Currently we store all of the shapes and frames in memory, might be nice to be able to read from a file and only keep in memory the current frame and N frames around it, then adaptively load them. This would of course only work for -f. This might require us to adopt a secondary format that has an index table for all frames so we can seek towards them. Another option is to have the same format but generate a file that has all the indices of the #FRAME_START commands.
Currently we store all of the shapes and frames in memory, might be nice to be able to read from a file and only keep in memory the current frame and N frames around it, then adaptively load them. This would of course only work for -f. This might require us to adopt a secondary format that has an index table for all frames so we can seek towards them. Another option is to have the same format but generate a file that has all the indices of the
#FRAME_START
commands.