Noise in my images - Forums

Topic starter

Jul 20, 2023

Maybe somebody has or is experiencing this issue. No matter how many subs I stack into a final image ( I’m in the process of redoing M51 with 1104 subs at 14secs each, totaling over 4hrs) I’m still getting red and blue “noise” bands in my images. This is after stacking in Siril, doing my crop, then a background extraction, then I do SPCC in pixinsight. At that point I’m seeing red and blue noise bands which I’m not able to extract using EZDenoise or NoiseXTerminator. Also unable to get rid of the banding using the CanonBandingReduction script. It then is causing issues with my GHS stretching. To extract fine detail with GHS I’m getting the noise also, mostly red scattered through the image

. I’m capturing my subs with an unmodified Canon 60D. Shooting my subs between 15-30 secs each depending on whether I’m shooting at my house in a bortle 6 sky or my dark sky site which is bortle 4, no light pollution filter. I’ve been in the hobby for almost a year now. 90% of the time I’m using my 12” SCT LX200. I don’t guide, the mount is too old and not responsive enough in Dec for guiding, at least that what PHD2 reports back when I try, that’s a separate issue I’m not concerned about at this time, just do a good polar alignment and let it track and recenter the object when needed, looking to update to an ZWO AM5 soon. Before you say I need longer exposures to increase my SN ratio, I understand this, but I feel there must be a way I’m not using to minimize the noise in my images, primarily in the red and blue channel. I use various means of back ground extraction, from DBE, to GraXpert. I also tried various ways in photoshop. Removing stars and working on the starless image in camera raw and other methods I’m familiar with doing photography for over 50 years and using photoshop since the early 2000’s. Any help would be greatly appreciated, and I’ve tried to supply as much info as I can. The above image has an has an STF applied and a noise reduction using EZDenoise. I don’t have NoiseXTerminator at this time, the 30 day has expired, but I get this issue in every image and I’ve not been able to use anything to get it out, outside of clipping the blacks to get rid of it that way, which isn’t a good way.

Chris McGrew

Jul 20, 2023

Could it be some issue in your calibration frames? Maybe the bias master isn't doing its job properly.

Well Written

Anthony (Tony) Johnson

Topic starter

Jul 20, 2023

Could it be some issue in your calibration frames? Maybe the bias master isn't doing its job properly.

I guess that’s a possibility, like I said I’m new to the hobby. I shoot my bias the way I’ve seen on YouTube and read. Fastest shutter speed, which for the canon is 1/8000 and telescope covered. I’m trying to do a noise reduction on each channel separately and the recombine to see if that helps. I just get this red scattered noise through the entire almost finished image image. Could be a flat problem too, I’m trying to understand better how to shoot them. Just not sure. There are conflicting methods on taking flats.

andrea tasselli

Frequent contributor

Jul 20, 2023

This banding is typical of many Canon cameras and it is hard to tackle. I suggest you keep the processing entirely in PI and see what comes out of that using a proper normalization reference frame which has minimum banding.

Well Written Concise

Anthony (Tony) Johnson

Topic starter

Jul 20, 2023

andrea tasselli:
This banding is typical of many Canon cameras and it is hard to tackle. I suggest you keep the processing entirely in PI and see what comes out of that using a proper normalization reference frame which has minimum banding.

Not too familiar with doing that in PI. Only been using it for a few months now. Had some issues stacking in WBPP so went back to Siril and without going through cr2 to fits conversion and then editing the fits header to have a keyword that references a different night (which btw I’ve done) doing multi night in WVPP regardless of what’s been said about it with cr2 files, I’ve never had any luck. So I do sirilic and Siril for the stacking, but I’ll look into it like normalize scale gradient or such as that. Thanks though. I’ll give that some thought.

AwesomeAstro

Jul 20, 2023 · Edited: Jul 21, 2023

This is absolutely normal with DSLRs unfortunately, and isn't "noise" in the traditional "variation from the actual value" or "uncertainty in measurement" sense. Stacking more images will not fix this in an of itself, although dithering aggressively and getting more stacks can help mix it around.

