Floating-point numbers often add a small rounding error. Maybe the slightly negative number isn't a bug but such a thing...
On Sun, 24 Feb 2019, 00:55 AKS-Gmail-IMAP, <[email protected]> wrote: > I can comment on some of this. I am currently using the "method 3” to > create “contact sheet” images of 35mm photograph film rolls archived in > style “35-7BXW” PrintFile archival negatives preservers. These are clear > plastic negatives preserves holding seven rows of six 35mm image negative > strips. These scan really well on a desktop flatbed photo scanner in > transparency mode, but it takes three scans to do one sheet. Therefore the > scans have to be stitched together. Hugin using method 3 does the job. > Furthermore, because the negatives might be at radically different > exposures, Hugin’s stack handling where the stacks are multiple scans at > different exposures allows exposure fusing all the images in the contact > sheet to be satisfactory images. > > When using this mosaic XYZ mode I always see the “Panorama Tools” > "Panorama must have positive height” message when the “edit script before > optimizing” check box is checked where the “Continue with these changes” is > pressed at the “Edit Panorama Tools Script” form. I would be clueless > trying to edit the script. I never see any negative numbers in the script. > Hugin does show “-0” in the Z plane parameters when it calculates the > parameters. The number is actually a small negative number when viewed in > the edit box. The "Panorama must have positive height” message will show > even after reseting all parameters to 0. Perhaps this is a bug. I would not > know since I am pretty much cook booking my way through this part of Hugin. > > By the way, Masks are essential to use in “method 3”. > > Regarding engineering drawings, the type and source of which has not been > described, I do not see how one could automate the task because each image > requires at least one unique mask and possibly four unique masks and also > because the image control points are best placed by hand. The time spent > removing and correcting falsely placed automatic control points is many > times more the manual effort to create the very few required control > points. The situation is similar to the fallacy digitizing scanned drawings > into computer aided drawings. Incidentally the time spent personally > observing each image in the manual method pays off with additional > knowledge about the images. That knowledge might explain causes for > problems seen in the output. > > The type and source of the engineering drawings is not mentioned. I’ve > worked in that business for many years and I have tried on many occasions > to reassemble plans using every imaginable technique from “xerox art” to > Hugin. I cannot recall any time the source images did not have a different > distortion in both the x and y directions. The results were always > unsatisfactory to some degree, often terrible. I could see x y scale issues > with any first generation roll plotter sources and small media copy > machines. Images created on full sized flat bed equipment were the best but > what company has one of those these days? > > Output from a roll plotting or reproducing device is going to have a > slight stretch distortion as the media passes over the roll. I would see > this in recent plots and also back in the day when running vellums or cloth > through a diazo machine. I would foresee difficulties trying to stitch > plans back together without being disappointed in the results. Even > recently the best results with the least amount of effort and frustration > were always using what we called “xerox art”. This is cut and paste using > Scotch magic tape with perhaps a bit of pencil or pen touchup followed by a > run through a large format copier. > > Regards, Allan > > > On Feb 23, 2019, at 2:40 AM, Bruno Postle <[email protected]> wrote: > > > > > > > > On 21 February 2019 14:10:41 CET, jim cullen wrote: > >> > >> I can't seem to make it fix the roll and pitch at zero successfully > during > >> optimization. > >> > >> I'm setting the fast preview mode to Mosaic plane - is that proper? I > ask > >> because the python script > >> stitch-scanned-images.py which one can find on github constructs a .pto > >> that seems to use Panosphere. > > > > Hugin has so many options that you can do this scanned images thing in > three completely different ways. I'm wary of prescribing a particular > method over another, but you can't mix them. They are: > > > > 1. Using very narrow field of view, with roll, pitch, yaw optimisation. > > Imagine that you have taken these images using a very long telescope, > Hugin will let you stitch them like a 'normal' panorama, as long as you set > the field of view to a small number (<1°). > > The result won't be perfectly correct, and there will be odd curvatures, > but for most purposes you can get away with this. > > > > 2. Using the d,e lens shift parameters to move the images around a > rectilinear 'canvas'. > > These shift parameters are intended to let you work with photos created > with a shift lens, or photos where one or more sides have been cropped away > (nearly all historical/Victorian/magazine photos are uncentred like this). > > So you can stitch scans by pretending that they are all the same photo, > each just cropped differently. > > > > 3. The 'mosaic' XYZ mode has been added to Hugin specifically for > fitting images to a rectilinear plane, so this is what we recommend, but it > is a powerful tool that can do lots more than stitching scans, so it is > conceptually a bit harder. > > The mosaic mode assumes that your camera moves to different 3D locations > for reach shot, hence the XYZ positions. > > It is possible to conceive of your scans as being taken by a camera from > different locations, but with the Z distance (the distance from the camera > to the object) never changing, and your 'camera' is always perfectly > perpendicular, so pitch and yaw are always zero. > > > >> I'm now setting the first cols of the optimize panel to unchecked and > zero > >> for yaw and pitch and leaving > >> the roll free (checked). The script (edit prior to optimize lets me > look) > >> seems to have sane vars viz > >> Tx Ty and r for each image. > > > > This is correct for method 3. > > > >> I must be blundering some place else as it > >> keeps saying Panorama must have positive height. > >> I think this is from the "batch" side of things during optimization. > > > > I've have no idea how this happens. > > > >> I only get the option to set these things if I select Custom on the > >> geometric optimization pull down. Is that > >> the place to set the vars to participate in optimization? > > > > Yes, I would set 'custom' geometric optimisation, and control everything > from the Optimiser tab (for methods 2 & 3). > > > > -- > > Bruno > > > > -- > > 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/7CF8818D-08E1-46CC-B919-320BA64FF7EE%40postle.net > . > > For more options, visit https://groups.google.com/d/optout. > > -- > 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/3FF884DC-A3EE-4247-9997-B16A5CEF9DAA%40gmail.com > . > For more options, visit https://groups.google.com/d/optout. > -- 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/CAMcFxV3FKFNVZhwYjYO1zj5QyPO1Z87iTcBTN6uAL9q6S79ioQ%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
