[hugin-ptx] another paper, maybe useful for masking
Found another paper, this one includes GPL source code: http://grail.cs.washington.edu/projects/photomontage/ video shows masking out of unwanted elements from a sequence of images -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: strahov gigapixel
Excellent work. I'll be sure to visit the Strahov Monastery Library the next time I'm in Prague. Congratulations, you made it onto wired.com: http://www.wired.com/underwire/2011/03/strahov-monastery-panoramic-image/?pid=3251&pageid=64399. -Jeff -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: "stitch large images in just a few seconds"
Hi all, I wrote to Zeev. He is doing some even more interesting stuff: ..."To tell you the truth, at this point I wouldn't bother to implement MVC in order to do image stitching. Right now we are working on a method which allows to approximate solutions of Poisson equation (which is what state-of-the-art image stitching methods use to create the panoramas). The new method will be very fast and even simpler to implement than MVC - no triangulations and stuff like that. We are going to submit it to Siggraph Asia in May and I intend to make the code publicly available as fast as possible after the submission/acceptance."... I invited Zeev to join us and provided the link to this thread too. Anyway here is the source code for MVC he provided to me by email: http://www.diy-streetview.org/data/development/20110329/MVCDemo_by_Zeev_Farbman.zip Jan On Tue, Mar 29, 2011 at 6:41 PM, kevin wrote: > On Mar 29, 12:12 pm, Agos wrote: > > the results are indeed very impressive. I'll leave to those who know math > to > > say if this could really be useful for hugin :) > > Yeah, I don't understand the math either! But it looks like they are > solving the laplace portion in a different way that's faster. I know > from speed tests I've done before if that could be used to speed up > enblend then stitching large panos would be much faster. > > -- > 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 hugin-ptx@googlegroups.com > To unsubscribe from this group, send email to > hugin-ptx+unsubscr...@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/hugin-ptx > -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: "stitch large images in just a few seconds"
On Mar 29, 12:12 pm, Agos wrote: > the results are indeed very impressive. I'll leave to those who know math to > say if this could really be useful for hugin :) Yeah, I don't understand the math either! But it looks like they are solving the laplace portion in a different way that's faster. I know from speed tests I've done before if that could be used to speed up enblend then stitching large panos would be much faster. -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: "stitch large images in just a few seconds"
On 11-03-29 01:48 AM, kfj wrote: Interesting abstract. I couldn't get the full paper downloaded, though. Just an empty page came up :( The link is a bit slow, but the full paper is definitely there; try right clicking and "Save link as". Very cool stuff. Cheers, BBB -- Bob Bright Vancouver Island Digital Imaging (250) 857-9887 bbbri...@victoriavr.ca http://VictoriaVR.ca -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] [REVIEW] Pushing Hugin changes to VIGRA upstream – Hugin changes
On 26 March 2011 16:42, Yuval Levy wrote: > On March 26, 2011 06:05:56 am Lukáš Jirkovský wrote: >> I'll branch out the repository when (and if) the proposed changes get >> their way to upstream. Till that the branch would be useless. > > I beg to differ. Branching in Hg is extremely cheap. Branch out, and save > the patches to that you want to contribute upstream in the branch as well. > > The use is that it makes your contribution persistent and easy to reproduce. > hg pull your branch, apply the patches to the vigra source code that is > already on the system and run. As opposed to: search for the patches to the > vigra source code in a mailing list archive that will forget them over time; > search for the patch to the Hugin codebase on a little known server that may > or may not be available a few months down the road; apply patches to Hugin; > apply patches to vigra source code that is already on the system and run. > > Anyway, I hope your changes get accepted upstream asap. Then it will take > time for the updated vigra version to trickle down the distributions and in a > few years we can clean up another part of the foreign source from the Hugin > repo. > > Yuv > Hello Yuv, I'm not yet convinced it's a good thing to do at this point in time. Still the users would have to manually apply several patches to VIGRA unless there's a public clone of VIGRA with changes applied so applying one more patch doesn't really hurt anything. I plan to send the patches upstream on Thursday or Friday, so we may find out whether the changes to hugin doesn't go in vain quite soon. Lukas -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: "stitch large images in just a few seconds"
