Say you have 500 stills comprising a video, and you also have the original blend file that created the stills. Is there a way to do a render that skips the time-consuming Image pass but still provides various other passes (like Mist, Shadow, etc.)? And such that those passes are stored in a way that can be input to the compositor to apply the result to the stills? I have done very skimpy passes (low bounces, etc.) that sort of achieve this, and stored them as multilayer EXRs, but haven't found a way to apply the output to more than one frame at a time.
I think the way to do this is as you said: low bounces, no caustics, etc.
You can bring the already rendered image sequences into the compositor with the image node:
Add an image node (ShiftA> Input > Image)
Open the first frame of your image sequence in the image node
Change the source type to Image Sequence:
Set the Frames, Offset and Start frame values.
- Frames is the number of frames to use.
- Start frame is the scene frame on which to display the first frame
- Offset is the amount to offset the current frame. (e.g. with Offset =
5, the image for frame 6 will be read instead of the image for frame 1)
You can load your sequence of Multilayer images alongside your original render into the compositor in this fashion. If you rendered the Multilayer images with the render passes enabled you should be able to use the passes from the Image node.
You can also use this alongside the Renderlayer node to composite with the render directly, e.g. with a
150 frame image sequence: