Hi,
I notice a lot of japanese games like this. Notice the 3D art that look like anime, but does not seem to have any toon shader. What do you think? Does any one know how this is achieved?
Any help is greatly appreciated.
Hi,
I notice a lot of japanese games like this. Notice the 3D art that look like anime, but does not seem to have any toon shader. What do you think? Does any one know how this is achieved?
Any help is greatly appreciated.
Most effects are a combination of custom shaders and artwork made specifically for the style.
P3P is only a diffuse shader.
You donât have to do anything to get it to look like that. Just use a unlit diffuse shader and paint the model to look how you want it to look in game.
Thank you! I guess the textures has a great role in this. I am a programmer and not an artist, but want to create models like those in Daz3D for example or another 3D software that makes it easy for non-artists to create 3D models.
Sounds interesting. Then I guess I have to do the textures in a certain way.
I have this art, which does not satisfy what I want to do.
I guess I have to edit the textures, and switch to unlit shaders?
An unlit diffuse shader means you would have to draw the shadows onto the models wouldnât it?
Yes. Here is an example.
2b5712d2bdf644678e5de79fc25be0d5
Yes. That will never look like anime because fundamentally itâs just some realistic looking dude with colors to look realistic and not stylized. Not only that but I donât think normal mapping really helps for the technique you are trying to go for.
For an example of how to do this ârightâ I would point to the latest entry in the Guilty Gear series. (Xrd Sign) In this game, theyâve gone to great lengths to emulate their 2D hand-drawn style using 3D models.
However, it is worth pointing out that Guilty Gear has a distinct advantage. Their game is designed to be seen from a single perspective. This gives them far more control over how the models are viewed. They can tweak the shading and the lighting given a largely fixed perspective. Having this level of control over the lighting is a huge advantage for attempting to achieve this effect.
Painting the shadows onto your model is one approach. But it is also one of the least flexible approaches, and wonât take advantage of dynamic lighting. Your best bet is to write some custom shaders. This is a much more technically demanding solution, but it will offer you the best results. The objective should be to translate shading into ramped tones of the underlying UV map colors, as opposed to hard shadows.
@dmitsuki Thank you for the valuable information!
Persona 3 does not have any sort of toon shader. Highly detailed artwork (including intro sequence) you see is hand-drawn.
Persona itself uses standard vertex lighting and pretty much it.
This will never look anime, no matter what you do with the textures, because facial proportions and character style are wrong and do not match anime style. Anime characters have less facial details, different eyes, faces, etc.
For example:
For references see guilty-gear games, Valkyria Chronicles, etc
The model you posted wonât even work in ârealisticâ anime style (see ghost in the shell, and patlabor).
If you want your game to look like anime, just cut as many corners as possible, make the story convuluted, but lack any actual substance, usually by having the protagonist trying to achieve a vague goal that is likely unreachable, and make the english voice acting sound as awkward and censored as possible.
Job done.
You canât have anime, but you can have a comic style look.
Maybe you can take some cue from this
There is a bit of automation to turn a real model into a comic style character that you might leverage to approximate good result
I want to elaborate on that, itâs one hour video, the relevant part start at ~20:00
Their creation steps:
automated step â with normal map
Ambient occlusion is generally baked from hi poly, with the absence of hi poly you can baked the hi poly, then find the normal map ambient occlusion and blend the two.
Shader can be ndotl masking a pattern projected base on screen UV, then multiplying the result with the diffuse. I advise having a step based fresnel (ndotv) â if fresnel > threshold then show diffuse else show white. You can also use a ramp to control both the fresnel and the diffuse (basically a texture with a height of 1 and as long a width as you want for storing the gradient, you can store the diffuse ramp and the fresnel ramp on a 2 pixel height, the use the height 0 for applying the nodtl and the height 1 to control the ndotv). Ramp are basically using the dot result as an indexing, You need to scale the ndotl though to get the full range, ie it normally range from -1 to 1, generally clamp between 0 and 1, we need to offset first to the range 0 to 2, then bringing back into the 0 -1 range by multiplying by 0.5. So it will index the UV width (which is in the 0-1 range) with its result, 0 meaning away from light and 1 meaning directly in front, so the texture beginning is away (generally shadow) and the end is front (lighted part), the middle being the transition.
Otherwise the unity toon shader is enough, it give you the outline and you can draw the ramp to get the desired result. If you have a smooth ramp you get smooth shadow, if you make step ramped, it give you stepped shadow, and so on. Control the look with the ramp.
Oh, you CAN have anime look. It is just very difficult to achieve.
Those are 3d models. Look 99% identical to handdrawn anime.
http://www.gdcvault.com/play/1022031/GuiltyGearXrd-s-Art-Style-The
I meant that he canât have anime with his current model and if he donât want to model (he could make a proxy mesh to project the normal and have the smooth lighting of guilty gear for example).
The basic behind Guilty gear is not very hard either, just clever!
The Guilty Gear approach relies on the 3D model itself for most of the detail. Iâm guessing that the Guilty Gear approach doesnât use much, if any, normal mapping. They simply throw enough polygons at their character models to provide the necessary geometric detail. The rest is clever use of shaders.
As I pointed out earlier, this works for Guilty Gear because of the nature of the game they are creating. Only having two characters on screen at one time makes very detailed character models more feasible. For a game with more 3D objects on screen at once, this would not be feasible.
The real issue you should be addressing is exactly what kind of art style you are going for. Iâd sketch up some concepts of what exactly youâre trying to achieve, and then balance out your design based on that.
Guilty gear do use normal, in fact they transform them using a proxy mesh to control lighting
More visual break down here:
http://polycount.com/discussion/121144/convincing-3d-that-looks-like-2d-wow/p1
http://www.4gamer.net/games/216/G021678/20140703095/
http://www.4gamer.net/games/216/G021678/20140714079/
Their gdc vault talk
more cel shading
https://simonschreibt.de/gat/zelda-wind-waker-hyrule-travel-guide/#history
similar rendering
@neoshaman Thank you very much!
I really appreciate the time you took to explain to me that approach.
Iâve spent a lot on art and cannot afford making it again, and it just does not look right.
Itâs not very good as realistic art and does not look cartoonish.
Iâll apply your approach to the model and the environment and post the result here.
It wonât make miracle but it might be better than just regular lighting ⌠I hope
Nice model and texture - minus the awkwardly backwards modeled elbow joints.
But textures are top shelf.
Is it just me - the sad face smiley faces look better than the final shaded faces on the right.