It’s actually not that bad when you keep the ‘macro’ images in a different Dataset.
When you want to move the images to a new group, the webclient will ask if you want to move all the images from the fileset (since you can’t split the fileset between groups). The images will get moved into one Dataset in the new group, but that’s not too painful.
Similarly, if you try to Delete just 1 image from a fileset, the webclient won’t allow it and this time will ask you to actually select ALL the images to delete, in order that you don’t accidentally “Delete All”.
One of the options considered at https://github.com/ome/design/issues/107 is to allow partial deleting of a fileset, to permanently remove the images you don’t want.
The advantage of using separate Datasets compared to a custom filter is that other clients will show the same Images in the Dataset (without needing to implement the same filtering). For example Insight or even Orbit.
Because we have to deal with hundreds of different file formats, it’s unlikely we’d support behaviour that is specific to one or two formats.
It’s certainly possible that we’d support the ability to save a filter on a Dataset. We’d need some ‘standard’ way of storing the filter, possibly as some sort of annotation on the Dataset.
However, even in that case, we would likely ask the user to choose which images are filtered, since making that decision is very dependent on the types of images, the file format and personal workflow preference. So, it may not have that many advantages over the existing option of moving images to a different Dataset.
If you wish to ‘automate’ the hiding of images in a different Dataset, you could consider scripting this move after images have imported (run the script over a recently-imported Project or Dataset). Or if you are importing the images via a client-side script (CLI, Python or Java) then that could hide some images in a different Dataset as part of the import workflow.
Let us know how that approach works and what are the ‘pain points’ that need addressing,