Acceptable death pixels?

3 replies166 views
Thomas avatar
Hi altogether,

I need a second opinion regarding a death pixel circumstance. I just recognized that my stacked lights (new ZWO ASI 2600 MC) contain some strange pixel pattern, which look like caused by death pixel.
Is this normal and in a range, which is acceptable? I mean, it looks like a complete pixel block out of order...

Kind regards
Thomas

dkamen avatar
Hi,

This is a hot pixel (one that goes at 100%). A dead pixel is one that stays at 0%. Collectively they are called bad pixels.

I would say up to a few hundred in your whole image is completely imperceptible unless you are looking at 100% magnification so  you don't have to do anything about it, especially if you intend to downsample your final image for presentation.  Up to 1% (several tens of thousands) can be dealt with modest dithering between subs. There are many reasons why it is a good idea to dither a little (say every 10% of your subs), bad pixels are one of them.

Between 1% and 5% needs either more aggressive dithering a procedure where hot pixels are replaced by the average value of their neighboors, this is extremely efficient and the only "price" is a very small penalty in overall sharpness. PixInsight and Astropixel processor support this procedure by applying a bad pixel map, which is a map of the locations of your bad pixels created using darks and flats. Rawtherapee supports automatic detection of hot/dead pixels in each sub whiile developing the raw file, and it also supports BPM for the difficult cases.

Overall, nothing to worry about unless you have a very small sensor and a very narrow FOV. This image has more than one hundred thousand hot and dead pixels on each sub and about half of them made it to the integration despite the crazy amount of dithering. See if you can spot them

Cheers,
Dimitris
Helpful Engaging Supportive
Thomas avatar
Hi Dimitris,

thank you very much for your helpful answer!
I was thinking that it's a "dead pixel", because I read somewhere that a dead pixel appears in different colors then adjacent pixels (because of bayer filter) and stays permanently; and hot pixels would rather appear like small crosses when zoomed in; actually the bad pixel shown above, is a pixel block of at least nine pixels.
I used dithering for the subs, which produced the bad pixel pattern on the integrated image (image with bad pixel pattern), but in this case I didn't used darks (which I think would perhaps prevent the pixels from creating the pattern). Initially I was thinking that the dithering is responsible for spreading the bad pixel and producing this pattern ... but that makes no sense, or?
Checked your image: Did not found any bad pixel!  

CS
Thomas
Helpful Insightful Respectful Engaging
Thomas avatar
Minor add-on: The subs do not contain the bad pixel pattern shown on the left most image. The subs only contain the bad pixel cluster shown right.
Well Written Concise