Building mosaic - getting panels to same levels

[cmassa]

I'm having trouble building a 4 panel mosaic of the Veil Nebula because the background lightness varies between the panels. When put together with gradientmergemosaic, the panel differences still show. What is a good way to balance the panels so they are even? I am assembling the mosaic after cropping, ABE or DBE, background neutralization, color calibration and before doing any stretching. Images are shot with a Baader IR  modified Canon T2i.

Thanks for any advice.
Chris

[pfile]

i usually just define previews over areas of background and check the mean values, then scale one of them by the ratio of the means. seems to work ok.

rob

[cmassa]

Hmm,

That does make sense, I just don't know how to do that. What process is used?

Chris

[pfile]

first - the images should still be in linear form

so i just make previews on each image, then use the Statistics process to look at each preview in turn. then using pixel math, i apply the following expression to the one with the dimmer background:

$T*(mean_brighter/mean_dimmer)

that should scale the dimmer one it by some value > 1.

in fact you could probably come up with a pixel math expression that just directly calculated the mean values of the previews, probably with Mean(brighter->preview1) or Median(brighter->preview1) etc. need to try this to verify that it works.

rob

[mads0100]

One thing I found when my mosaic was doing this was that I had a small 1-2 pixel black line down where it was trying to merge the mosaic.  It was playing havoc with GradientMergeMosaic's process and really messed it up.  I didn't notice it until I really zoomed in.

[pfile]

that's a good point - any strong discontinuities (like a bright star, or a black border) can mess up GMM's gradient computation…

rob

[cmassa]

I image at 20-25% overlap and make sure to crop the borders.

Chris

[georg.viehoever]

If you share images, I can have a look.

Generally, GMM will not get rid of differences in texture (such as noise "patterns") or differences in scale (contrast). The corrections GMM does are purely additive. To arrive at a similar scale, try LinearFit.

Georg

[cmassa]

That's what I am having trouble with is differences in contrast. I tried to use the linearfit tool but can't seem to figure it out. The one time I got it to work, one image came out black, no data left. I've searched online and can't seem to find how to use it and there is no documentation in PI.

Chris

[georg.viehoever]

Chris,
see http://pixinsight.com/forum/index.php?topic=2302.msg15204#msg15204 .
Georg

[cmassa]

Thanks. The example there uses two single frames. Will this also work with stacked images? If so, should this be done right after stacking? Or should some background extraction be done first. I tried using images that were slightly different size (cropped to remove stacking artifacts) but it gave me errors regarding geometry differences and stopped.

Chris

[georg.viehoever]

The images for LinearFit (and GMM) need to be aligned to each other (and thus have identical dimensions). I usually do it (LinearFit and GMM) after stack+DBE, before any non-linear operation. But I know that others have gone different paths.
Georg