Thanks for the help with my row filter problem. I almost don't dare to say it but I'm facing yet another problem. This time I want to join a label that has been filtered and reassembled by the "Image Group By" node with the original image. The joiner node just cant join these two columns and I have no idea why.
In brief I joined a label and it's corresponding image stack, then I used an "image segment features" node to be able to filter the labels based on their max value. After applying the respective row filter I use the Image Group By node (Compose labeling mode) to create a new labeling containing only the filtered labels. Now I would like to join the original image stack with the newly built filtered label to use them in a segment overlay viewer. And this step always fails. I've tried all possible settings in the "Labeling Compose Options"-tab of the Image Group By node but the joiner node just can't do the job.
in the Image GroupBy node we still have the problem, that we can't keep the images rowid. We hope to fix this in the future. The new RowIds of these Labelings are Row0, Row1, ... RowN
What you can do for now: Use the rowid (or row id) node after the image you want to join the composed labeling with. Then these rows will also have a rowid from 0 ... n. You should be able to join the image and the composed labeling now.
PS: Feel free to ask whenever you want. This is very valueable feedback, as I guess many other users face the same problems and we maybe can learn from these issues to make things easier in the future! :)
if you use the solution proposed by Christian D. please make sure to check the "Retain row order" option in the image group by node (which might slow down the grouping a bit). But there is also an alternative solution: instead of joining the tables by their rowid you can just join them according to the labeling (if available in the tables to join), i.e. the original labeling and the labeling you selected as grouping column. I attached an example workflow.
Okay thanks again guys for your help. It works perfectly now.