~~Potentially unpopular statement: I largely fix this problem by exposing my flat frames to 95% of the histogram, rather than the normal 50%.~~ Edit: turns out I was exposing to 50% the whole time this way, the histogram in the camera isn't linear, even when set in raw mode! See posts below. I still do to this day. I did extensive testing with flats and found that this kind of garbage color signal is substantially reduced when my flats are done in this way. ISO values also influenced it, even though they "shouldn't". DSLRs are weird. Consider trying it and see if it helps, after all, what's the worst that could happen? I use flats exposed at ISO 200 (same as the lights), lasting longer than 1s each. I no longer use a dew shield while imaging as well. Both of these have helped.

Also when you start getting more total integration time, this kind of stuff tends to fall "behind" the signal and only minimal dark pixel clipping is needed to hide it completely.

Helpful Insightful Engaging Supportive

Anthony (Tony) Johnson

Topic starter

Jul 20, 2023

This is absolutely normal with DSLRs unfortunately, and isn't "noise" in the traditional "variation from the actual value" or "uncertainty in measurement" sense. Stacking more images will not fix this in an of itself, although dithering aggressively and getting more stacks can help mix it around.

Potentially unpopular statement: I largely fix this problem by exposing my flat frames to 95% of the histogram, rather than the normal 50%. I still do to this day. I did extensive testing with flats and found that this kind of garbage color signal is substantially reduced when my flats are done in this way. ISO values also influenced it, even though they "shouldn't". DSLRs are weird. Consider trying it and see if it helps, after all, what's the worst that could happen? I use flats exposed at ISO 200 (same as the lights), lasting longer than 1s each. I no longer use a dew shield while imaging as well. Both of these have helped.

Also when you start getting more total integration time, this kind of stuff tends to fall "behind" the signal and only minimal dark pixel clipping is needed to hide it completely.

I totally appreciate you input and will give your suggestions a try. I’m curious though about one of your statements, you no longer use a dew shield. I’ve been using my SCT for 27 years and have never used it visually or photographically without the dew shield. I see it as that and a light block given the fact that the corrector plate sets at the very end of the tube. I mean I use dew heaters, always have but a dew shield, not sure where that plays into it but hey I’m totally open. I will give the flats as you suggest a try. I’ve been thinking of lowering my iso to the recommended 800 but I could see going to 200. I do know the lowest iso setting is where there is no amplification to the sensor involved. Photons to pixels recommends 800 but like I said I could see doing 200. And getting the flat frames over 1 or 2 seconds. Food for thought. Thank you very much. I’ll take your suggestion under advisement and give them a try. The freakin camera takes great pics, just that noise is a real pain.

Respectful Engaging Supportive

Arun H

Frequent contributor

Jul 20, 2023 · Edited: Jul 20, 2023

I’ve been thinking of lowering my iso to the recommended 800 but I could see going to 200.

On an older Canon cameras, you really, really don't want to be lowering ISO and certainly not to 200. You ideally want to be running at 1600. The problem with the older Canon sensors was that the analog to digital conversion adds a lot of noise. Raising the ISO amplifies the signal from your sensor before it is converted to digital and hence reduces the impact of this conversion noise. Raising the ISO past 1600 on these cameras will not impact signal to noise but unnecessarily cut dynamic range. Your fundamental problem is that you are using short exposures on a camera with very bad read noise - and read noise of the worst kind, one that has a pattern. There just isn't enough random noise coming from the sky to overcome this atrociously noisy sensor.

Helpful Insightful Engaging

AwesomeAstro

Jul 20, 2023 · Edited: Jul 21, 2023

Yes to Arun's point, my recommendations do not include using the same ISO value as me. ISO values for DSLRs should be determined with two considerations.

1) You want a value high enough to cause the signal to swap noise that the camera introduces later in the image processing sequence after the ISO "gain" has been applied.

2) You want to use the lowest value above this level, in the interest of preserving dynamic range.

