Aligning images (2)

I have been making some progress on image adjustment, I am now able to estimate the distortion parameter and to correctly position the images:

The main problem I got was that I was trying to estimate the distortion parameters with two different set, one for the first image, and one for the second. The goal was to be the most general as possible, in case some one wants to create an image with pictures taken with different lens (or with a different focal distance), but it’s definitively not the most common use case, so it probably don’t fit in Krita’s use case for panorama creation.

The difference of color is caused because I use a naive blending algorithm, I will need to have a look at enblend to see how to improve that.

This entry was posted in Krita, Open Source and tagged , , , . Bookmark the permalink.

4 Responses to Aligning images (2)

  1. Ants Aasma says:

    You should probably also check out < HREF="http://smartblend.panotools.info/" REL="nofollow">Smartblend<>. The algorithms themselves are surprisingly simple for the results that they give.

  2. Matthias says:

    Sorry if it sounds stupid, but why not just use (and maybe improve/contribute to) libpano and enblend instead of reinventing (worse and slower) algorithms?With libpano and enblend all the basics are covered and you could concentrate on a super perfect never seen before GUI.

  3. Samuel S. Weber says:

    That smartblend is pretty cool.

  4. Cyrille Berger says:

    Hum yes, I forgot about smartblend, I wonder how it behaves against eblend 3.0 (which include deghosting), the comparison linked from smartblend’s website are with an older version.@matthias, I didn’t use libpano for two reasons, one there is no documentation (no example of use, and not of the API, so I usually don’t fight with a library unless I believe it will be much much faster to use that library instead of reinventing the wheel), the second is I like to discover new stuff, new algorithm, and I strongly believe you need to write an implementation before you can consider you know how the algorithm works. As for worse, well the result of positioning and lens distortion are of the level of libpano, and as for speed, optimization alone (which is the only things I could scavenge from libpano) is a very fast operation, I didn’t find a difference between the two.In fact, my reference is http://www.cs.ubc.ca/~mbrown/panorama/panorama.html, but unfortunately they don’t give source and they charge for the software.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>