Help with CCD flats

[traveller]

Been experimenting taking CCD flats and have been very frustrated.
My old image of NGC 2070 without flats clearly shows gradient (photo 1), but when I applied my flats, it made it worse (photo 2)
When I examined my flats, I think I found the issue, only the first flat had a pixel value of 10.5k average (photo 3), but all other flats (I took 35 of them), showed pixel values of 31.3k (photo 4) average, even though the parameters were the same!
So the questions I have are:
1. Is a flat with average pixel value of 10.5k acceptable? (I am using QHY8, using white t shirt over computer screen).
2. Why do the values change for all other flats even though I took them with same settings?
3. What else am I doing wrong?
Thanks
Bo

[Atmos]

If all of the flats bar one are around the 30k mark then that one outliner shouldn't have much of an effect. Technically you can do flats with 10.5k ADU value but you need to take a lot more of them to get the same quality.

If it was the first one that had the issue, it could have just been a random issue that caused it. No idea what but maybe a mistimed exposure or something.

I wouldn't mind seeing a properly stretched one (running a DDP) to see the full range. Looking at the shape (not round) I have a feeling that you may not be getting even illumination. This will cause your flats to do more harm than good. Over correcting in some parts and under correcting in others.

[rustigsmed]

maybe have a crack at twilight flats, there is less to worry about then, just the exposure length rather than t-shirt and laptop as well.