My D5300 is among the lowest noise cameras I've found, and according to testing performed here, I can image as low as 200 without read noise becoming proportionally worse after the ISO adjustment. Looking at the canon's data (link on the same page), it looks like your 60D shouldn't be used at ISOs lower than 800, in line with advice given above. That is, 800 is the minimum for that camera.

My point above was that changing the ISO of the flats themselves, actually caused differences for me in the effectiveness of the calibration routine, even though it "shouldn't" have such an effect. However as of late, I've found that the histogram is even more influential, and despite the typical advice, I have found time and time again, with direct and very careful and meticulous testing, that exposing the histogram 95% of the way to the right is substantially better than the typical 50% for MY setup/camera. Edit: Oops! Turns out this was 50% exposed linearly all along... See a few posts down! I love when theory meets experiment. I no longer vary the ISO of my flats, as the histogram change has been more influential. Otherwise I get weird color gradients (although you should note that they're not horizontal like yours, mine are either annular or splotchy). I'm not advising you to match my ISO values, and in fact you shouldn't, based on the link above.

This may not solve your issue, however to support the above, here is the exact same stack of data with no post-processing except background neutralization, with flats of identical ISO values with identical steps taken, ~~the only difference being the first one has the flats histogram peaking around 50% (usual advice), and the second one with the flats exposed a bit longer to around 95% of the way to the right.~~ Edit: This histogram doesn't show the linear data, even if the camera is set to raw only!! See a few posts below. Literally only 1 thing changed here: the flat exposure length. The difference is almost absurd, but I've gotten the same findings on many different nights on many different targets. I've also had splotchy color gradients remain as well (this one is annular). Ignore the satellite trails, I didn't use pixel rejection (to isolate this issue). Needless to say, ~~I haven't gone back to exposing my flats at 50% anymore, because these results are repeatable across many nights. I doubt this is universal advice: YMMV.~~

Helpful Insightful Respectful Engaging

AwesomeAstro

Jul 20, 2023 · Edited: Jul 20, 2023

So you definitely shouldn't lower your ISO, my nikon has lower processing noise. But you can definitely fiddle around with your flat exposure histogram and see if that fixed pattern color material gets calibrated out better from the flats. No guarantees unfortunately.

Regarding the dew shield, I also used it for years for blocking light pollution (I don't get dew), however years ago I realized it caused non-linear reflections when putting the flat panel over the shield (so flats couldn't help), so I always removed the shield for flats, with much better results. However, years later after a failed experiment (to reduce gradients) in which I made an absurdly long shield out of black velvet, I found the gradients even worse, by far (more vignetting was expected, gradients from the darker material were not...). Then I thought to remove my commercially-bought dew shield altogether, and to my surprise the gradients were even lower. I personally experienced no additional light pollution (YMMV if you have lamps near your imaging location- I do not), but I did experience a flatter field. I no longer use a dew shield at all (I have emergency dew heating rings if needed). I mention this to answer your question, although this almost certainly isn't your issue.

In short, don't lower your ISO value on that camera, get more integration time total (so you can get that stuff "suppressed" behind the SNR of the galaxy without clipping the blackpoint too much), consider seeing if varying your flat exposure time/histogram peak location helps calibrate it out better (these cameras can be really weird with color-dependent contributions), and don't be afraid to clip a percent or two of dark pixels anyways; whatever it takes to get results that you like, visually. I also think it's usually better to use fairly long-length flat exposure times (1s or greater), to fight noise similarly. If you get curious though, you could see if fiddling with that changes the calibration of this fixed pattern stuff... When experimenting, only change 1 thing at a time.

Chris McGrew

Jul 20, 2023

Not that you're asking, but the day I got a dedicated astro cam and stopped using a DSLR or mirrorless camera was the best day of my astrophotography life. I winged about the decision for a while but am so happy I finally did.

