Is this an over exposure issue or my lack of processing skills?

[sspear]

Hi all,

I'm a noob on PI but I think I've learned a bit from tutorials and experimentation.  I processed a Beehive capture I did and realizing that I'm having issues with the stars, here I am.

I typically do a masked stretch as the first foray into nonlinear land. This typically works great. I saw typically as due to a variety of circumstances, I've had limited time with PI. But, I have processed tutorial and challenge data and that data *always* seem to work. My last two images with my Canon 7D (nonmodded) and AT72ED have led to stars that are plateaus of white, mostly blown out cores.

The star image reference below is a blue star, 38 Cancri, that is blue-white.

#1 - This is how it looks after integration (98, 120sec lights, drizzle.) The imaging train is AT72ED - Zhummel Sky filter (not sure of which one and it's raining) - Hotech SCA flattener - Canon 7D.

#2 - After ABE.

#3 - After BN.

#4 - After CC.

#5 - After SCNR green.

#6 - STF

#7 - I split the channels, registered, aligned, combined. I see the star pattern looks like a tilt exists in the system somewhere.

My question, does this look "normal"? Perhaps it is coming from the sky filter or my lack of processing skill. It just seems weird and it is hard to control the stars.

Maybe time for a true APO.

[jkmorse]

A couple of questions.  Is your scope achromatic or apochromatic.  If the former, color issues are normal due to failures to focus colors at the same point.  Second, where on the image is the star in question, dead center or off in a corner.  That sometimes has an impact due to field rotation, though it looks relatively minor.

As to the blown out cores, that just goes to interation time and the well capacity of your camera.  If you are looking to avoid blowing out the cores, try shorter subs and also try using histogram transform so you can watch what is happening with the histogram as you stretch the image.  Also, see what the stars look like with STF turned off.  If you are seeing the stars without any stretching, those are blown out cores.

Best,

Jim

[sspear]

The scope is a good achromatic one. This star is near dead center. I can see the bright stars before stretching so that answers my over exposed question. 2 minutes subs at 400iso, argh.

Thanks for the help!

[jkmorse]

Not all is lost.  Depending on what you are shooting, that may be just fine.  Actually you often see the brightest stars showing color only on the edges.  That does not mean that you are necessarily over exposed.  Rarely are you able to avoid any fully exposed cores.  Especially in something like the Beehive, those stars are just too bright.  If you want to avoid any oiver exposure however, more subs with shorter iterations are the way to go.

[Greg Schwimer]

This is a guess, but it may be in the right direction...

Taking the the RGB values of your image above at face value the star is probably not overexposed. However, I think it is. I've seen the same with my DSLR. It has to do with the bit depth of the camera. The red is probably at max for the bit depth, resulting in a pinking star core.

I think Vicent showed me a trick at the PI class last May but I don't remember the details completely. I believe the idea was to clip the whites down such that the red was no longer at max before any processing at all. Maybe this is just below the red value you are reading for the core of the brightest stars. I'm not completely sure. When you do that the pink cores should look more white. Give it a try and see if it works.

I don't know if this will help you recover the natural color of the cores though.

[sspear]

Taking the the RGB values of your image above at face value the star is probably not overexposed. However, I think it is. I've seen the same with my DSLR. It has to do with the bit depth of the camera. The red is probably at max for the bit depth, resulting in a pinking star core.

Interestingly, the stacked images are very red with the Canon 7D. My previous Olympus camera stacks came out near black with no discernable color tint in the linear image. I will look around to the clipping tip.

I did find this and will give it a go tomorrow. It looks promising.
http://www.pixinsight.com.ar/en/info/processing-examples/28/maskedstretch-stars-sores.html

[sspear]

Not all is lost.  Depending on what you are shooting, that may be just fine.  Actually you often see the brightest stars showing color only on the edges.  That does not mean that you are necessarily over exposed.  Rarely are you able to avoid any fully exposed cores.  Especially in something like the Beehive, those stars are just too bright.  If you want to avoid any oiver exposure however, more subs with shorter iterations are the way to go.