the results are indeed very impressive. I'll leave to those who know math to say if this could really be useful for hugin :) -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] strahov gigapixel
Hi Everybody, Just thought you might enjoy this http://www.360cities.net/gigapixel/strahov-library.html It wasn't made with Hugin though. Maybe when I can get cpfind to reduce images to a specified pixel size ;-))) seriously though, I would like to make one of the next ones using hugin, to give hugin some nice publicity! :-) cheers, Jeffrey -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] GSOC 2011 Student Introduction
Awesome! I for one am happy to hear that someone is interested in working on vertical line detection. I review tons of panos (stitched by hundreds of people all over the world) every day, and lots of them are not level. This task is currently quite difficult and non-intuitive for people, and it's the most common error with panos that are otherwise stitched ok. It will be cool when I can point them to Hugin to stitch level panos more easily :-) cheers, Jeffrey -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: "stitch large images in just a few seconds"
On 29 Mrz., 01:32, kevin wrote: > Not sure if people have seen this paper yet: > > http://www.cs.huji.ac.il/~danix/mvclone/ > > Found it by looking at Gimp's Google Summer of code. They state in > the video that using their method they can stitch images in just a few > seconds. Interesting abstract. I couldn't get the full paper downloaded, though. Just an empty page came up :( Kay -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: I want a parameter in cpfind for "reduce image"
On 29 Mrz., 09:56, Jeffrey Martin <360cit...@gmail.com> wrote: > Hi Kay, > > I like the idea, but it seems overly complex (and expensive) to me, to be > honest. And won't work in cases when the FOV is unknown. I am only using my experience with my script to argue your point that tailoring the CPGs input beyond the simple --fullscale is desirable. Of course I'm trying to make everyone interested in my software as well ;-) If the FOV isn't known, my approach won't work. But hugin requires the FOV to be known or at least guessed. And as far as woa is concerned, keep in mind it's scope: it's meant for refining prealigned panoramas for purposes like multi-lens panoramas and lens calibration. > My idea is much, much more simple. > > Couldn't we start with my idea? ;-) Let's. > Or, if I convinced someone (say one of my colleagues who is a competent > programmer) I guess I could simply add it to cpfind, yes? I'd be happy to see progress along these lines. You go ahaead and do the convincing. A simple scaling factor or size parameter should indeed by simple to implement, and using external logic to feed it depending on the FOV of one's images is probably even cleaner than a pixel-per-degree parameter, though of course the latter would be nice to have as well. How about allowing both? If pixels-per-degree is specified and FOV isn't, the program could refuse to operate. Have you done tests with upscaled/downscaled images to see what gains there are to be had? Kay -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: Use RAW Photos with Hugin ?
On 28 Mrz., 23:27, KenO wrote: > Can Hugin use RAW photos? I'd like to shed a slightly different light on the topic. Why does anyone want to process RAW in hugin? There are different motivations: - to wrench the last bit of quality from the images that would be lost in a TIFF: the green channel effectively yields 1.4 times the resolution, for example. Potentially very useful. - to save the RAW conversion - the RAWs are there, you want a pano, you'd like to not have to launch the RAW converter before feeding the data to hugin I'd assume the first motivation will be relevant for special purposes, e.g. calibration or focus stacks. But I don't think it's awfully relevant in most paographers' every day work - converted RAWs or even good JPGs will do the trick just fine. So my suspicion is that the main motivation to use RAW is the second: save the conversion. Maybe it's enough to just defer the conversion? I don't know about other RAW formats, but Canon's CR2 contains en embedded 'thumbnail' - this sound pokey, but in fact it's something like a good 1.5 megapixels in size, and has well enough detail to run a CPG on it. In fact it's very crisp, having been scaled down from the original 12 MP my camera yields. So it's entirely feasible to extract this image, using something like ufraw-batch --embedded-image IMG_.CR2 and make a panorama from the preview images. Don't laugh. The embedded preview has the same geometry as the RAW image. So you can - later, and maybe in a batch run overnight, have the RAWs expanded