Arun H

Frequent contributor

Jul 20, 2023 · Edited: Jul 20, 2023

It helps to know that @AwesomeAstro is using a Nikon D5300. That is a very different (and much better) sensor than the Canon 60D. You can see this in the DR graph below (from Bill Claff's website), that the D5300 is a really good sensor. It shows high dynamic range at low ISO and each doubling of ISO shows a 1 stop reduction in Dynamic Range. This means it has really low analog to digital conversion noise - which means you can use it at ISO 200 with no penalty. The Canon 60D on the other hand shows a flat curve and lower dynamic range until ~ ISO 800 - which means you have to amplify the signal considerably to get it past the read noise floor.

The bigger problem with the 60D which the graph doesn't show is that the analog to digital conversion is done external to the sensor and different rows of pixels take different paths to the external ADC. This results in a banding pattern since the different paths have different noise characteristics. It is possible to significantly reduce this noise by using longer exposures and adding enough random noise that the banding becomes less significant, but using short exposures will certainly exacerbate the problem.

I should add: these things matter very little for day to day photography because there is enough light coming in (hence enough shot noise) that the banding is not usually an issue. But for astro, our signal is so low, and the shot noise comparatively low, that these things matter a great deal.

Well Written Helpful Insightful Engaging Supportive

AwesomeAstro

Jul 20, 2023

Arun H:
It helps to know that....

+1 to everything Arun has mentioned.

I think to myself how astonishing it is to me that they would use different electrical pathways for different parts of the sensor, but then I remember that these cameras weren't actually built with astrophotography in mind... Canon probably wouldn't have ever expected these weird issues in making that design choice, because for daytime photography (or even non-astro nighttime), the effects wouldn't have been seen or relevant...

Given this reality, longer individual subframes should be of help as well, like he is saying (not just longer overall integration time as I advised- Note that with fixed-pattern stuff, adding more and more frames can't eliminate it entirely, if it's present in all subframes), because the pattern occurs on a per-image basis; longer subs will overwhelm this pattern more on each subframe before integration, regardless of total integration time. Reading back on your original post, I see that 1) you cannot guide right now (although that new mount could allow these longer subframes!) and 2) you wanted advice aside from this. Here are some other options, although none of which are particularly ideal...

~ Use longer exposures anyways (see how long can you go without star trails), possibly throwing out more subs, but having lower noise in the ones you keep.

~ Aggressively dither and/or rotate the entire optical train throughout the night (since you're not needing a guide star, rotate that thing across the total 360deg throughout the night!) Spewing all the fixed pattern stuff around can make it more removable in the end since local variations become smaller, and pixel rejection can function better.

~ Change cameras (I know....)

Do note that any noise reduction algorithm will not handle the color patterns well, as you've found. Those are for noise on a smaller pixel scale, but yours are more along the lines of "gradients", although sadly on too low a pixel scale to be easily handled by a background/gradient extraction either. They're right in the sweet spot...

Helpful Insightful

AwesomeAstro

Jul 20, 2023

Not that you're asking, but the day I got a dedicated astro cam and stopped using a DSLR or mirrorless camera was the best day of my astrophotography life. I winged about the decision for a while but am so happy I finally did.

Can you speak more to this and the benefits you obtained? I've been considering it back and forth similarly to yourself... I just haven't been able to justify it yet given all of my light pollution...

Chris McGrew

Jul 20, 2023

Don't let light pollution be your guide. I live in downtown Houston and get great results with monochrome narrowband imaging.

my ZWO Astro cam is cooled, seamlessly integrates with the AsiAir, and produces cleaner images than any traditional camera I've ever tried. If I could go back a few years I would encourage myself to stop hesitating and just go for it.

