PixInsight Forum (historical)
PixInsight => Tutorials and Processing Examples => Tips & Tricks => Topic started by: Jules on 2011 June 27 10:51:54
-
Hi
Please have a look at the attached image.
I have been collecting Ha/OIII data for this image, I have used Ha for red, OIII for blue and a blend of OIII with a small amount of Ha for green. I have two questions:
1. I have applied a background neutralisation routine to this image. I do not understand the histogram. The image is predominately red, however the amplitude of the red curve is significantly lower than that of blue/green?
2. How would you apply a DBE to an image containing mostly nebulosity?
Julian
-
Add 1): The histogram is only an indication of the balance you have in your image. It does not tell you anything about the spatial distribution. What this histogram tells you is that you have much more variation in the red channel (i.e. a broader distribution of red pixel valúes). Those values may be in locations where there is no blue or green, giving you the red impression.
Add 2): Impossible. You need to have regions where there is no nebulosity. ABE may be an alternative (which removes low frequency gradients, leaving high frequency structures).
Georg
-
Hi Georg, that is interesting, and what happends if you apply DBE on separate channels which may have region whitout nebulosity.?
Hola Georg, es interesante, y qué pasa si se aplica DBE en los canales por separado los cuales es más posible que tengan regiones sin nebulosidad?
Saludos, Alejandro.
-
Hi Jules,
Georg has nailed it. The basic idea is that we cannot do miracles (well, we actually do, but don't tell anyone :)): if there are no background pixels in your image we cannot model the background.
However in the image you've posted you have many 'relatively free' background areas to generate a reasonable model of your background IMO. With 'relatively free' I mean that these areas can be considered as background to the depth you've achieved in your image, or in other terms, there are no significant structures recorded on these areas. To use DBE with this image, you shouldn't place more that 10 - 20 samples in total. Then we have the upper left corner where there are definitely no free background regions, but DBE should be able to extrapolate appropriate values. As noted, the generated background model won't be perfect but it can be reasonably good. If you have gradients to fix it will be better than nothing, anyway.
N.B.: I have to fix that broken 'Auto Clip Setup' button on HistogramTransformation in Mac OS X PI versions! :-[
and what happends if you apply DBE on separate channels which may have region whitout nebulosity.?
You can try, and sometimes that helps, but DBE works on a per-channel basis so it actually builds three separate models, one for each color channel. So in general you'll get quite similar results.
-
Hi Julian,
Just as a side remark - I wouldn't add Ha to the green channel, but add it to the blue channel as a simulated contribution of Hb. That gives a much better rendition of the true color of glowing hydrogen. You can see it during a total solar eclipse as the color of the chromosphere.
Best regards,
Gerhard
-
Gentlemen
Thank you for your comments.
1.That does make sense with respect to George's comments about a broader distribution of red pixel values.
2. Juan, I will use your methodology. The problem I have is I do not know if I have a gradient, to me it is not obvious?
3. Gerhard, I will do that.
Regards
Julian
-
Hi Julian,
Just as a side remark - I wouldn't add Ha to the green channel, but add it to the blue channel as a simulated contribution of Hb. That gives a much better rendition of the true color of glowing hydrogen. You can see it during a total solar eclipse as the color of the chromosphere.
Best regards,
Gerhard
Gerhard
Would you agree with this
Red - Ha
Blue - Ha/OIII, with 10 to 15% Ha
Green OIII
Julian
-
2. Juan, I will use your methodology. The problem I have is I do not know if I have a gradient, to me it is not obvious?
If you're not sure, you may try one of following approaches:
1) Apply brutal STF and examine image, ignoring nebulosity.
2) Try to extract background with Juan methodology (sparse samples) and examine extracted background with help of STF. Apply same STF to your image and compare it to extracted background - do you see same variations?
3) Extract just green channel and examine it with help of STF. There should be very little to none Ha information, so background gradients will be easier to spot.
cheers, Zbynek
-
Yes Julian, this is the usual configuration I use.
Best regards,
Gerhard
-
Hi
DBE can work wonders even with very few samples have a look here http://www.harrysastroshed.com/pixuser/DBEpart2.html (http://www.harrysastroshed.com/pixuser/DBEpart2.html) for a similar example
Regards Harry
-
Harry, Zbynek and Gerhard.
Thank you Gentlemen, lots to think about!
Julian