and the pto file upsacaled to work on the full-size images, and do a stitch from them. I've written a pto upscaler precisely for the purpose - batching the other stuff is (relatively) trivial, but a nice batch file would of course help... The upscaler is here: http://bazaar.launchpad.net/~kfj/+junk/script/view/head:/main/scale_pto.py This one isn't hsi/hpi but plain Python. I admit I haven't looked at the script in a while, but it should work nevertheless. The approach has a few good points, like saving time and space - a bonus for laptop users and travellers. Instead of the emberdded preview image, it's of course just as well to use output from the RAW converter at smaller scaleand in a compact format - in fact this is even better. Why? In the proposed workflow I always felt there was one thing missing. Hugin can very nicely take care of exposure, colour and lens geometry, but there is one point which can be crucial which is left in the domain of the RAW processors: tca. Correcting tca with a RAW procesor changes image goemetry and stands in the way of my 'upscaled thumbnail' approach. You can use an external tca corrector in your toolchain on both the thumbnail and the processed RAW, but this approach has always had less appeal to me since it involves yet another warp with the accompanying degradation in quality - not to mention the extra processing time. Kay -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Google Summer of Code 2011: Call for Mentors
I've contacted a professor here at CVUT (the czech technical university) and asked if any students might be interested in participating. People at CVUT invented MSER http://en.wikipedia.org/wiki/Maximally_stable_extremal_regions and is very highly regarded in the realm of computer vision. -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: I want a parameter in cpfind for "reduce image"
Hi Kay, I like the idea, but it seems overly complex (and expensive) to me, to be honest. And won't work in cases when the FOV is unknown. My idea is much, much more simple. Couldn't we start with my idea? ;-) Or, if I convinced someone (say one of my colleagues who is a competent programmer) I guess I could simply add it to cpfind, yes? -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Use RAW Photos with Hugin ?
Agree, RAW input for stitching is a bad idea. At the very least you want to set white balance on your raw images (although it's true hugin can fix white balance, this is not ideal) Even if Hugin did have raw input, it would simply use dcraw for this most probably. I'm not in favor of adding this as a feature, there are many more important things Hugin needs! -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: "stitch large images in just a few seconds"
So can Autopano Giga not sure if anyone here has used it lately but you can throw a 40-image handheld series of shots at it, and it will align it in a few seconds. Rendering is a different issue, although it is also incredibly fast in comparison to hugin and ptgui, and it has "smart" blending to boot. -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
[hugin-ptx] Re: I want a parameter in cpfind for "reduce image"
On 28 Mrz., 23:55, Jeffrey Martin <360cit...@gmail.com> wrote: > Hi, > > the --fullscale flag in cpfind is ok, but really makes no sense considering > the size of input images could vary by huge amounts. > > wouldn't it make more sense to have a setting for the target pixel size that > cpfind should reduce the image to? This reminds me of the discussion we had in http://groups.google.com/group/hugin-ptx/browse_thread/thread/1f2f441d35598cbe/65f7816f5263746f?lnk=gst&q=pixels+per+degree#65f7816f5263746f where I proposed to allow scaling of the images to a certain pixels- per-degree-of-fov value. So I propose this again. I've had good success with scaling images (of different lenses) to the same pixel- per-degree-value (and the same geometry) before feeding them to the CPG - I do this by warping them with nona to a representation that is as similar and undistorted as possible, calculating the CPs on the transformed representation and mapping them back to original image coordinates. See http://bazaar.launchpad.net/~kfj/+junk/script/view/head:/main/README.woa http://bazaar.launchpad.net/~kfj/+junk/script/view/head:/main/woa.py The approach is computationally intensive, but fully automatic, so you can batch it and do something else. The images have to be prealigned, but the prealignment can be rough (like, by shooting pattern, give or take a couple of degrees). The software is early beta and I've only tested it on Linux. You'll need hsi/hpi and a few other bits and bobs to run the software, but you may find the readme interesting anyway. In my experience it helps to shape the input images of the CPGs - their size as well as their geometry. Kay -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx