BackgroundNeutralization: upper limit not working as expected

[Nocturnal]

Hi,

I'm trying to make my BN sessions more accurate by setting the upper limit properly, based on actual pixel data. I noticed that my faint stars really are quite faint and easily fall under the default 0.1 upper limit value. It appears that my background is all below 0.0015 so that is what I selected. When BN is applied the image hardly changes. The green color remains. I would use the default 0.1 which gives better results but the resulting image can does get colorcalibrated correctly and the background is most certainly not grey.

How do I do this correctly?

[Nocturnal]

I should add that when I proceeded to CC and used the same ROI the background reference mask selected almost all pixels and the result was not good. This led me to go back to BN and find the correct limit for that process first. Of course BN does not have a mask output so it's a bit harder to debug what's going on.

[Harry page]

Hi

Does it not say green is 0.0019 

I would set at 0.002

Harry

[Nocturnal]

right but I thought it looked at the L value.

[Harry page]

Hi

Why would it look at the lum value when balancing the RGB 

Harry

[Nocturnal]

Because you can specify only a single cutoff value. RGB has three values. For BN cutoff it makes sense to look at the brightness of the pixel, not at individual color components. Well, it makes sense to me anyway and reading the documentation didn't change how I thought it would work. What does it do when G is greater than the cutoff but R and B are not? Ignore the whole pixel? Ignore just G?

If it's working as intended the documentation should probably get a slight update to address this part. A warning that a 0.1 cutoff is much much much too high is probably appropriate as well.

[Ignacio]

FWIW, I always use the highest value of the RGB channels, rounded upwards, and never had a problem. With a prior adjustment of the black point (without clipping data) I usually end up with numbers in the range 0.005-0.01.

Ignacio

[Nocturnal]

FWIW, I always use the highest value of the RGB channels, rounded upwards, and never had a problem.

When you do that doesn't it use all the pixels in the image? That doesn't seem to be how BN should work. Maybe I misunderstood?

[Ignacio]

I do this within a preview of what I consider to be a good portion of clean background. When the image is flooded with nebulosity, I use several small previews of the background (sometime using histogram equalization to detect the best ones), and then use the preview aggregation utility. This is fed into the background neutralization process.

Hope it helps,
Ignacio

[Nocturnal]

The point is if your preview has only background and no stars at all then you can safely set the upper limit to '1' without bothering to find the maximum first. If your preview has stars then the maximum value will include stars or nebulosity and is therefore not the correct value. In that case it is necessary to determine a reasonable maximum value that excludes most stars.

[Ignacio]

The preview has stars, plenty of them, but the choice of upper limit is based on background levels on each channel. I use the mouse to explore and look for the highest value, as sometimes there is some gradient to account for.

Ignacio

[topboxman]

Hi Sander,

Have you tried bumping the Upper Limit setting until the image begins to change? Sometimes guesswork works better than measuring. 

Peter

[Nocturnal]

Hi Peter,

till now I'd mostly used the default parameters and it worked reasonably well. With my current image I did not like the results so I figured I'd spend a bit more time on it. I now used measured values and fiddled with the other CC parameters as well. I think the result is better.

I still think the BN limit should apply to the L value of the pixel.

[topboxman]

Hi Sander,

I just started imaging with mono camera and NB filters. Because of very weak signals with NB, I find it works very well by lowering the Upper Limit setting of BN. I have used default values in BN in the past with SXVR-M25C OSC camera. Now with mono camera and different filters is a whole different ball game.

I just finished processing Bubble Nebula using Ha, RGB and I had a hard time. When I saw this thread, I started experimenting by lowering the Upper Limit of BN, it worked very well. Thanks for creating this thread.

This is my HaRGB result of Bubble Nebula:

http://peternagy.smugmug.com/Telescopes/Nebulae/15823498_3vBdTQ#!i=2099824883&k=cjWhMxF

The next image is bi-color of Ha-Oiii-Oiii mapped to RGB.

I still have a long way to go to understanding processing images.

Peter

[Nocturnal]

That looks really great Peter, congrats. I have the ambition to get a mono camera with filters but man that gets expensive really quickly. Many interests competing for the same dollars. I'm glad this thread was useful to you. I learned a lot myself.