Problems with flat calibaration of light frames. - Forums

Topic starter

Mar 5, 2022

Hello,

I am trying to get back into Astrophotography after quite some while and I am encountering problems that I did not have before :-/

I made a first test night with only a couple of shots from the city center just to try out the equipment and workflow.

After preprocessing though I have quite some ammount of unflattnes left which is incredibly visible after a quick background extraction.

Equipment:
Lacerta 72/436
EOS RP
Lightrack II

The flats were taken with a milky acrylic glass on top of the lens in room-light. An approach which I used for years before without problems.
I did not try to tweak anything yet.
I also repeated the calibration without any flats. ABE substraction vs division doesnt fix it.
What I find weird is, that the Flat seems to have this extra "dark ring" and not just "vignetting" but the dustbunny is perfectly removed by the flats.

(boosted STF for better visibility)

I would highly appreciate your advice.
I'm also happy to provide further information :-)

Clear Skies,
Robert

Well Written Respectful Engaging

andrea tasselli

Frequent contributor

Mar 5, 2022

You should consider taking dusk flats instead. Easy to do and requires no additional hardware. Having a diffusive screen close to the scope inlet pupil is often cause of wrong flattening profiles, as it is in your case. Alternatively you should get a EL panel and take the flats at night.

Helpful Concise

Linwood Ferguson

Mar 5, 2022

Several things come to mind:

First, it is rare that with a CMOS sensor you want both bias and darks for the same calibration, that causes issues on sensors with really low noise. I couldn't tell from your description if you did that. It is fine to use bias for flats, and darks for lights (it is also fine using dark flats for flats, and darks for lights). Try avoiding though all three for one calibration – this was common for CCD, even needed, but usually not for CMOS. Similarly, avoid scaling darks, if using darks match the exposure, do not scale (Pixinsight causes this dark optimization, making it sound like a good thing - usually it is not).

(I am sure there are CMOS cameras especially ones with amp glow that need bias and darks separately – but it is work experimenting if you are unsure your camera does. A LOT of documentation just automatically presumes you need all three, but most is held over from the CCD days and older technology).

Since it is color, the flat has all three channels. Look at the RAW histogram (you cannot see that in camera!) and make sure that all three channels have their peaks well away from either end, and are not clipping. It's very easy in color to have a nice looking flat but in actuality one channel gets clipped high or low.

I don't know Canon, but on my Sony when i was using it, there were settings that impacted raw that should not have: lens compensation was one, and yielded a circular patterns in darks which ended bleeding through after calibration because of their sharp edges. Try turning everything "helpful" off, it should be as much an unprocessed raw image as possible. Be very sure all long exposure noise reduction is off (actually high ISO noise also). Nothing "smart" should be engaged.

Try with and without lens hood, look inside for anything shiny that could be causing reflections. Light leaks in taking darks can also be a serious issue often not that visible in looking at them. Many lens hoods do not seal all light out where they mate with the lens, and anything shiny inside the hood can then reflect back toward the sensor.

If, for the same optics, your lighting techniques yield different looking flats – especially with varying asymmetries– it will simply not work well. You have to solve the flat problem.

While inconvenient, if you do not want to try a good flat panel, try sky flats – longish exposures untracked (so stars streak) on a brightening sky in the morning, or even early dusk. Takes some experimentation to get rid of stars, but if shooting mostly up you should be able to avoid gradients. I suggest this because it is easier to avoid an uneven light source. If you still get weird rings and especially asymmetry, there's an optics issue somewhere.

Again, I really have no idea what is causing it, but some of the above can screw up flats, so FWIW.

Helpful

AstRobert

Topic starter

Mar 14, 2022 · Edited: Mar 14, 2022

Thanks for your answers. I took your hints into consideration and spent another night testing. For processing I did only use darks on the lights and left the Flats uncalibrated for now. Also all automatic image coreections and so on were turned off and I made sky flats. Unfortunately as you can see, the problem still persists.

If you look closely you can see, that the raw as well as the flat have horizontal strips on the top and bottom of the image fields. Is that supsicious? Maybe some refletion issue?

I also dug up some older images where I used the same setup + a 2" Filter.
Left a current raw, right an old raw. Due to a reflection on the upper right side, i had in the old, i got a new flattener.

It feels like that the profile of the old and the new differ. The new one on the left seems to have "two rings".

The old one I was able to process without any issues using my standard Flat method.

Linwood Ferguson

Mar 14, 2022

Why do your sky flats have a ring inside the lighter portion? That in itself looks suspicious to me. Do you know what causes that?

Well Written Engaging

AstRobert

Topic starter

Mar 14, 2022

No, unfortunately I do not know…

AstRobert

Topic starter

Mar 15, 2022

If anyone is interested, here is a small batch of my pictures. I would be very keen on seeing, if you also encounter the problems. This way I could find out if it is a software or a equipment/capture issue:

LeoTripletSubset

Cheers Robert

Rick Veregin

Mar 15, 2022

There is definitely something odd in both your image and your flats, those interior rings are very strong and odd. In theory though, if you get your flats perfect, they should be able to correct any oddness. Note as well, if the signal strength in your image is low (too short subs, too low gain, too low total exposure) then the more you have to stretch and the more likely you will pick up problems with flat correction.

Flats will not be great with a CMOS camera unless you use a dark flat with the same exposure as the flat. A bias might do in a pinch, but biases at very short exposure in CMOS can be weird. You are using a dark with your light, that too is required–must be same exposure and temperature as your lights. These two are a must, I never got good CMOS flats until I started using dark flats and darks for the lights. So if you don't do both you are not likely to get a good result.

Make sure light is not getting in anywhere it shouldn't. If you have a dew shield make sure it is on and in the same place when you do your flats as when you did your lights.

As mentioned by others, make sure your histogram is not clipping on either end for your flats. Typically the red will be near the bottom, blue in the middle and green at the top, try to ensure the bottom of the red is not clipped, and the top of the green is not clipped. Depending on the light source for the flats, if it is too bright or not bright enough in an area of the spectrum you may have trouble not clipping at some colors. If you clip at all, your corrections can show weird rings like you are seeing.

I wasn't clear if you are using a filter or not in all your tests. I would suggest taking the filter out for some lights and flats and get that looking right. Then add the filter and confirm it is not the filter causing vignetting or multiple internal reflections. Make it is as simple as possible as you troubleshoot.

Good luck
Rick

Helpful

AstRobert

Topic starter

Mar 16, 2022

So i made some furhter experiments using old data, where i did not have these problems. The working theory was, that if I use one half of the flats to create a master flat and the second half of the flats to be my "lights", if everything was ok then the resulitng image should be perfectly flat.

Indeed that was the case, but only for the old images and not the new ones. The process of flat creation and WBPP were identical.
The upper row is the old MasterFLat and next to it the stack of calibrated flats, perfectly even.
Below the new MasterFlat and next to it a stack of calibrated flats.
(no darks involved at all)

Differences between old and new flats:

Exposure time and ISO (old: ISO 3200 1/125sec; new: ISO400 1sec),
Different FLattener
the old one I used an additional light pollution filter in.

So what does that tell me?

kuechlew

Mar 16, 2022

First of all I would support Linwood's suggestion to check for reflections. I had a similar ring in my flats a while ago, caused by exactly that. You took your old flats at 1/125s ISO 3200 which corresponds to 1/15s ISO 400. You claim your new flats are taken with 1s at ISO 400, so you're pouring 15x the amount of light into your system, assuming the same equipment (you mention you only exchanged the flattener). This certainly makes the amount and impact of reflections more likely.

Your latest test indicates an inconsistency in your flats. I would suggest to look at the individual images of your flats whether there are some outliers which you may want to remove before creating the master flat. Proper flats should pass your test as long as you only apply the flats to the "lights" if I'm not mistaken.

Good luck and clear skies
Wolfgang

Helpful Respectful Supportive