Thanks for the info. I wondered if a combination mode could be made, which corrects the colours using the Poisson equation (something for me to learn about - got any links? Is it a frequency domain thing?), but still uses a hard seam to connect the images without trying to blend any features (which blend mode appears to do, as I get smears very similar to Enblend/Multiblend when I use deliberately misaligned images).
I've been updating Muliblend and while it is faster than both Verdandi and Enblend, I haven't yet come up with a reliable way to avoid overblending (Enblend uses different numbers of levels for every image, as the overlap allows, but as well raising the possibility of inconsistent blending this also seems to result in later images being blended more than earlier ones). On Friday, 22 January 2021 at 15:44:05 UTC T. Modes wrote: > Hi Monkey, > > verdandi has 2 blending modes: > > - hard seam: this is the fastest one and relies to 100 % on Hugin > photometric optimization. For me this works fine in about 3/4 of the cases. > If there are bigger color/exposure differences at the image edges between > the images then I use blend seam (or enblend) > - blend seam: it uses the same seam finder as in the hard seam case. The > remaining differences between the images are blended with a Poisson > equation solver - something completely different from enblends pyramids > approach. This algorithm can also blend away some higher color shifts > between images. But it still benefits from Hugin photometric optimization - > especially when the photometric optimizer has corrected the vignetting > (color and exposure is not so important in this case). > > But I got some reports where verdandi blend produced bad results. This > happens mostly when using bigger images. This issues are not there when > reducing the output size. So it is very difficult to debug in the case. > Fixes for this are always welcome ;-) > > Because of this issue and for backward compatibility I have hesitated to > change the default. But each user can change the default blender in the > preferences. > > PS: > > Hugin may have done in balance the exposure of control points > > The control points are only used for geometric optimization. They have no > effect for the photometric optimizer. Here only the current arrangement of > the images is used for probing the photometric differences between the > 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 [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/107d740f-084f-4a24-ba26-b808c73f48d4n%40googlegroups.com.
