Histograms

[+Stu]

Hi folks. The last couple of sessions I've had have appeared to be fruitless. They were both hampered by technical problems but I managed to get a few subs, but I think they're over exposed or something. One session the moon was quite bright so I wasn't expecting too much. 

I'm sure I read somewhere that when you open the stacked image in processing software, the histogram should spike somewhere in the first 1/3 of the window, or the image isn't much use. The first was in the last 1/4 and the second was just over halfway. Both looked like a mess when I adjusted the RBG black levels. Am I over exposing? (If that's the right term)

[Bottletopburly]

I doubt your over exposing the moon may ruin things washing Dso away but  image away from the moon and maybe try star clusters rather than galaxies or nebula, post your set up , camera , iso etc number of subs , did you take calibration frames?.

[Clive]

I set the histogram peaks in the final image as far to the left as possible without clipping the data to the LHS of the peak or 10% of the maximum image intensity, which ever is the greatest.  When you pull in the stacked image the peaks can be virtually anywhere. The further to the right they are just means that the dynamic range of the image is reduced resulting in more stars becoming saturated (have a flat top rather than a nice peak). The width of the histogram peak indicates the level of the background noise and a bright moon or light pollution will tend to increase the width.  A wider histogram peak indicates more background noise which will reduce the faintest detail you will be able to see in the image.

Edited November 12, 2022 by Clive
Typo

[+Stu]

I'm not getting many subs due to clouds, trying to make things work etc. So probably only 10x 3min, ISO1600 with calibration frames.

 

150PDS

700D mod

IDAS LPS P3 filter

 

I was going to drop down to 2 minute subs and then just need a decent session to get enough of them. Hopefully tonight so don't want another "wasted" session. 

1 hour ago, Bottletopburly said:

I doubt your over exposing the moon may ruin things washing Dso away but  image away from the moon and maybe try star clusters rather than galaxies or nebula, post your set up , camera , iso etc number of subs , did you take calibration frames?.

 

[+Stu]

4 minutes ago, Clive said:

4 minutes ago, Clive said:

A wider histogram peak indicates more background noise

 

I think they were relatively wide. I'll put some screenshots up. 

[Clive]

Do you stack using DSS? If so what background level does it indicate. I try to aim for a maximum of 10% but have used images with up to 15% when I'm convinced the high background level is not caused by light cloud.

[+Stu]

52 minutes ago, Clive said:

Do you stack using DSS? If so what background level does it indicate. I try to aim for a maximum of 10% but have used images with up to 15% when I'm convinced the high background level is not caused by light cloud.

I do, although it's not something I've been checking. There could have been some cloud in the mix. 

 

 

 

[Clive]

The histograms peaks do look very wide. Probably due to a combination of moonlight, light pollution and a low number of subs in the stack.

As I've found out, Affinity doesn't appears to show the peaks of the histograms so it's difficult to set them all to the same ADU.

[+Stu]

6 minutes ago, Clive said:

As I've found out, Affinity doesn't appears to show the peaks of the histograms so it's difficult to set them all to the same ADU

I'm not sure what you mean by ADU. Is that something that can be done in DSS before saving the file?

 

 

 

[+Stu]

I also noticed that I did my flats at ISO800 and not 1600, could that be affecting it?

[Clive]

9 minutes ago, Stu said:

I'm not sure what you mean by ADU. 

 

Analogue to digital units. Your Canon 700 has a 14 bit sensor so will have an ADU range of 0 to 16383 (2^14 -1). I believe DSS scales the images to an equivalent ADU range of 16 bits (0 to 65535) by effectively multiplying the captured data by 4.  For some reason, a lot of image processing software assumes an ADU range of just 8 bits (0 to 255, presumably because JPEG images are only 8 bit). So that 16 bit data suppled to Affinity by DSS is converted to 8 bit data by dividing by 265 before displaying the histograms?

 

An incorrect flat ISO would not affect the position of the histogram peaks. Flats are normally taken at a different exposure length than the lights so why not a different ISO?

[+Stu]

I noticed you said 16bit, I've been saving as a 32bit (saw it on an Affinity tutorial). I've saved one as a 16 and got a much better result. 

[Clive]

I believe DSS scales the RAW images to 16 bit no matter what the initial image resolution (12 bit, 14 bit etc). When it does its stacking business it uses floating point maths (probably 64 bit?) to maintain accuracy so after the stacking process it offers the opportunity to save the result as a 32 bit floating point image or 16 bit integer image. To be honest, I have yet to be convinced that if you start with 14 bit resolution images, you gain much benefit my saving them as 32 bit floating point stacked images if in the end you save the processed image as an 8 bit JPEG image.

[+Stu]

I checked background levels and the highest was <7%. At least I can be a bit more confident going into tonight that I'm not falling at the first hurdle with bad data. 

Thanks for the help as always 😁

[+Stu]

Just been doing some reading while the scope does it's thing and I might've been using too many dark frames.

[Clive]

I didn't think you could use too many darks. A huge quantity won't give you any additional benefit but surely won't do any harm.