In the past, I used MaskedStretch with good success which has helped to dilute star core issues. I've also used layers with TheProgramThatShallNotBeNamed in order to correct the issue. I realized it is a problem and I'd like to do all my processing in PixInsight and so I've been trying to tackle this.

I reprocessed the Beehive. The original was with layers and the new without. I will still be looking for improvement as my Pleiades integration has some bright star cores!

http://www.astrobin.com/275011/B/

[Greg Schwimer]

Funny - I'm processing some data on M103. Turns out I overexposed the blue channel so I have blue cores pre-STF, and when stretched they have the typical hard ring core.

I returned to the pre-stretched state and used HT to expand the dynamic range. I adjusted "high" and applied several times until the blown cores (visible when not stretched) almost lost in the dark of the pre-stretched image. Masked stretch away - cores look good.

[sspear]

How do you adjust "high" with HT?

[sspear]

I'm guessing high range. I'll try that later. Thanks!

[vicent_peris]

This is a guess, but it may be in the right direction...

Taking the the RGB values of your image above at face value the star is probably not overexposed. However, I think it is. I've seen the same with my DSLR. It has to do with the bit depth of the camera. The red is probably at max for the bit depth, resulting in a pinking star core.

I think Vicent showed me a trick at the PI class last May but I don't remember the details completely. I believe the idea was to clip the whites down such that the red was no longer at max before any processing at all. Maybe this is just below the red value you are reading for the core of the brightest stars. I'm not completely sure. When you do that the pink cores should look more white. Give it a try and see if it works.

I don't know if this will help you recover the natural color of the cores though.

Hi,

Yes. In this camera, the saturation point is higher in the red channel, so it shows those saturated areas as red. You have two options:

- Adjust the highlight clipping separately in the three RGB channels. DON'T touch anything else, specially the midtones.
- Adjust the highlight clipping in the joint RGB histogram so you clip to the level of the color channel with a lower saturation point. This will clip some of the stars in the channels where the saturation point is higher, but this makes sense because color is meaningless when any of the color channels are saturated.


Best regards,
Vicent.

[Greg Schwimer]

This is a guess, but it may be in the right direction...

Taking the the RGB values of your image above at face value the star is probably not overexposed. However, I think it is. I've seen the same with my DSLR. It has to do with the bit depth of the camera. The red is probably at max for the bit depth, resulting in a pinking star core.

I think Vicent showed me a trick at the PI class last May but I don't remember the details completely. I believe the idea was to clip the whites down such that the red was no longer at max before any processing at all. Maybe this is just below the red value you are reading for the core of the brightest stars. I'm not completely sure. When you do that the pink cores should look more white. Give it a try and see if it works.

I don't know if this will help you recover the natural color of the cores though.

Hi,

Yes. In this camera, the saturation point is higher in the red channel, so it shows those saturated areas as red. You have two options:

- Adjust the highlight clipping separately in the three RGB channels. DON'T touch anything else, specially the midtones.
- Adjust the highlight clipping in the joint RGB histogram so you clip to the level of the color channel with a lower saturation point. This will clip some of the stars in the channels where the saturation point is higher, but this makes sense because color is meaningless when any of the color channels are saturated.


Best regards,
Vicent.

Thanks Vicent! So I was close...

With option #2 you present, if RED saturates at a pixel value of 0.8, clipping the highlight/whites to just below 0.8 should do the trick, correct?

For option #1, how would you determine the correct clipping point for each channel?

[vicent_peris]

Thanks Vicent! So I was close...

With option #2 you present, if RED saturates at a pixel value of 0.8, clipping the highlight/whites to just below 0.8 should do the trick, correct?

For option #1, how would you determine the correct clipping point for each channel?

Hi,

Sorry, I made a mistake. I think the best way is to simply clip the image channel by channel. That way you better preserve the full dynamic range of the camera. So the good option, in my opinion, is option #1. The clipping point for each color channel is found simply by reading out the image over the saturated areas.

Best regards,
V.

[sspear]

I attempted these without success today. I don't know enough PI yet.

I did have some success in other areas of PI. But, still, a long way to go! http://www.astrobin.com/273190/B/