thank you for your kind words about BiofilmQ We are always grateful for feedback.
As you already figured out correctly, the Track IDs are based on the first frame. In the dropdown menu (marked red in the image below) you have the options to either assign each individual cube/cluster in the first frame a different ID (this is the option you currently have selected) or assign all cubes in the first frame the same ID (thus only newly emerging cubes would be assigned a different ID).
The next important choice for controlling the track IDs is to decide whether you want each individual cube of the first frame to be assigned a unique ID or if you want to assign the IDs based on clusters of cubes (marked cyan below). Clusters of cubes are cubes which are neighbors as defined by the search radius (marked magenta). So, if you want to have fewer track IDs, one possibility is to check the checkbox next to “Assign same TrackID to connected clusters” and then play around with the search radius.
From your images I can see that you have a lot of small objects - possibly single cells - and a couple of bigger clusters. If only the large clusters are of interest to you, you may also consider increasing the minimal object size (in the post-processing part of the segmentation) such that all the single cells are filtered out and only the clusters remain. This would be a different way of decreasing the number of initial track IDs, the usefulness of which is of course very dependent on your research question.
I hope that this advice is helpful to you. Let me know how your analysis works out and if you encounter further questions, don´t hesitate to write again!