New Material Type For Threejs #28543
Conversation
📦 Bundle sizeFull ESM build, minified and gzipped.
🌳 Bundle size after tree-shakingMinimal build including a renderer, camera, empty scene, and dependencies.
|
|
Can somebody help me with the unit test. Thanks ! |
|
Looking at the code, it seems the computations are less expensive than the standard material but more expensive than the normal lambert implementation. So I don't think it would be a candidate for replacing the so far cheap lambert material. Besides, do you think it's possible to implement this in Or even better, do you think you can implement this material with the new node based material system? If you write the shader code with TSL, it will work for WebGL and WebGPU at the same time. At the current state of the project, it's probably better to not add more default materials in |
Yes that is true but it should not be candidate for replacement for MeshLambertMaterial. As that is extremely cheap, this is just a few steps more but a lot better approximation .
I think I can do that but should not it be a new materia in the Three.js library?
I have not looked into the node based material yet. I will look into it .
Sure, I feel it's a good candidate as a separate cheap back-scattering material . Tell me if you feel otherwise . |
|
Related #8354. I think it would be nice to have a diffuse material that exhibits retro-reflection at grazing angles. The moon has often been used as a canonical use case. Some interesting images here: https://www.rombo.tools/2021/12/15/lambert-sphere/. If the material could be added as a Node Material, I think that would be great. If you need help, there are contributors here who can help you. |
Well, not in the core. But putting it in That said, a node-based implementation would be the preferred, more future-proof approach. |
| vec3 wo = directLight.direction; | ||
| vec3 wi = geometryViewDir; | ||
| vec3 irradiance = shadeLambertianSphereBRDF(wi,wo,geometryNormal,kd); | ||
| reflectedLight.directDiffuse += irradiance; // no need to multiply with BRDF_Lambert( material.diffuseColor ); |
There was a problem hiding this comment.
This is going to need to be corrected.
The quantity on the left of the equals sign has units of radiance, but the quantity on the right you are calling irradiance. So that is incorrect.
(The BRDF has units of inverse steradians, and converts the irradiance units to radiance units.)
Units matter.
There was a problem hiding this comment.
Changing a variable name does not fix the problem. The units on the left side and right side of the equals sign must match.
|
|
||
| } | ||
| /* struct IncidentLight { | ||
| vec3 color; |
There was a problem hiding this comment.
The color and intensity of the incident light is missing from this implementation.
There was a problem hiding this comment.
This is fixed by modifying the implementation to support the incident light color and intensity. It is not fixed by removing a comment.
| @@ -9,6 +9,7 @@ import { MeshPhongMaterial } from './MeshPhongMaterial.js'; | |||
| import { MeshToonMaterial } from './MeshToonMaterial.js'; | |||
| import { MeshNormalMaterial } from './MeshNormalMaterial.js'; | |||
| import { MeshLambertMaterial } from './MeshLambertMaterial.js'; | |||
| import { MeshSphereLambertMaterial } from './MeshSphereLambertMatertial.js'; | |||
There was a problem hiding this comment.
- import { MeshSphereLambertMaterial } from './MeshSphereLambertMatertial.js';
+ import { MeshSphereLambertMaterial } from './MeshSphereLambertMaterial.js';
This material is a much better approximation for diffuse materials for dusty/porous
surfaces that exhibits back-scattering and saturation effects that are missing from prior diffuse models like Lambert. It has a very low cost as compared to MeshStandard /MeshPhysical material. Can be used as a cheap alternative for surfaces like skin/clouds/pots...etc .
The method :
It has been directly taken from the recent 2021 paper : https://d1qx31qr3h6wln.cloudfront.net/publications/lambertsphereBRDF.pdf