On Sep 26, 6:04 pm, "T. Modes" <[email protected]> wrote: > You can use the multi-row detector also with other cp detectors (like > panomatic, select then one-step detector). > But using the two-step approach (involving autopano/generatekeys from > autopano-sift-c) has advantages. In this case each image is only > probed once to find keypoints. The keypoints are cached to disc. At > the end the temp files are deleted.
I'd like to understand this process properly. I ran a test with a panorama from 8 brackets of 3 images each, using align_image_stack as the stack aligner and the generatekeys/autopano combination for the interstack alignment. First CPs were generated for the 8 brackets and put into my panorama. Then the second image of each stack was put to the SIFT by generatekeys, resulting in 8 keyfiles. Next, there was a pairwise comparison between sequentially adjacent keyfiles. What the advantage of such a process over treating all second-in-stack images with a one-pass CPG would be I cannot figure out, particularly since finally there was another use of the previously generated keyfiles, putting them all together into the autopano process. If they had been looked at by a one-pass CPG, this step would have been unnecessary. Over all, the SIFT algorithm was performed once per image, as would be in a one-pass CPG, but rather than just lumping it all together into one tree and doing a global match, there was some pairwise comparing done. What does one gain from the additional pairwise comparison? Or what am I missing? If the SIFT keyfiles had been generated when the images in the stacks were looked at by align_image_stack, there could have been some gain, but I could not see anything along these lines happening. Did I not look in the right place [my TEMP directory]? Or is there something wrong with setting the whole panorama-of-stacks process like I did? Since I'm just in the process of asking questions concerning CPGs, let me add these questions: Is there a way to automatically sort the images into stacks by telling hugin it's a series of brackets with N exposures each? I'm thinking here along the lines of enfuse_align_droplet.bat Is there a way to let hugin pick the first image of each bracket rather than the second for the positioning run? My Canon does them +/-0, -X, +X, and the default behaviour is rather inconvenient since it will look at the underexposed images only which are really just there to get the sky right. Alternatively, is there a way to make hugin also find CPs for the other two expoure levels [apart from manually]? This would make sense, because different parts of the images become good sources for CPs with different exposure levels. with regards KFJ -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/hugin-ptx
