A solution for the 128 tag limit.

[EBE49704]

Hi MrClock8163,
I found that Meshroom's StructureFromMotion(SFM) node only checks the cctag.desc file at the end of the SFM process. The SFM node ignores everything in the cctag.desc file except for the first 8 bytes (which represent the feature count), and it allows the feature count to exceed 128.
This means we can bypass the maximum limit of 128 tags by manually constructing the matches.txt file (which is output by the FeatureMatching node).
The construction of matches.txt is as follows:

f"{view_ID_A} {view_ID_B}"
"1"
f"cctag4 {len(tags)}" # The number of tags co-observed in the image pair.
f"{feature_index_A} {feature_index_B}" # The "feature_index" is the line number minus 1 for each tag entry in the cctag.feat file.
...

I've tested this solution, and it works. But I don't know how to code Meshroom nodes. Could you make an update for this?
The image below shows the results of using this method for motion capture calibration:

[MrClock8163]

Hi!
I remember looking into something like this when I created the nodes here, but as this was eventually outside the scope of the project, I set it aside. I appreciate the info.

A potential problem, that I see with this is, that it might be problematic to implement this in a way, that can be used in combination with Meshroom's own feature matching (with SIFT features for example).

I'm currently busy with another project, but I can certainly take a look at this again after that's more or less done.

(As far as I can tell from the work in the Meshroom repository, Meshroom is hopefully gonna finally get a long overdue update. By the looks of it, it's gonna come with a better plugin system. So I'll also have to look into porting my nodes to the new system.)