Return to Answer

One option would be to render each model against the material's individual leaf nodes into attachments, then use a composition graphics pipeline to create the final image. Different materials can be numbered using an integer attachment, and they can be composed using a method simimlar to green-screen keying. Looking around, this seem to have a name: "deferred rendering".

GLSL doesn't support VLA as C99 does, so if it were to be done in shader code, there has to be a lot of custom codes. This is not flexible enough for me.

Please do refute me in the comment if there's any shortcoming of the above approach.

One option would be to render each model against the material's individual leaf nodes into attachments, then use a composition graphics pipeline (combined with a separate sub render pass) to create the final image. Different materials can be numbered using an integer attachment, and they can be composed using a method simimlar to green-screen keying. Looking around, this seem to have a name: "deferred rendering".

GLSL doesn't support VLA as C99 does, so if it were to be done in shader code, there has to be a lot of custom codes. This is not flexible enough for me.

Please do refute me in the comment if there's any shortcoming of the above approach.

One option would be to render each model against the material's individual leaf nodes into attachments, then use a composition graphics pipeline to create the final image. Different materials can be numbered using an integer attachment, and they can be composed using a method simimlar to green-screen keying.

GLSL doesn't support VLA as C99 does, so if it were to be done in shader code, there has to be a lot of custom codes. This is not flexible enough for me.

Please do refute me in the comment if there's any shortcoming of the above approach.

One option would be to render each model against the material's individual leaf nodes into attachments, then use a composition graphics pipeline to create the final image. Different materials can be numbered using an integer attachment, and they can be composed using a method simimlar to green-screen keying. Looking around, this seem to have a name: "deferred rendering".

GLSL doesn't support VLA as C99 does, so if it were to be done in shader code, there has to be a lot of custom codes. This is not flexible enough for me.

Please do refute me in the comment if there's any shortcoming of the above approach.

One option would be to render each individual leaf nodes into attachments, then use a composition graphics pipeline to create the final image. Different materials can be numbered using an integer attachment, and they can be composed using a method simimlar to green-screen keying.

GLSL doesn't support VLA as C99 does, so if it were to be done in shader code, there has to be a lot of custom codes. This is not flexible enough for me.

Please do refute me in the comment if there's any shortcoming of the above approach.

One option would be to render each model against the material's individual leaf nodes into attachments, then use a composition graphics pipeline to create the final image. Different materials can be numbered using an integer attachment, and they can be composed using a method simimlar to green-screen keying.

GLSL doesn't support VLA as C99 does, so if it were to be done in shader code, there has to be a lot of custom codes. This is not flexible enough for me.

Please do refute me in the comment if there's any shortcoming of the above approach.