Reference based motion correction jobs can be quite time consuming, so failures part way through the dataset (e.g. due to a particular micrograph with very many particles overflowing GPU memory) are relatively costly. Two issues (2nd may be a feature request).
- Mark as complete currently doesn’t work for RBMC particle stacks, subsequent jobs fail with errors like these (input particles were re-linked after force completion)
- Mark as complete also doesn’t populate the failed micrographs output slot, which would allow particle extraction to be continued with a new job