What’s the best approach to configure different batch sizes for ROP USD Output vs USD Render TOP? We’re trying to generate USD files in batches of 20, then render them using USD Render TOP in batches of 5.
The issue we’re encountering is that batch index/attributes appear to get cleared between nodes, preventing downstream TOPs from inheriting these values anyway. Currently, we’ve managed to implement batching for the ROP USD Output, but we’ve had to fall back to single-frame processing for the USD Render TOP.
Any suggestions on how to maintain batch attributes through the dependency chain or create new ones? Custom Python partitioner?