Dear group! I introduced HDR blending with pv some time back, and since that time I felt tempted to tackle both exposure fusion and image stitching as well. Hugin usually delegates these tasks to enfuse and enblend, and I have used both programs to good effect for a long time, so I was aiming at using similar techniques in pv. enfuse and enblend rely on multiresolution blending, as published by Peter J. Burt and Edward H. Adelson in their article 'A Multiresolution Spline With Application to Image Mosaics', which was used for exposure fusion by Tom Mertens, Jan Kautz and Frank Van Reeth, as described in their article 'Exposure Fusion'. pv uses b-splines for interpolation, and it's use of image pyramids relies on a variant of image pyramids based on b-splines. This has proven effective for viewing images with pv for many years now, and it's also what I use for image pyramids in my implementation of the adapted multiresolution blending algorithm. This provides a fresh, modern implementation based on my library vspline, <https://bitbucket.org/kfj/vspline> which is fast because it is multithreaded and uses SIMD code. I have pushed a prototype to the master branch of my pv repo <https://bitbucket.org/kfj/pv> which offers both exposure fusion and image stitching using this new code, so for now the code is Linux-only - if you feel adventurous, do your own merge to the msys2 or mac branches. To try it out, simply load a PTO containing a registered exposure bracket or a panorama into pv, select the ROI and press 'U' for an exposure fusion or 'P' for a stitch. You may want to pass --snapshot_magnification=... on the command line to get larger output, other snapshot-related parameters should apply. Consider this a 'sneak preview' - the code is, as of this writing, not yet highly optimized, but should be functional. The output is rendered in the background and may take some time to complete. Best work from the command line, where there is some feedback on the process. I'll tweak the code in the days to come, and I intend to provide a version with alpha channel processing as well. I'll post again once everything is 'production grade' and fully documented. One large benefit of having these new capabilities in pv is that it does away with the need for intermediate images and external helper programs: the intermediates are produced and processed internally in RAM, so there is no disk traffic, and the internal intermediates have full single precision float resolution, which makes for good image quality. On the down side, due to the RAM-based intermediates, output size is limited. Kay
-- A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ --- You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/b5684495-9bb9-4558-bc72-e3c4fb6d9b91n%40googlegroups.com.