Helpful Concise Supportive

Anthony (Tony) Johnson

Topic starter

Jul 20, 2023

I really appreciate the help that you guys having given me. Unfortunately a couple things will have to remain the same till I can get enough money to move up in the hobby, 1. Long exposures are not possible with my setup. I might be able to get 30 sec subs but without a light pollution filter my sky glow at my home site is too bad to allow longer exposures. I’ve tried and the subs get totally washed out. So there’s that. 2. I don’t have a mount good enough to guide. The LX200 from the 90’s just doesn’t have the ability to do it adequately enough, at least not that I been able to do, firstly it’s a fork mount and polar alignment can be challenging, I have a very good method that gets me close but without a polar alignment scope such as the pole master then it comes down to the drift method each and every time I set up. Time consuming even if you do get good at it. So until I get a new mount, which I’m going to do if nothing more that to lighten my load, reaching 70 years of age and my setup is a big one. And yes a dedicated camera is in the works. All ZWO so it works together with few issues hopefully. I do have an old Nikon D200, not the greatest either but I don’t get the banding with it. I used it last year when I first got started but I’ve advanced light years since then, not the best at noise either but no banding. I was under the assumption given guys on YouTube that the canon 60d was a good camera for this, and I’ve got good results and don’t feel I’m spinning my wheels, but I may have to rethink a few things. All ya all been very helpful. Thanks again for the help.

Helpful Respectful

AwesomeAstro

Jul 21, 2023 · Edited: Jul 21, 2023

I really appreciate the help that you guys having given me. Unfortunately...

I wouldn't necessarily assume that just because they look washed out doesn't mean you can't use them. For subs this short, I don't think the light pollution prevents you from increasing exposure time since half of the exposure means half the light pollution, but also half the signal. Similar effects. I use 5-min subs (must be guided of course) in a bortle 7-8, and without stacking and background extraction, sometimes the target isn't even visible in the frames! You might experiment there too, all the way to a processed image however. Of course there are specific considerations regarding how long an exposure should be to 1) overcome internal noise and 2) to hit the light-pollution point of diminishing returns, however my point is that simply looking "washed out" is not a problem in itself. It inevitably will with light pollution, so long as your raw ADU values aren't getting extreme.

One other tip came to mind, when I have aggressive background color garbage (mine are splotchy but similar!), I use a lightness mask to cover the stars and target (after stretching), and use MultiscaleLinearTransform to kill the color detail in the background. Use the "chrominance" option, choose 5 layers, and X out the 5 layers before "R". It'll kill that color.

Best of luck with your setup!

Helpful Respectful Engaging Supportive

AwesomeAstro

Jul 21, 2023

Don't let light pollution be your guide. I live in downtown Houston and get great results with monochrome narrowband imaging.

my ZWO Astro cam is cooled, seamlessly integrates with the AsiAir, and produces cleaner images than any traditional camera I've ever tried. If I could go back a few years I would encourage myself to stop hesitating and just go for it.

Your results in bortle 9 are definitely great! It might not work for me in imaging galaxies at higher focal length though... It's definitely good to see what you can obtain in those conditions though, it's definitely something for me to consider.

Die Launische Diva

Jul 21, 2023 · Edited: Jul 21, 2023

However as of late, I've found that the histogram is even more influential, and despite the typical advice, I have found time and time again, with direct and very careful and meticulous testing, that exposing the histogram 95% of the way to the right is substantially better than the typical 50% for MY setup/camera.

The "histogram peak at 50%" advice refers to the linear histogram and not the one provided at the back of the camera. Take a look at this guide. Try targeting for flat exposures of ~1sec and use the same ISO as your lights (yes, I know, in principle different ISO for flats shouldn't matter). For your flats use a light source of good uniformity and dim it as required with sheet(s) of paper.

I have seen some banding with my Canon 550D (not high-tech too) when my darks temperature differs a lot from that of my lights and when shooting lights of 15s or less. But generally it calibrates fine with the notoriously fast Samyang 135mm lens.

Helpful

AwesomeAstro

Jul 21, 2023

Die Launische Diva:
However as of late, I've found that the histogram is even more influential, and despite the typical advice, I have found time and time again, with direct and very careful and meticulous testing, that exposing the histogram 95% of the way to the right is substantially better than the typical 50% for MY setup/camera.

