engine to not use 3rd party libraries and finish / port my 3D fighting game engine to unreal. If you need any help on the technical side hit me up. 2D can be fun and if your aware of how to animate 2D then that is the best place to come from as that is the most difficult part of the project imo.


#14

Rad! That would be all kinds of awesome :slight_smile: I’m going to start making assets as soon as the studio I work for closes for the holidays.


#15

Just for reference, each of Skullgirls’ DLC characters cost about $175,000 per character and it took about 4 or so months for each character to be finished. Take note, that Lab Zero actually saved a bit by outsourcing the animation to hundreds of individual contractors, so they didn’t have to spend on staff salaries. Of course, this is also helped them deal with the fact that they were drawing each frame at 2160p and each frame had 3 separate layers outline, color and shading (since their Z Engine allows for real time sprite shading and lighting).


#16

@d3v yeah, for sure, I think as an exercise I’m going to do different versions of a single idle, and then share to see what takes better


#17

Skullgirls also has the highest number of unique frames per animation whose title was held by 3rd strike beforehand. You can design some movies like a Tatsu which may last 2 seconds (120 frames) but is in reality like 12 unique frames of animation. A single person doing what Skullgirl’s did is unlikely (impossible) but you can make ‘a fighting game’ with less work it just won’t be of the same quality. If you look at Vanguard Princess you’ll notice many animations are choppy and that is because there aren’t many unique frames of animation.

http://twvideo01.ubm-us.net/o1/vault/GDC2014/Presentations/Cartwright_Muriel_Animation_Bootcamp_Fluid.pdf

Here’s some insight someone from the Skullgirl’s art team wrote up about some of that stuff and how more frames isn’t even always better.

Here’s some more reading about animation and how it work in fighting games that is a bit more in depth.

Honestly 2160p vs 1080p isn’t straight up 4x more work for the artist. It all depends on the level of detail they want to go in with modern tools. The reason bigger sprites often are more expensive is an artist spends more time on details for each sprite. If you went for a flatter less shaded look your resolution wouldn’t matter as much. So art style is a bigger aspect of the amount of work imo then resolution.

Honestly, I think outline, color, and shading textures being seperated is a pretty standard fighting game workflow not just for dynamic sprite shading / lighting but it is the only way to really do things like color edit modes and the like. All that stuff is fluff though and not 100% necessary in a small project. Skullgirls went above and beyond and created a really high quality project.


#18

The difference is that these aren’t flattened in Skullgirls. So you only palletize the base colors and then you change the colors, the shading is done automatically as explained in the video below.
https://www.youtube.com/watch?v=5VkDXBsIXso
In other games, the colors that make up the shading and highlights have to be edited manually (since the shading isn’t real time). You can tell by how much simpler the palette editing is in Skullgirls, compared to the color editors we see in other games (the fact that Mike won’t give us access to that is a shame, IMO).

Also, I recall that the Z-Engine also uses something similar to bump mapping that they use for dynamic lighting (whereas something like KoFXII/XIII just lights/colors everything within a certain radius)


#19

Biggest thing we’ve learned from Cerebrawl so far: if you’re a solo artist working on a 2D fighter, you’re going to need a great deal of time. We initially thought it would take 4 months (of Eliot working 12 hours a day) to complete our first character, which ended up taking a little over 10 months (still 12 hours of work per day give or take a few days due to travel or whatever) and the character still isn’t totally complete - all of the extra bits are sketched out but she’s not fully polished. The first two months of the last year involved us just nailing down our art pipeline.

Also, I don’t want to get into the secret sauce of how our spriting process works, but we only have to produce 2 layers for each sprite and our engine tools/shaders handle the rest. I cannot stress enough how important good tools are going to be for your success.

Wish you all the best @dawidstrauss! Look forward to seeing updates.


#20

Yeah they are breaking it up like that largely for color editing as can be seen at the end of that vid you linked, I wouldn’t be surprised if they had 1 more map after outline, color, and shading just for the normals. You need the shading separated from the color or else the color editing is going to be bad / messy. If you do it like that you get infinitely many color palettes for 1 set of sprites.

https://www.youtube.com/watch?v=W0H4vza9knM

Here is a pretty basic example of what you are talking about with normals. Even with all those maps you can bake it all down to a single channel per map except the normal map to keep memory usage low.

They also did 3D models for some parts where I imagine they got some normals. Mike Z talked about how a lot of things like Eliza’s staff and Painwheel’s … wheel … were done in 3D at times to help the artists be consistent. You also have games like BB that do everything in low quality 3D then basically trace the renders for their 2D line work. That is actually what lead them to doing Xrd, they got so good at making the 3D models for BB they figured they could put a little more work into them and completely cut out the 2D.