Help With Gradient Analysis

Giddy

Giddy

Active member
Hi. I'm totally new (<2 months) and I started using PixInsight this week. I'm using a Nikon Z7 II and this was a Tamron 150-600 at 505mm F6.8. I'm shooting from Bortle 4-5 skies. Not sure if any of that helps, but it seems those are common questions.

After stacking 180 lights (60s each, ISO800), 100 bias (.000125s ISO800), and 40 flats (ASIAIR Plus Auto Flats ISO800) and running SPCC, a screen stretch revealed a mid-image change in the brightness from left to right. A straight line down the center. I wasn't able to remove it using any of the background extraction methods I know (ABE, GradientCorrection). I'm not sure what can cause that, so that is my first question. Is this an issue with my data or my processing? Is there anything I can do to recover the image?

My second question is regarding the circular gradient which is very noticeable in the lower left and which ABE and GradientCorrection don't seem to touch much. I've noticed this circular gradient or noise pattern (not sure which) in all my images so far (only about 10 targets shot). I've tried using darks and this was without darks. The pattern is still there. I can usually process some out, but I end up cropping a lot to exclude the outer 1/3 of the image.

I've tried several different background extraction methods with nothing really helping. Any advice would be apprecated. Is this bad data, a mistake on my part during stacking, or is this the best I can hope for with my current equipment?

split_gradient.png
 
Giddy said:
Hi. I'm totally new (<2 months) and I started using PixInsight this week. I'm using a Nikon Z7 II and this was a Tamron 150-600 at 505mm F6.8. I'm shooting from Bortle 4-5 skies. Not sure if any of that helps, but it seems those are common questions.

After stacking 180 lights (60s each, ISO800), 100 bias (.000125s ISO800), and 40 flats (ASIAIR Plus Auto Flats ISO800) and running SPCC, a screen stretch revealed a mid-image change in the brightness from left to right. A straight line down the center. I wasn't able to remove it using any of the background extraction methods I know (ABE, GradientCorrection). I'm not sure what can cause that, so that is my first question. Is this an issue with my data or my processing? Is there anything I can do to recover the image?

My second question is regarding the circular gradient which is very noticeable in the lower left and which ABE and GradientCorrection don't seem to touch much. I've noticed this circular gradient or noise pattern (not sure which) in all my images so far (only about 10 targets shot). I've tried using darks and this was without darks. The pattern is still there. I can usually process some out, but I end up cropping a lot to exclude the outer 1/3 of the image.

I've tried several different background extraction methods with nothing really helping. Any advice would be apprecated. Is this bad data, a mistake on my part during stacking, or is this the best I can hope for with my current equipment?

View attachment 23020
Click to expand...
Can't really see anything in a JPEG. You need to post your XISF master light (unprocessed) to a cloud service and give us a link to it.
 
Thanks, here it is:

Stacked, Unprocessed

I'm sorry if the file is too big. I'm not sure what options to pick or if I should strip things out when sharing. Please let me know if there is a protocol to sharing these correctly.

P.S. I love your website, Chris! Very clean design.
 
Last edited:
Giddy said:
Thanks, here it is:

Stacked, Unprocessed

I'm sorry if the file is too big. I'm not sure what options to pick or if I should strip things out when sharing. Please let me know if there is a protocol to sharing these correctly.

P.S. I love your website, Chris! Very clean design.
Click to expand...
The gradient across the image looks like typical sky gradient, and corrects nicely. I don't think I see the circular gradient you're referring to. Is the central line common to all your images, or just this one? What does your calibrated flat look like? There have been problems with flats generated automatically by ASIAIR being significantly underexposed.
 
I apologize... I might be using the wrong terminology. Would it be possible for you to share with me what you're seeing?

I have tried to annotate in this JPEG where I see the circular pattern shift... am I seeing things? Where the red arrows end, I see the background become darker, while the area in the arrows is "noisier" or perhaps redder? This is just with SPCC. There is sort of a dark donut and then it starts again towards the middle.

circular_gradient_arrows.png


I will research your other question about if the line is common across images. I don't recall seeing it all during my Blink check but I wasn't looking for it. I will also check the calibrated flat and report back.

Is there a way for me to check within PixInsight that my flats were properly exposed? I was worried I was trusting ASIAIR too much, but I didn't know how to get a good exposure manually in the app... the interface is confusing to me in that regard. I would have just captured using an intervalometer and the exposure meter, but I was told in another thread that mixing data acquisition methods can also cause stacking issues. I guess I could get the exposure on camera and then use that in the app.
 
