Is it worth normal-mapping smaller props?
polycounter lvl 8
So, after doing some research on the topic, I'm wondering - is it worth giving individual props normal maps?
When I say props, in this case I'm referring to small objects, such as desk ornaments and items on shelves, that a player could realistically get pretty close to. I'm researching and planning a scene that is quite small, self-contained and structurally quite simple, but the inclusion of props should add some much-needed character and story to the scene, and I'm wondering about the best way to approach them.
I'm tending towards making smaller resolution normal maps for the props, given that it's quite a simple scene and so the inclusion of normal maps shouldn't be too much of a hit performance-wise, but I'd be interested to hear how you would approach, say, an indoor scene (think Fallout 3 or Far Cry 2 interior environments in terms of prop density) with quite a few objects in it.
Is normal mapping props commonplace (or feasible), given today's tech? Or is it a little indulgent? Is it even worth it?
Thanks in advance.
When I say props, in this case I'm referring to small objects, such as desk ornaments and items on shelves, that a player could realistically get pretty close to. I'm researching and planning a scene that is quite small, self-contained and structurally quite simple, but the inclusion of props should add some much-needed character and story to the scene, and I'm wondering about the best way to approach them.
I'm tending towards making smaller resolution normal maps for the props, given that it's quite a simple scene and so the inclusion of normal maps shouldn't be too much of a hit performance-wise, but I'd be interested to hear how you would approach, say, an indoor scene (think Fallout 3 or Far Cry 2 interior environments in terms of prop density) with quite a few objects in it.
Is normal mapping props commonplace (or feasible), given today's tech? Or is it a little indulgent? Is it even worth it?
Thanks in advance.
Replies
Then yes.
technical: will the player get close enough to notice?
timing wise: will gameplay allow the player to even have enough time to look at the prop properly?
psychological: what will keep the player's attention? will he even care or notice the prop consciously? (e.g. some stuff you only notice if it's missing, but not when it's there)
And I think the quality of the asset will always be higher if you're baking it down from highpoly source - it will be more consistent with the rest of the environment too, if the rest of the environment is primarily baked from highpoly.
The only cases where I'd tend to make exception are simple organic things which often need tiling textures, like boulders/rocks or simple trees.
For actual mechanical or hard-surface objects I'd always make a highpoly these days.
Thanks again.
performance wise it would be best if everything in the scenes used the same shader and better yet the same material on everything
so the answer from a technical stand point is YES normal map those fuckers !
and merge as many objects together as you can get on one atlas map ! (seriously as much as possible)
When you do a proper HP, you get all sorts of things for free, like a correct AO bake, a RGB color bake to help create masks, when you consider these things, its likely going to take you longer to do this stuff by hand in photoshop. I always chuckle when people say they save bolts and smaller details for adding in photoshop, when throwing some floaters on your mesh is literally the easiest and quickest possible thing to do in 3d, and you get proper ao, proper masks, detail projected correctly onto your uvs even if they are slightly distorted etc.
Like mop says, really only tiled texture stuff would i leave to texturing alone, and organic stuff at that, as you can do modeling work to speed up a lot of tiling-type textures as well.
One really important thing to consider is that if you're doing a simple props with low resolution textures, you can really whip out HP models quite quickly, as the resolution of the textures dictate that:
A. You wont need nearly as detailed of a HP, infact you do not want a detailed HP, as you'll just end up losing a lot of detail.
B. Scale is very important when dealing with smaller, simple secondary props, so you dont want an excessive amount of detailing, and especially not a lot of very fine detail, as it will end up being noisy and distracting.
C. Scale and detail dictate that you can get away with errors(pinching, slight inaccuracies) that you otherwise wouldn't if the object was going to be held up to close inspection with a high resolution texture, so you gain some speed here too, there are only so many pixels to define your normals, so little errors pretty much get "averaged" out.
I think people get this idea that normal maps = detail, which to me is a little silly, as the best possible thing normal maps can do is accurately represent the form and shading on your mesh, something you could never to do the same degree of quality without normals or with a converted bump map.