We have a set of hand selected empty frames with problematic background. We were running including those frames, but things were fragile. @bw4sz fixed the underlying fitting issues with empty frames and added random likely empty frames from Lindsey's counts, but it looks like we haven't actually done runs with either just the hand picked empty frames or a combination of hand picked and model generated random ones.
We have a set of hand selected empty frames with problematic background. We were running including those frames, but things were fragile. @bw4sz fixed the underlying fitting issues with empty frames and added random likely empty frames from Lindsey's counts, but it looks like we haven't actually done runs with either just the hand picked empty frames or a combination of hand picked and model generated random ones.