well i mainly wanted to use it for a rock wall like ive seen it used before, i know that it looks alot better than a normal map if it's used right i like the shadowing effect it gives to make it really look like parts of the wall stick out further than others and vice versa
@Xendance when you take it that far you would be better off performance wise to put more polys in your mesh or use tessellation and displacement.
since that is a lot of instructions, for just one material and it could get expensive fast.
just checked it and it takes 212 to 246 instructions to do that. so that is a lot of gpu cycles to eat up and in almost all cases it would be cheaper to use some more memory for some more faces on your model.
@Xendance when you take it that far you would be better off performance wise to put more polys in your mesh or use tessellation and displacement.
since that is a lot of instructions, for just one material and it could get expensive fast.
just checked it and it takes 212 to 246 instructions to do that. so that is a lot of gpu cycles to eat up and in almost all cases it would be cheaper to use some more memory for some more faces on your model.
Depends really, for terrain it can be useful
Though I agree that that particular implementation is rather heavy, you can achieve similar stuff with 100-150 instructions, it's just there might be stepping problems if the displacement is too large.
Nope, he's right, Relief mapping/P-Occlusion is too heavy and very useless in games, even for terrains, they contribute nothing much. Plus, with recent Post-Process FXAA, you're not getting any benefit out of it since they'll show off very jaggy, while polygons will not. Polygons are easier to work with and process much more easily in most engines, and smaller power required.
Let me put it this way, for every polygon you use, applying a specialized material to it adds the processing power required about thousand fold, so all you're really doing for one polygon is asking 1000 more 'power' for it with such a material, which isn't very beneficial.
Also, Bumpoffset is more then enough for terrain. Just disable your compression (Uncompressed) with importing your normal map and Offset map, and use a constant to control the depth of it, while applying a power node if you want harsher contrast. Simpler solutions sometimes are the best.
Nope, he's right, Relief mapping/P-Occlusion is too heavy and very useless in games, even for terrains, they contribute nothing much. Plus, with recent Post-Process FXAA, you're not getting any benefit out of it since they'll show off very jaggy, while polygons will not. Polygons are easier to work with and process much more easily in most engines, and smaller power required.
Let me put it this way, for every polygon you use, applying a specialized material to it adds the processing power required about thousand fold, so all you're really doing for one polygon is asking 1000 more 'power' for it with such a material, which isn't very beneficial.
Also, Bumpoffset is more then enough for terrain. Just disable your compression (Uncompressed) with importing your normal map and Offset map, and use a constant to control the depth of it, while applying a power node if you want harsher contrast. Simpler solutions sometimes are the best.
Yes well, I suppose it depends on the scene. If all you have is an open terrain, you can use a bit heavier materials there. Whatever performance allows.
And I wouldn't say that every polygon with specialized material adds to the processing power requirements. These are mostly pixel shaders, no?
Bump offsets tend to look pretty crumby on terrains due to the likelihood of visible texture/pixel swimming with varying view distances and heights. I'd stick to regular normal mapping and using meshes where more visible sillhouettes are required.
Let me put it this way, for every polygon you use, applying a specialized material to it adds the processing power required about thousand fold, so all you're really doing for one polygon is asking 1000 more 'power' for it with such a material, which isn't very beneficial.
No it doesn't. The polygon density of your mesh isn't going to have a massive effect on the performance of the material shader. The number of visible world-space pixels on the screen using said shader however, is a different matter.
Replies
http://udn.epicgames.com/Three/MaterialsCompendium.html
gives the usage on every node including the bumpoffset that your looking for.
http://udn.epicgames.com/Three/MaterialsCompendium.html#BumpOffset
http://eat3d.com/materials
since that is a lot of instructions, for just one material and it could get expensive fast.
just checked it and it takes 212 to 246 instructions to do that. so that is a lot of gpu cycles to eat up and in almost all cases it would be cheaper to use some more memory for some more faces on your model.
Depends really, for terrain it can be useful
Though I agree that that particular implementation is rather heavy, you can achieve similar stuff with 100-150 instructions, it's just there might be stepping problems if the displacement is too large.
Let me put it this way, for every polygon you use, applying a specialized material to it adds the processing power required about thousand fold, so all you're really doing for one polygon is asking 1000 more 'power' for it with such a material, which isn't very beneficial.
Also, Bumpoffset is more then enough for terrain. Just disable your compression (Uncompressed) with importing your normal map and Offset map, and use a constant to control the depth of it, while applying a power node if you want harsher contrast. Simpler solutions sometimes are the best.
Yes well, I suppose it depends on the scene. If all you have is an open terrain, you can use a bit heavier materials there. Whatever performance allows.
And I wouldn't say that every polygon with specialized material adds to the processing power requirements. These are mostly pixel shaders, no?
both which would look better and take less resources.
only find the bumb offset worth it for more subtle things where you dont need to to self shadow like that.
No it doesn't. The polygon density of your mesh isn't going to have a massive effect on the performance of the material shader. The number of visible world-space pixels on the screen using said shader however, is a different matter.