Ideal Steps for DSLR Frame-Flickering Issues

[cameronleger]

I'm experiencing a quirk and I'm not sure if it's with cameras, DSLRs, or my Sony A-99 in particular. I thought it was with my flats, but now I think it affects my lights too; perhaps it's an issue with the sky!

The issue manifests itself nearly at the end of my processing, so it's quite a pain to experiment with combinations of processing. I'm hoping that anyone here will know what I'm talking about and/or have some tips to eliminate it.

For my setup, I try to aim for around 30-50s exposures which usually means that I need an ISO somewhere from 800-1600 and a somewhat open aperture; full-open would be 5.6 and I can usually get 8 or 10. At 300mm, the lens has a fair amount of vignetting which is the main source of this problem, I believe. Fully open, it's basically a gradient starting at the center of the frame, but where I'm at there's a nice flat circle almost constrained by the vertical region of the image, and it fades off starting from there.

I made a small flat-box, but I've also tried sky-flats with the same issue. The main source of this issue is that not all of my frames come out with the same illumination. If I 'blink' the flats I notice a flicker as most of the frames vary slightly in their brightness, and a handful of them get quite a bit brighter or darker. I thought this might be the fact that they're shot around 1/400s and slight variations of the shutter might show this, but it happens on 30s light frames as well! This is a problem because I can see that the vignetted portions of the frames do not change that much, but this central circle varies more intensely.

So, I pre-process the flats and lights and there's not much to show here. You can slightly see the pattern in the rejection maps as there's a central circle that's rather consistent and more/less rejections with the gradient outwards. Since I usually have 200+ frames I'm using the Linear Fit rejections. I've tried Local Normalization by my best/average frame and it slightly changes the issue but it's still there.

With a fully-integrated light frame, the issue is very slightly noticeable. There's a mostly flat illumination profile, but there's a darker ring where the vignetting would start. After the background extraction the issue is fully pronounced. At this point I have to be super careful with the extraction process to try to get that ring included and make it smoother. The corners are fine, the center is fine, but where the center starts to vignette outward there's a ring where there were, I guess, inconsistent rejections.

Has anyone seen or overcome this? I have a few ideas to try with rejections and normalizations, but since the issue is not really noticeable until the post-processing it takes a lot of time and space to experiment. I've attached a recent example of a post-background-extraction where this issue is most present.

[dld]

Hello Cameron,

Maybe your flats exposure is too short? You may have a shutter shadowing effect i.e., impartial/uneven illumination of your flat frame due to the shutter movement. Dim your flat light source and increase the exposure time. I use a thick paper in front of a laptop monitor which I randomly move between capturing flats in order to avoid recording any inhomogeneities of the paper. I also lock the mirror up just in case...  8)

 

[cameronleger]

I will keep that in mind, but I might have to compromise to make it happen. I don't normally use darks on the flats since they're so quick, and maybe it's time to drop the ISO all the way down for them and expose them closer to the white-limit instead of just above the noise floor. I found at ISO 100 some pre-processing can make the results worse!

I tried to find one light frame that showed this symptom after pre-processing, and then did some experiments with that one frame in order to speed up the testing. I tried high/low rejection values for the flat integration between 0.5 and 5.0 sigma units; while the flats and their rejections came out differently, the effect persisted almost unchanged.

I just tried Local Normalization on the flats and integrated again with default settings (except for LN), and this is the best result on that one frame so far. It might take a few times to find which flat is the 'best' to LN as a reference, but I don't have many other ideas at this point.

[pfile]

you have to use darks, or bias on the flats or else they will not correct the lights properly. understood that the sony sensors have very minimal dark signal but they still have bias/offset that needs to be removed...

rob

[cameronleger]

I definitely do the bias on every frame; I have a recent and very large integration for each of my ISOs from 100 - 3200, skipping the ones that are not 'standard' like 120.

It's the darks on the flats that I'm thinking of skipping. Since the dark is 45s and the flat is 1/400s, even the 'optimize' doesn't know what to do with it. I could do darks at 1/400s also, but I was under the impression that it's rarely done since the contribution is negligible.

For example, once I tried moon shots at the night so very quick shutter times were involved. Even after adding pedestals which was necessary, the frames were definitely worse when using darks.

[pfile]

OK - since the flats are doing funny things i just wanted to confirm that they had at least been bias subtracted.

you are right that there's negligible dark signal in the flats, so it's probably OK to skip the darks.

1/400 is pretty short for flats and i'd second dld's advice about trying to get the shutter open longer. on my STT-8300M i never do flats shorter than 1s, even though it has an "even illumination" shutter.

this video is of a canon DSLR, and the longest exposure they did was 1/500s, but you can see that at 1/500 the 2nd curtain is chasing the first, so it's at the point where the shutter stops opening and closing completely and starts becoming a "moving slit"...

https://youtu.be/CmjeCchGRQo?t=2m12s

for a sky flat short exposures might be OK but with a panel, there's going to be some amount of flickering and you want to make sure the shutter is open long enough to average out several cycles of flicker.

rob

[cameronleger]

Glad to know I'm probably on the right track. I have a small flat-box and I'll try to put in some more paper or something to dim my LEDs even more. I thought I was finding it hard to get it perpendicular to my lens to eliminate any vertical/horizontal gradients, but I notice that I still had the same issues with my sky flats. Either it's very hard to get it 'perfectly flat' (haha...), or perhaps the lens introduces it by some misalignment. In any case, I didn't have significantly better results either way so I'll go back to my box.