Many 3D graphics libraries like OpenGL model 3D objects like cube using a set of vertexes and 3D triangles. But I am wondering how images are modelled in 3D space so linear transformations can apply. I want to make my 3D software rasterizer to do something like this
I think the one possible way to model images is represent every single image pixel as a point in 3D space with RGB colour value as additional attribute. But it will be slow because images can have more than thousands of pixels.
Any better ideas?