The "histogram peak at 50%" advice refers to the linear histogram and not the one provided at the back of the camera. Take a look at this guide. Try targeting for flat exposures of ~1sec and use the same ISO as your lights (yes, I know, in principle different ISO for flats shouldn't matter). For your flats use a light source of good uniformity and dim it as required with sheet(s) of paper.

I have seen some banding with my Canon 550D (not high-tech too) when my darks temperature differs a lot from that of my lights and when shooting lights of 15s or less. But generally it calibrates fine with the notorious fast Samyang 135mm lens.

Wow, you absolutely blew my mind with this, and I'm really glad you shared it! I was familiar with the ADU measurements, even more so with CCD cameras, but I had always assumed that the DSLR shows you the raw histogram on the back if you set it to only take raw, no jpegs, in which case one would aim for 50% exposure based on the histogram itself. I am completely wrong on that; even when set to raw mode, the DSLR still shows the histogram of the automatically-produced jpeg that it stores with the raw file (and in fact, you also only view the jpeg on the window!) and not the raw .NEF file. You can imagine my amazement when I exposed a raw file completely, for a value of ~16,400 ADU, and then looked at the flats I had already settled on, which even with the camera histogram literally around 95% of the way to the right, had an ADU value mean of 6,200! Despite still being slightly too low (which is also surprising), they ended up very close to correct! But what's crazy is I found them to be better visually by trial and error, thinking I was deviating from the normal practice. Apparently I had accidentally proved the common advice to be best, without even knowing it. What a crazy revelation, thanks again.

In short, the DSLR histogram is not linear, even when in raw mode, and can never be used at 50% to properly expose the flats. I guess I have to go to 98% on the non-linear histogram now if I need a reference in the field

I'm actually laughing at myself because going back to the flats I used in my example above that I thought were only 50% exposed, the raw values show they were only 10% exposed... That's pretty funny, no wonder they didn't work well!

Helpful Insightful Respectful Engaging

Anthony (Tony) Johnson

Topic starter

Jul 21, 2023

I use Lightroom to transfer files from my dslr while taking them, backing up to an external hard drive as well as my laptop. I do not store on the card. So if the histogram on the camera, which I’ve never used btw is not linear then what is the histogram in Lightroom showing. That’s the one I’ve been using to get my reading on my flats. And yes I’ve used the 50% rule for the histogram because that’s what I heard on YouTube from so many guys. Actually I’ve just been reading about all the other values for the flats just recently. And kinda having a difficult time understanding all of it. I use a homemade flat panel for my 12” and don’t really have a way to dim it outside or the white t shirt, to get the 1 sec exposure time. Seems as though everything I’ve heard might be wrong. Shoot darks at the same iso and time as the lights, shoot flats at the same iso adjusting expose time to get 50%on-the histo, shoot bias at the fastest shutter speed possible at the same iso as the darks, flats and lights. And don’t worry bout dark flats. Thinking I need a new recipe for my calibration frames reading you posts here, because as far as flats are concerned I think I’ve been doing it wrong. And could be part of my noise problem as well as blotchy background that takes a long time to edit out.

AwesomeAstro

Jul 21, 2023 · Edited: Jul 21, 2023

I use Lightroom to transfer files from my dslr while taking them, backing up to an external hard drive as well as my laptop. I do not store on the card. So if the histogram on the camera, which I’ve never used btw is not linear then what is the histogram in Lightroom showing. That’s the one I’ve been using to get my reading on my flats. And yes I’ve used the 50% rule for the histogram because that’s what I heard on YouTube from so many guys. Actually I’ve just been reading about all the other values for the flats just recently. And kinda having a difficult time understanding all of it. I use a homemade flat panel for my 12” and don’t really have a way to dim it outside or the white t shirt, to get the 1 sec exposure time. Seems as though everything I’ve heard might be wrong. Shoot darks at the same iso and time as the lights, shoot flats at the same iso adjusting expose time to get 50%on-the histo, shoot bias at the fastest shutter speed possible at the same iso as the darks, flats and lights. And don’t worry bout dark flats. Thinking I need a new recipe for my calibration frames reading you posts here, because as far as flats are concerned I think I’ve been doing it wrong. And could be part of my noise problem as well as blotchy background that takes a long time to edit out.