Last edited:
I figured out the gradient! I needed to lower the scale. The shifts I was seeing are all but gone. This software is amazing! Totally worth the investment.

Unfortunately, I still have a noticeable line in the middle, even at Scale = 1.0. My external drive is being used for swap space by PixInsight on my laptop for another stack so I can't get to my flats or lights right now. I'm stacking this same data with the darks to compare results.
 
Giddy said:
I figured out the gradient! I needed to lower the scale. The shifts I was seeing are all but gone. This software is amazing! Totally worth the investment.

Unfortunately, I still have a noticeable line in the middle, even at Scale = 1.0. My external drive is being used for swap space by PixInsight on my laptop for another stack so I can't get to my flats or lights right now. I'm stacking this same data with the darks to compare results.
Click to expand...
Although that faint structure is correctable with GradientCorrection, it is probably a remnant that should have been corrected by your flat (that is, not an actual sky gradient).
 
Thanks so much for your help.

I stacked another set of data. I'm starting to think I'm either taking flats or bias frames wrong. I've uploaded my master flat, bias, and light frames. It looks horrible, but when I look at each master individually they look ok to my untrained eye. I was not able to get rid of the bright center area with any background removal tools.

Can you tell what is happening and which step I'm screwing up?

Master Bias
Master Flat
Master Light

This is a screen stretch after SPCC:
Screenshot 2024-05-05 092127.png
 
Giddy said:
I did not and I didn't notice. Thanks! I will fix that and restack.
Click to expand...
ISO information is only visible in the subs. It isn't propagated to either the limited FITS header or the full XISF metadata. That seems like a problem.
 
Giddy said:
Got it. Do my flats look properly exposed to you?
Click to expand...
Pretty good. Could have a little more exposure, but probably that wouldn't make much practical difference. One thing I don't like is the short exposure time. What is your light source? It's usually a good idea for a flat to be exposed for a second or two in order to avoid aliasing effects involving sensor readout, shutter movement, light source flicker, and other factors that can crop up with very short exposures.
 
I'm using a lit tracing tablet. I place a pure white cleaning cloth over the lens hood and then put the tracing tablet on top. I have just been using the brightest setting. Would it be better to start with the dimmest setting?
 
Giddy said:
I'm using a lit tracing tablet. I place a pure white cleaning cloth over the lens hood and then put the tracing tablet on top. I have just been using the brightest setting. Would it be better to start with the dimmest setting?
Click to expand...
I'd go as dim as possible, and as low ISO as possible.
 
Giddy said:
Would it be better to start with the dimmest setting?
Click to expand...
That is difficult to say. It may be better to cut the level down with a few sheets of white paper.
The problem is we don't know how your pad reduces brightness. You may think it is just "winding down" the LED backlight brightness, but it is more likely using something called PWM (Pulse Width Modulation), where it switches the light on and off very quickly, controlling the ratio of "on" to "off" to set a particular brightness. This can interfere with how your sensor readout works - so dimming down with a few sheets of paper may be safer.
 
cloudbait said:
I'd go as dim as possible, and as low ISO as possible.
Click to expand...
Just to make sure I understand that last part... I will also need a set of bias frames at that ISO as well, right?

Is the process to stack the flats and biases at the same ISO to create a master flat first? Or if I include biases for the flats and for the lights, will WBPP sort it out automatically by matching up the ISOs?

fredvanner said:
That is difficult to say. It may be better to cut the level down with a few sheets of white paper.
Click to expand...
Thanks for the tip! Interesting about pulse width modulation. I'll see if I can find out how my tablet works. I can always permanently tape a layer of paper over the front.

I
 
Giddy said:
Just to make sure I understand that last part... I will also need a set of bias frames at that ISO as well, right?

Is the process to stack the flats and biases at the same ISO to create a master flat first? Or if I include biases for the flats and for the lights, will WBPP sort it out automatically by matching up the ISOs?


Thanks for the tip! Interesting about pulse width modulation. I'll see if I can find out how my tablet works. I can always permanently tape a layer of paper over the front.

I
Click to expand...
It is almost certainly modulating the lights by PWM (fixed frequency, variable duty cycle). You may be able to see this by simply waving the tablet around and noting the pulsation. This is a primary reason to try and get your exposure time up on the order of a second, because then each pixel will average out to the same (large) number of cycles.
 
You must log in or register to reply here.
Back
Top