iitter_solve with gcps

[James Hollingsworth]

Hi Oleg,

Is it possible to allow the use of --fix-gcp and --use-lon-lat-height-gcp-error in jitter_solve? (I have a case with very good gcps)

And would it be too unstable to free up solving intrinsics from within jitter_solve too? Currently, I'm iterating between jitter_solve and bundle_adjust for these different functionalities, so I was wondering if it could be simpler to do everything from jitter_solve?

Cheers,

J

[Oleg Alexandrov]

I added the options  --fix-gcp and --use-lon-lat-height-gcp-error to jitter_solve, with the same logic as for bundle_adjustment. This will be in tomorrow's build.

Solving for intrinsics in the jitter solver would require a large amount of code work, as there's a lot of book-keeping involved in the bundle_adjust logic for that (and many controls for how fine-grained the sharing of intrinsics should be). It would be a big undertaking to factor out that bundle_adjust code and then call it also in the jitter solver.

For now, if you are not doing that already, one can pass the lists of images and cameras as output by each of these tools to the other tool via --image-list and --camera-list.

BTW, recently I added some more options to the dem2gcp program (as documented in the news). This is quite handy when fixing local horizontal registration errors. Particularly the newer option --max-disp for this program is of help when too much changed on the ground and a DEM-to-DEM disparity (and corresponding GCP) ends up having inaccurate portions.

--
You received this message because you are subscribed to the Google Groups "Ames Stereo Pipeline Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ames-stereo-pipeline...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/ames-stereo-pipeline-support/fa8a954c-3e36-4428-b279-9bf89d7e5c54n%40googlegroups.com.

[James Hollingsworth]

Hi Oleg,

(Sorry for the delay, I was away for a few days.)

This is great news! Many thanks for the updates. I suspected adding solve intrinsics would be complicated. 

I didn't try dem2gcp yet, but l might try that too for this case, as it could be appropriate.

Thanks again for your amazing efforts...

James