Can't find why my flats are not even and change as I rotate my flat panel

[ollypenrice]

8 hours ago, newbie alert said:

No idea 💡 as just read the text, can't remember seeing the images , didn't realise any was there🙈

They are in the opening post.

We do see rotation but what is rotating looks unlike a flat to me.

Olly

[aleixandrus]

Thank you for all you comments, they really help. I didn't have good weather yet and next days seem even worse. I run son 'garage light flats' instead of 'sky flats' and the uneven flat pattern shows similar than before. It is true it is still LED lighting so I need to run sky flats anyway.

What I also found is a very strange grid pattern in Ha flats, which only shows with very-very strong stretching and strong flat illumination. None of that shows in the lights and I didn't find any of this when processing data. Even ZWO suggest this is normal with the IMX183 as it is a sensor-related issue, but it calibrates well. Oh, keep in mind this is not related with debayering as it is a mono sensor.

I read some data in forums that indicates this unevenness in the field may be 'normal' in some sensors, specially the first with BSI technology, such as the IMX183 (ASI183) or IMX492 (ASI294). This is not a brand-related issue but a sensor-related issue. I still have not a definitive clue on what it is happening but it seems feasible that the combination of sensor and glass window coating and filter glass coatings and band-pass size *may* be interfering in an strange way. I saw similar cases with OSC cameras and dual/quad band filters, where a large 'X' pattern shows even in the light subs. What people suggest with this 'lemon' sensors or equipment combinations is to take flats in the very same conditions than the lights (gain, offset, temperature...), with a proper illumination (30-40% histogram peak) and large exposures (>3s very minimum, even 50-60s) and stack >30 flats with no optimization at all. They also suggest use a even illuminated flat pattern, both in intensity and color hue as these sensors are quite sensitive. Contrary to what I thought, an iPad seems not to be a good flat panel, although my sensor is mono, so not sure if this applies. Finally, they also suggest "don't judge flats for their appearance but for master light final result".

 

This is what I've being doing since day one, so I don't know, maybe everything is going the way it is supposed to go. In any case, the flat unevenness and the varying flat illumination doesn't seem related issues, which is what worries me. I can understand the first, but not the last. Maybe I should just keep shooting narrowband and process nebulae to taste and forget all of this if the images keep going well to my eyes. I still want to know what is happening as this is usually my way of doing things (fully going down the rabbit hole) so I'll post an update if I get more evidences.

Edited December 26, 2023 by aleixandrus

[Elp]

If your flats are calibrating fine then there's no issue.

I haven't personally read anything about flats issues with the 183, I've been using it for 3 years and also an uncooled one for DSO work and both work fine. Both my samyang lenses work similarly to each other also. The 294 however is a completely different beast and does cause strange issues in relation to flats dark bias signal and DBE, due to this sensor is why I invested into perspex sheets to darken my panel and take much longer flats, I haven't had much issue since and the panels I use are cheaply sourced.

[aleixandrus]

Well... I 'reopen' this thread as I'm taking a second try to solve this flat issue...

I remove the whole optical train and try from flats, piece by piece. I've finally illuminate the naked sensor (just the camera resting over the flat panel) and what I found is that the illumination varies as the camera rotates (forget about the horizontal stripes, I was using a very low exposure length as the sensor is completely unfiltered and I can't lower more the illumination). Please, check how corner illumination varies and the gif above.

Is this an expected result, that the camera orientation affect illumination evenness? Has this ever happened to you? Maybe I'm doing something wrong but I am out of ideas right now...

[900SL]

Are you using some kind of polarised light source?

[aleixandrus]

I tested different light sources. One, an old LCD tablet, it is almost for sure. Other, a LED flat panel, I supposed it is not but to be fair, I didn't check it.

[ollypenrice]

So it looks as if the camera itself is very slightly vignetted. I've never tried this experiment so I don't know how my cameras would behave. To be honest, I don't see it mattering because the variation is minor and  this vignetting will find its way into the final flats (shot through the optics) anyway.

By the way, you posted some ABE-processed images earlier. What we see in them is a common ABE fault which regularly affects galaxy images. ABE puts background sky markers too close to the galaxy (or roundish nebula in your case) and treats faint outer glow as background sky. When applied, it lowers the pixel values close to the object and creates a dark region around it. I only use ABE when there is really no workable way to identify background sky anywhere in the image, as often happens in very deep stacks.

Olly

[ONIKKINEN]

Both cameras i have used for flats have a slight sensor vignette (EOS550D and IMX571), so dont find that too strange.

[aleixandrus]

Sorry for the late reply. Thank you for your insights. I suppose it is what it is, including sensor vignetting, flat panel unevenness and (probably) some user error. I'll skip this 'issue' for a while unless it becomes a problem during processing. As I said, thank you so much!

[TiffsAndAstro]

On 20/12/2023 at 09:29, aleixandrus said:

I tried with the flat panel and an old 10" tablet. I'll try with a new tablet (iPad) as soon as I can borrow it and do some sky flats. I can understand the flat panel is wrong... but that tablet? It has no sense to me.

I stretch an old white t-shirt over few shield and but that up to an old pixel c tablet. Seems to work well in Nina's flat wizard, especially when I lowered the brightest when Nina told me off.

Is your panel really diffuse enough? It suggest edge lit LEDs from only one edge or a big difference between them. Defusing that light more might help reduce that gradient.

[TiffsAndAstro]

On 22/03/2024 at 13:51, ONIKKINEN said:

Both cameras i have used for flats have a slight sensor vignette (EOS550D and IMX571), so dont find that too strange.

Vignette would be  direction/orientation independent though?