I honestly wouldn't say you have a noise problem, but rather a fixed-pattern color issue from the camera (I subscribe to the camp that only calls "noise" random deviations from a perfect measurement of what the camera actually sees- I wish we had a better term for fixed pattern "noise"/signal- but of course this is just a terminology issue and doesn't matter much in the end). The data looks good overall. If you're unsure about the histogram in LR, use the guide in the link posted a few posts earlier to see, in Pixinsight, if your flats are properly exposed (compare to a fully blown out image pointed at a light for a few seconds). I'd actually be curious to see if exposing your flats better helps correct at least some of these color issues you're having; I understand it stems from the sensor design, but I wouldn't be surprised if it could still be calibrated out at least somewhat with properly exposed flats, assuming yours were also underexposed. Not all things can be, but who knows. Definitely show us the results if you try this, I'd be curious.

It doesn't sound like you need more dimming too, if you find your flats are underexposed, then simply use a longer exposure time on them until the ADU values are correct. Definitely do continue shooting darks at the same ISO as the lights, and same for bias. I do the same for the flats now too even though this is theoretically less important. Your issues are not caused by lacking dark flats either.

Helpful Insightful Respectful Engaging

Anthony (Tony) Johnson

Topic starter

Jul 21, 2023

I use Lightroom to transfer files from my dslr while taking them, backing up to an external hard drive as well as my laptop. I do not store on the card. So if the histogram on the camera, which I’ve never used btw is not linear then what is the histogram in Lightroom showing. That’s the one I’ve been using to get my reading on my flats. And yes I’ve used the 50% rule for the histogram because that’s what I heard on YouTube from so many guys. Actually I’ve just been reading about all the other values for the flats just recently. And kinda having a difficult time understanding all of it. I use a homemade flat panel for my 12” and don’t really have a way to dim it outside or the white t shirt, to get the 1 sec exposure time. Seems as though everything I’ve heard might be wrong. Shoot darks at the same iso and time as the lights, shoot flats at the same iso adjusting expose time to get 50%on-the histo, shoot bias at the fastest shutter speed possible at the same iso as the darks, flats and lights. And don’t worry bout dark flats. Thinking I need a new recipe for my calibration frames reading you posts here, because as far as flats are concerned I think I’ve been doing it wrong. And could be part of my noise problem as well as blotchy background that takes a long time to edit out.

I honestly wouldn't say you have a noise problem, but rather a fixed-pattern color issue from the camera (I subscribe to the camp that only calls "noise" random deviations from a perfect measurement of what the camera actually sees- I wish we had a better term for fixed pattern "noise"/signal- but of course this is just a terminology issue and doesn't matter much in the end). The data looks good overall. If you're unsure about the histogram in LR, use the guide in the link posted a few posts earlier to see, in Pixinsight, if your flats are properly exposed (compare to a fully blown out image pointed at a light for a few seconds). I'd actually be curious to see if exposing your flats better helps correct at least some of these color issues you're having; I understand it stems from the sensor design, but I wouldn't be surprised if it could still be calibrated out at least somewhat with properly exposed flats, assuming yours were also underexposed. Not all things can be, but who knows. Definitely show us the results if you try this, I'd be curious.

It doesn't sound like you need more dimming too, if you find your flats are underexposed, then simply use a longer exposure time on them until the ADU values are correct. Definitely do continue shooting darks at the same ISO as the lights, and same for bias. I do the same for the flats now too even though this is theoretically less important. Your issues are not caused by lacking dark flats either.

Dude, I have check my flats via this article; https://www.myastroscience.com/proper-flats-with-dslr going off this article my flats from my M51 shoot had a mean of 2971 and a medium of 2988, I'm not sure what a fully saturated flat or over exposure frame from the Canon 60D would be, I've not check but will, but going from his article which tracks pretty close to what you are saying my flats are seriously under exposed. Thereby not even close to calibrating my lights properly.

