That looks almost right, but you should mask out that darkened corner as well. And for the other images you just need to make sure you haven't left any of those darkened parts in (even if they are small they can disproportionately affect the blend).
Given the nature of the images you can be pretty aggressive in masking out, at least when using Multiblend - it only needs a tiny overlap to avoid leaving a blank area, and will blend images even if they don't overlap. On Tuesday, 13 April 2021 at 22:59:41 UTC+1 Jared wrote: > With enblend. I just re-tested with multiblend and get roughly the same > results. Yes, reasonably sure I masked that out. Here's an example, just > to verify I didn't misunderstand you again: > https://boxdog.legroom.net/public/ffmap-mask.jpg > > I created a mask like that on all 8 source images to mask the square. Is > that what you mean? > > On Tuesday, April 13, 2021 at 3:49:15 AM UTC-5 Monkey wrote: > >> Was that with Enblend or Multiblend? Have you made sure to mask out the >> shadowed/darkened strip at the edge, as well as the actual square? >> >> On Tuesday, 13 April 2021 at 05:19:41 UTC+1 Jared wrote: >> >>> Ah. Well that makes much more sense. Sorry, that's my bad. So what >>> you're seeing is a square that I use to align stuff on my scanner - just >>> aligning against any of the edges around the glass results in the sides of >>> the image being cropped off. I didn't think it'd have an effect on the >>> blending process, so I had planned to crop it out of the stitched image. >>> >>> Here's my latest attempt: >>> https://boxdog.legroom.net/public/ffmap-example3.jpg >>> >>> Sadly, not much of an improvement. Actually a bit worse, as the middle >>> section is darker now as well. After applying the mask, I generated one >>> image with no other changes (not shown - pretty much the same as example2, >>> though), and then for this image I clicked Calculate for Optimize, >>> Photometric, Low dynamic range, variable white balance. Is there some >>> other option I should try to apply proper blending after that border is >>> masked out? >>> >>> Again, I appreciate the continued support. >>> >>> >>> On Monday, April 12, 2021 at 7:13:08 PM UTC-5 Monkey wrote: >>> >>>> That's not the border I meant. It's the thin gray border with black >>>> lines which is at the top and/or left and/or bottom of *all* the >>>> images. It might be part of the scanner lid being visible or something you >>>> put behind the folded page. >>>> >>>> On Monday, 12 April 2021 at 21:34:06 UTC+1 Jared wrote: >>>> >>>>> Ha. So keeping in mind I'm new to hugin and figuring this out as I >>>>> go, my first attempt at creating a mask didn't exactly go as planned. >>>>> Here's a screenshot of the resulting image: >>>>> https://boxdog.legroom.net/public/ffmap-example2.jpg >>>>> >>>>> Technically I think I did mask that out, but I'm guessing that's not >>>>> what you had in mind. :-) >>>>> >>>>> So that's the result of an exclude mask that I added to the two >>>>> left-most source images. Can you provide any more detail on how I >>>>> *should* >>>>> create that mask? I tried referencing the mask tutorial ( >>>>> http://hugin.sourceforge.net/tutorials/Blend-masks/en.shtml), but it >>>>> seems like that's addressing a fundamentally different issue, so wasn't >>>>> sure how to apply that to this image. >>>>> >>>>> On Monday, April 12, 2021 at 6:25:13 AM UTC-5 Monkey wrote: >>>>> >>>>>> The problem is the grey border with black lines on the top and left >>>>>> of the images (and the shadow it casts on the page). Mask those out and >>>>>> it >>>>>> should then blend as expected. >>>>>> >>>>>> On Monday, 12 April 2021 at 03:46:41 UTC+1 Jared wrote: >>>>>> >>>>>>> Hello. Appreciate your continued guidance on this. I got the >>>>>>> canvas size down to a usable state so I can open it in GIMP now. Still >>>>>>> having trouble with the blending, though. I tried Bruno's white >>>>>>> balance >>>>>>> suggestion, and then spent a while fiddling with a bunch of different >>>>>>> options to see if I could come up with anything, but no luck. I've >>>>>>> uploaded scaled versions of the remmaped files, plus the full image and >>>>>>> PTO >>>>>>> file for reference, here: >>>>>>> >>>>>>> https://boxdog.legroom.net/public/transfer/ >>>>>>> >>>>>>> Looking closely at the source images, it looks like there is a >>>>>>> little color difference between the two left-most segments and the rest >>>>>>> of >>>>>>> the map (even though it was all scanned under the same conditions), but >>>>>>> nowhere near what's shown in the final image. I also noticed that the >>>>>>> second column is oddly darker as well - look at the water below point 2 >>>>>>> and >>>>>>> to the left of point 1 around the bottom center of the map. All of the >>>>>>> water should be a reasonably uniform blue, except for those 4 stains in >>>>>>> the >>>>>>> upper-left. >>>>>>> >>>>>>> On Wednesday, April 7, 2021 at 5:48:32 AM UTC-5 Monkey wrote: >>>>>>> >>>>>>>> If the remapped images don't show the difference in colour, but the >>>>>>>> blend still does, could you output a reduced size remapped image set >>>>>>>> and >>>>>>>> upload it somewhere? >>>>>>>> >>>>>>>> On Tuesday, 6 April 2021 at 22:00:38 UTC+1 bruno...@gmail.com >>>>>>>> wrote: >>>>>>>> >>>>>>>>> On Tue 06-Apr-2021 at 13:46 -0700, Jared wrote: >>>>>>>>> > >>>>>>>>> >I have one additional question, if you don't mind - the stiching >>>>>>>>> came out >>>>>>>>> >well, but the blending is off. Here's a much smaller version of >>>>>>>>> the >>>>>>>>> >stiched image for reference: >>>>>>>>> >https://boxdog.legroom.net/public/ffmap-example.jpg >>>>>>>>> > >>>>>>>>> >Note that the left column is darker than the rest. The source >>>>>>>>> images >>>>>>>>> >aren't like that - they're uniformly blue. I suspect it's those >>>>>>>>> dark >>>>>>>>> >splotches, I guess some kind of oil or water stains, that's >>>>>>>>> throwing off >>>>>>>>> >the blending. Is there any reasonably straightforward way to tune >>>>>>>>> that to >>>>>>>>> >get the original brighter blue across the full image? Both >>>>>>>>> enblend and >>>>>>>>> >multiblend produced similar results. >>>>>>>>> >>>>>>>>> Hugin will try and optimise the brightness and colour of your >>>>>>>>> images >>>>>>>>> to match if you ask it to, but your PTO project has default values >>>>>>>>> for photometric parameters. So it looks like your photos are >>>>>>>>> different somehow. >>>>>>>>> >>>>>>>>> In the Photos tab, optimise Photometric -> Low dynamic range, >>>>>>>>> variable white balance. >>>>>>>>> >>>>>>>>> -- >>>>>>>>> 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 hugin-ptx+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/14b9270b-d71a-41cb-bf3d-ec930974f3ddn%40googlegroups.com.
