LuxCoreRender Forums

The author of the technique byob shared is the same developer who wrote the tangent space normal mapping standard we use in lux. He has actually posted a new method that is state of the art. It also allows mixing of bump and normal maps, and to my knowledge doesn't require uvs.

He also provides source code.

https://mmikkelsen3d.blogspot.com/p/3d- ... s.html?m=1

I thought I would do a simple cube projection UV's but that also ruins bump mapping quite a bit. bump simply needs a good proper unwrap, or seam will show.

lacilaci wrote: ↑Thu Jan 09, 2020 6:54 am I thought I would do a simple cube projection UV's but that also ruins bump mapping quite a bit.

I was thinking the same... but if it's not solving and it needs a correct Unwrap...well we loose the meaning of triplanar mapping.

marcatore wrote: ↑Thu Jan 09, 2020 7:28 am

lacilaci wrote: ↑Thu Jan 09, 2020 6:54 am I thought I would do a simple cube projection UV's but that also ruins bump mapping quite a bit.

I was thinking the same... but if it's not solving and it needs a correct Unwrap...well we loose the meaning of triplanar mapping.

well, triplanar works. Bump mapping has issues

ok..but today..is very strange to use just diffuse and not bump...so if you use triplanar mapping for diffuse and not for bump mapping..it's not the best...or am I missing something?

marcatore wrote: ↑Thu Jan 09, 2020 7:54 am ok..but today..is very strange to use just diffuse and not bump...so if you use triplanar mapping for diffuse and not for bump mapping..it's not the best...or am I missing something?

Yes, I'm just stating that this issue is not because of triplanar not working. It's because bump mapping needs good UV's.

Understood but it still a big problem...in my opinion.

Dade, you say that you need a "space" to reference the axis...but why object space or world space is not suitable?
I know that it's not so useful but I've checked the Vray implementation and they have 2 mode: local space or reference node.
If you set reference node and you don't set any reference object, it uses the world space.

marcatore wrote: ↑Thu Jan 09, 2020 8:43 am Dade, you say that you need a "space" to reference the axis...but why object space or world space is not suitable?

It is not an object "space" but the intersection point "space" (i.e. Z along the point normal, etc.).

patrickawalz wrote: ↑Thu Jan 09, 2020 12:22 am The author of the technique byob shared is the same developer who wrote the tangent space normal mapping standard we use in lux. He has actually posted a new method that is state of the art. It also allows mixing of bump and normal maps, and to my knowledge doesn't require uvs.

He also provides source code.

https://mmikkelsen3d.blogspot.com/p/3d- ... s.html?m=1

However this looks like a totally new framework and it would require to change all bump map related code including all textures and materials. It is a lot of work.

Dade wrote: ↑Thu Jan 09, 2020 10:41 am

patrickawalz wrote: ↑Thu Jan 09, 2020 12:22 am The author of the technique byob shared is the same developer who wrote the tangent space normal mapping standard we use in lux. He has actually posted a new method that is state of the art. It also allows mixing of bump and normal maps, and to my knowledge doesn't require uvs.

He also provides source code.

https://mmikkelsen3d.blogspot.com/p/3d- ... s.html?m=1

However this looks like a totally new framework and it would require to change all bump map related code including all textures and materials. It is a lot of work.

And from the examples shown, skin bump aside, it doesn't look that great.

Some simplified approximation isn't possible for cases of no uv present? Just to have a better fallback than weird looking shading. Combined with a warning in blender it could be "good enough"