Processing of images taken by satellites CARTOSAT-1, SPOT 5, Corona, Hexagon

[nothing_more]

Hi all!
The situation is this: I have three stereo pairs of the same territory, obtained by the CARTOSAT-1 satellite. All this data is separately aligned, and the RPC coefficients are loaded. But only for one stereopair it is possible to build a dense point cloud. When trying to build point clouds for two other stereopairs, the error "Assertion "2389128301877, A.unproj()->B.proj()->B.unproj()->A.proj() error found=7727.49" failed at line 2167!" appears. TXT files with RPC coefficients for all TIFF files have a similar structure, images are not broken, tie point clouds are built without problems. Maybe someone has encountered a similar problem? I use Metashape version 2.0.2.
There are also a number of stereo pairs obtained by the SPOT 5 satellite. But they have some special format for the satellite camera model - files with the DIM extension. I've read that it could be possible to use GDAL to convert DIMAP to common formats, but this seems to apply to XML files. Does anyone have any ideas on how to generate files with RPC coefficients for SPOT 5?
And the last question: is it possible, using Metashape, to develop a universal, reliable method for processing reconnaissance satellite images taken on film that have complex distortion geometry (for example, from Corona, Hexagon satellites)?
Thank you for attention!

[Alexey Pasumansky]

Hello nothing_more,

It would be very helpful, if you can share minimal dataset (via any convenient file transfer service) that can be used to reproduce the problem on our side.

There were some user reports related to this assertion for satellite imagery processing, but they are not able to share the data, unfortunately.