He's saying full well on his example would be 15000 so half well and what to shoot for via his example would be 7000 and if that's close to my camera I'm truly seriously under exposed as you can see from the screen shot. I do believe now that I've had a day to think about what you are saying which lines up with the website I refer to, sounds legit. I'm going for what you and he are saying the next time out. I need to figure out how to dim my light panel so I can get those 1sec flats, and then I'll just shoot dark flats with the same ISO and timing and just use them as bias frames for the calibration. You know how sometimes, when your brain is finally ready lol, everything just starts to make sense, well this is starting to make perfect sense to me. So I'll give your advise coupled with Sergio's and take my flats to get that mean and median setting what they need to be to get half well mean and medium. Just need to figure out those numbers with a over exposed frame from my camera. Hopefully I can get out there tonight and shoot something. I'll keep you posted and your info has be invaluable. Oh BTW in going back in our thread I realize you are referencing the same article as I am. I've had that article for a few months now and was having a hard time wrapping my head around it, but I totally get it now after talking to you. Thanks again, keep you posted

Arun H

Frequent contributor

Jul 21, 2023

Anthony - you have received some great advice on flats.

Would you mind sharing your complete calibration procedure? Specifically, how/when are you taking darks? Are you using biases? Do you have the "Optimize Darks" box checked in PixInsight when you do your calibration?

Well Written Engaging

Anthony (Tony) Johnson

Topic starter

Jul 21, 2023

Arun H:
Anthony - you have received some great advice on flats.

Would you mind sharing your complete calibration procedure? Specifically, how/when are you taking darks? Are you using biases? Do you have the "Optimize Darks" box checked in PixInsight when you do your calibration?

I do everything one right after another the night of the lights. Darks are shot with same everything as the lights but with my telescope cover on, biases are shot at the fastest shutter speed the camera provides with the same ISO as the lights and the cover on, flats have been taken till now with same ISO as everything else and exposed to give a 50% histo in lightroom with my homemade light panel, I've been using 1250 ISO so that equates to 1/3200 sec, which now I'm learning is about 3200x too fast lol. As far as stacking, I use Siril. The reason for that is a simple one. I'm not familiar enough with WBPP to get the results I want. Not being able to guide my exposures I can get less that perfectly round stars. I've had WBPP discard half of my lights, in Siril there is a command you can input before your stacking begins that will "relax" having to have perfect stars. Now, and you can see from my posts, most of my stars are used, I've had 100% of my lights get used, M21 is an example of that, I sit and monitor my exposures as they are coming into Lightroom and can see if an exposure needs to be deleted due to wind or some unforeseeable problem that can arise during capture, so I guess I'm blinking while I'm shooting, not every single one, but I do glance at every single one because I have to monitor the drift so that I can stop the process and re center the object. Also getting 2 nights to work in WBPP with CR2 files had not worked for me without converting them to fits files and editing the fits header to include a session number. I've tried separate folders, session number in the file names and then using those keywords and nope not working, so if you are wondering why I don't stack in PI that's why. I hope to get to doing that, I will be upgrading my entire rig in the next year with a new mount, AM5, and then a new camera from ZWO but until then I'm in the Canon ecosystem, but honestly I have no problem with the way I'm doing my capture, just want to get that "noise" or whatever it is under control, I'm happy with my results, just want to do it to the best of what I can with what I've got. My LX200 is 27 years old and has been modified to basically be an Autostar, thanks to George Dudash in California, because my original motherboard fried and as you know Meade is out of business, so PHD2 doesn't like it and reports after I try to do a calibration that my Dec can't respond enough to guide adequately. I've stacked in PI but when you take 600 subs and the app throws out 250 of them, and Siril throws out 20 and then the result from PI so no better than Siril, I'll just go with Siril. I know it may not be the best way to get your final but it works for me, and the results show. So that's it in a nutshell up to now. I love PI and do 80% of my editing in it with the final image going to photoshop, but that stacking is another matter.

Helpful