Ask a question about implement coordinate space conversion from WebGPU to OpenGL in wgpu-rs

[ghost]

Hello,

I've been investigating how wgpu-rs handles the conversion between the coordinate spaces of WebGPU and OpenGL, and I'd like to confirm my understanding of the implementation details. I have outlined my findings below and would appreciate any clarification or confirmation:

Screen Space Y-Axis: In WebGPU, the y-axis in screen space goes from top to bottom, whereas in OpenGL, it goes from bottom to top. My understanding is that the final step in the wgpu-rs implementation uses BlitFrameBuffer to flip the y-axis of the internal FBO. Is this correct?

Texture Coordinates Y-Axis: Similarly, in WebGPU, the y-axis for texture coordinates goes from top to bottom, while in OpenGL, it goes from bottom to top. I believe that the wgpu-rs implementation uses Naga's ADJUST_COORDINATE_SPACE to indirectly achieve this by setting gl_Position.y = -gl_Position.y in the vertex shader. Could you confirm if this is accurate?

Z-Axis Values: WebGPU uses a z-axis range of [0, 1], while OpenGL uses a range of [-1, 1]. My understanding is that ADJUST_COORDINATE_SPACE modifies the z-value in the vertex shader with the transformation gl_Position.z = 2.0 * gl_Position.z - gl_Position.w. Is this the correct implementation?

I would be grateful for any insights or corrections you could provide. Thank you for your time and consideration.

Best regards,
Moyy

[cwfitzgerald]

Yeah, that all seems correct

See https://github.com/gfx-rs/wgpu/blob/trunk/wgpu-hal/src/gles/egl.rs#L1112-L1114 for the final blit on present.

And https://github.com/gfx-rs/naga/blob/master/tests/out/glsl/quad-vert.main.Vertex.glsl#L47 for what we insert in generated GLSL vert shaders.

[ghost]

thank you very much for help me.