Peter, You seem to have created a very good stitch and I don’t think I can make any suggestions that will help you improve it.
What I would mention, more for another time than because it will make any noticeable difference here, is that, if you are adding h and v control points to ensure that you have a rectangular, correctly oriented frame for the stitch, you should also allow the values of y, p and r for the anchor image (17 in your case) to be optimised. Otherwise the perspective of the anchor image may not match that of the stitch as a whole, although in this example the discrepancy is probably dissipated amongst the other images. On the image 4 problem, about which you wrote again today, I ran your b version of the project file on the images you supplied and did not meet the strange darkening of image 4 you showed. I also replaced the control points with a complete new set. The result stitched to give a panorama equivalent to yours, but again without the extreme darkening of image 4. Why it happened, or what can be done to avoid it, I do not know. I do wonder if possibly your stitch comes from an earlier version of the project file. Progressively extending the scope of the optimisation can be used to add images into the optimisation in the way Bugbear suggested, and I have been forced to do that. But if possible I include the full set of images and add classes of parameter cumulatively. My reasons for doing so are: 1. It can be helpful to restrict what the optimiser is trying to do at one time. If too much is asked of it, it may be overwhelmed (forgive me for being a bit anthropomorphic about this). 2. One can stop as soon as a good stitch is achieved and thus achieve the stitch with fewer parameters needing to be optimised. The more parameters you include in the optimisation, the more control points you need (I think). 3. At each stage I try to refine and extend the set of control points. My objective at this stage is as much to achieve a robust and comprehensive set of control points as to do the stitch itself. Once one has a good set of control points I find that one can usually reset all the optimised parameters to 0 (but leaving them optimisable) and then optimise the whole thing in one big bang. The result should be as good as, or possibly even better than, the original optimisation/stitch. I think the need to create control points oneself is a normal, if boring, part of stitching photos of old maps, where control-point detectors can be thrown by the splodgy nature of the features in the images. But there is a great resource to help solve this problem, as you may well know. Once the images are approximately positioned, by any means available, extra control points can be added by the ‘Create control points here’ feature that can be invoked in the Preview tab of the Fast Preview panorama window. See here: http://hugin.sourceforge.net/tutorials/hugin-2015.0.0/en.shtml You asked why I preferred to use a single lens. I am not at all clear why Hugin suggests here that each image should have its own lens. If images do not all have the same dimensions, separate lenses need to be used. They may be required in other circumstances, too. And if the settings of the lens are changed, the purist will say that each set of settings should have its own set of lens parameters. But here it seemed to me that your set-up was sufficiently fixed to mean that any variations that were present from features like automatic focussing were likely to be masked by variations in other factors, such as the exact positioning of the control points. In those circumstances it seemed to me that it was closer to reality to assume that only a single lens was involved. Then, even with the full set of parameters a to e, only 5 were needed. I notice that you have allowed b to vary between the different lenses, one for each image. That requires 20 parameters and they vary between roughly -0.016 and -0.020. I suspect that this range is more than is needed specifically to compensate for differences between the lens distortions for the different images and has to do with the fact that the more you give the optimiser to play with the better result you get (provided there are enough control points, of course). In other words variations in b are used to adjust the relationships between the images. Roger Broadie ======================================== Message Received: Dec 20 2016, 11:34 AM From: "Peter Cooper" To: "hugin and other free panoramic software" Cc: Subject: [hugin-ptx] Re: Stitching photographs of a large map into a whole map Thanks again. I have had a go at another map to see how the process can work when you know what you are doing! The input and output files are again in my DropBox. It went well and the end resulting average was less than 1 pixel. The map itself is different from the first one I did (it is adjacent, and I had fewer images) but I had a similar problem with some pairs (adjacent vertically) not being given any control points by the automatic creation process, and getting all control points removed by the cleaning control points process. I wonder if this is a feature of stitching maps or a bug? Incidentally this map had images taken without the camera's macro being on but I do not see much difference in the end product from my first map, which had the macro on. I cannot find out why my first stitched map had a black area at bottom right, but that does correspond to the part of the map that is in image 4 and not other images. -- 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 hugin-ptx+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/96989e65-ecd6-4e8a-9d54-9134b23f38fc%40googlegroups.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 hugin-ptx+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/21258152.39921482434769250.JavaMail.www%40wwinf3709. For more options, visit https://groups.google.com/d/optout.
