Hello Guys ,
I had this task few weeks ago to make next-gen weapon , I followed the normal next-gen pipeline ( That i know ) which includes :
HighPoly model creation > LowPoly Mode > Unwrapping > Normal Map and Ambien occlusion Baking > Diffuse and Specular maps painting in photoshop .
The final result was very satisfying ( according to the lead artist and art manager opinion ) ( sorry can't publish any images due to NDA ) .
But this was not the client opinion . They said that photoshop texturing is ( not a widely recognised Next Gen technique ). and instead of that , we should apply procedural shaders and maps to the HighPoly model ( Inside 3dsmax ) and bake it ( Both the diffuse and the specular map ) into the LowPoly model plus a minor adjustment in photoshop .
when i compare the model i did with the refrence model they send us ( based on procedural maps ) it's very obvious it's very blurry and looks past-gen to me .. while the one we did is looking just fine .
I was wondering if this is true and if procedural shaders and maps are the way to go to create next-gen art .
Replies
So if you think of "nextgen" especially for smaller studios it's somehow the way they say it is, not that it is better, i've seen this alot and i don't like it. Smaller sets with more love are definitely a thing i'd always prefer instead of giant world with thousands of generic assets.
EDIT: Here's something I read about in 3DWorld 99. http://www.mapzoneeditor.com/
It seemed pretty nice (nodes are awesome). Not sure how well it works out with more complex objects though.
You model and texture everything in high res, using procedural maps or photoshop maps tiles on all the peices. The reason you would do this is you can then rearrange everything into a new model if you wanted or make quick adjustments on detail.
Then just bake the normal, diffuse, spec, whatever else to the low poly shell.
I've only heard of Heavenly Sword using this method and personally don't think there is a lot of gain unless you have a lot of objects that look very similar.
I use them to start maps, concrete, wood and mostly metal. The advantage is that you can set a procedurals material to always follow a world axis and then bake that into your jumbled UV mess. Having the seams be (more or less) correct and the pattern facing the right way on the final model regardless of the way the UV pieces are laid out. But at best its normally a "decent start" not a quality final product.
Also depending on the app you use to do the baking, most will correctly color the padding making a final bake after painting and retouching a breeze to do padding.
One of the big disadvantages of procedural textures is that they tend to make the texture work throughout a game look very..."samey." Their use almost insures a certain lack of stylistic diversity. Also, I've found that procedural textures also tend to have an "artificial" feel to them. It is almost as if they look too clean. Moreover, procedurals are ideal for repeating textures, but next to useless for unique textures. I would never use procedural texturing for a non-repeating model. You might as well try texturing your model in Illustrator, you'd probably get a similar result.
I think procedural textures are good for environment modeling, especially for prototyping. (they are a great way to fill an environment with decent-looking textures quickly) But for the end product, you will always get the best result using painted textures.
It's just this sentence ( ( not a widely recognised Next Gen technique ) ) that worried me alot i suddenly thought that i have become an old fashion
Thanks again for the help .. all appreciated .
(click)
...but it takes a pretty deep tree of nodes to get results like this. Once you set it up the variations are a snap, and normalmaps, specmaps, etc. are generated automatically. Still, it's very hard to avoid that procedural look. Painting skills are still a very necessary part of texturing.
In my opinion, unless the project has a very defined pipeline that saves the production highpoly model for future editing and reworking, there is NO good reason to force an artist down a certain path of production.
-R
Now on a serious note, I'm sorry you have to deal with that kind of mentality. Procedurals work hand and hand with bitmaps (photoshop textures). They are not meant to replace bitmaps they are meant to work together. Also good materials with procedurals can take their sweet time to render, so this is probably not for games. Is this an art test? If it is you might be better off working for someone else if you can afford it. It would bother me to have to deal with that kind of nonsense.
On a side note they might be referring to combining normal maps so they look good up close and from far away, which is a slightly different than just using one 1024 x 1024 to texture the weapon. You have a smaller seamless texture that gets mixed in with a larger texture so it has sharp small details when you get close up.
Put down the crackpipe because that shit is never going to happen.
Procedurals can be a good base, just as you can quickly make a good base in Photoshop. They are an option for starting something. But not something that you can rely upon to get you 100% of the way there. Shit, in most cases they can't even get you 70% of the way there.