How to get world position from depth on orthographic camera?

[Eric.J]

Pixel position in world space from fragment postprocess shader issue

Welcome aboard! The depth renderer is rendering linear depth in the depth texture by default but you would need the non linear z value if you use it “as is” as the z NDC coordinate. You can instruct the depth renderer to render non linear z values by passing true as the second parameter to enableDepthRenderer: Note I’m not quite sure what you want to achieve with the vectorToPixel computation, I’m just returning the world space coordinate. If you want to use linear z values instead you can …

This goes well on perspective camera.

Babylon.js Playground

But get no effect on orthographic camera.

Babylon.js Playground

Is there any different?

[Eric.J]

I get the right effect like this.

Babylon.js Playground

So the depthMap for perspective camera and orthographic camera is different?

[Evgeni_Popov]

Yes, the projection matrix is different between perspective and orthographic mode, so the depth values are different too. In the orthographic case, the z values in clip/NDC space are in the [-1,+1] range, so your fix is the right one.

[Eric.J]

And the range is [0,+1] in WebGPUEngine ?

[Evgeni_Popov]

@Eric.J

And the range is [0,+1] in WebGPUEngine ?

Yes. You can refer to this page for the maths. It’s for shadow generation, but the orthographic/perspective matrices are the same for cameras:

• for orthographic case, refer to “Directional Light”:
  • WebGL: Babylon.js docs
  • WebGPU: Babylon.js docs
• for perspective case, refer to “Spot light”:
  • WebGL: Babylon.js docs
  • WebGPU: Babylon.js docs

[Eric.J]

Thank you so much!