[Gimbal]

Third averaging exposures is not related anyhow to reducing noise levels

It is, that is why you stack images in astro photography for instance. Random noise is averaged out and only true signals that are present in enough frames remains. But if the noise isn’t random it to will remain.

[MJKoski]

Average multiexposure in-camera (K-1) drops noise levels. Easy to see, those ones which have not been plagued by magenta-issue take considerably more punishment in post compared to single exposures. Same thing happens with PS-exposures. 150x 1sec ISO100 exposures have been ultra-clean. That cottage crop is result of 10x 30sec which is still better in IQ than single 30sec frame.

Loaner unit is available later today to pick up. It should be a brand new batch of cameras made in August.

[beholder3]

Random noise is averaged out and only true signals that are present in enough frames remains. But if the noise isn’t random it to will remain.

Yes, that is what I explained above. And long exposure sensor heat up does not create random noise. Stacking without DFS just removes random noise and keeps the heat induced static noise pretty much exactly as it is. The hot pixels do stay where they are on each shot. The average of ten shots which all show 255 or 16000 on one pixel is what? Zero?
Only you will see the heat noise much better because everything around it has been flattened by stacking but they remain shining as they did before.
That is why the Astro pros use dark frame subtraction.

It is extremely easy to understand if you think about what you are doing.

[Gimbal]

Yes, that is what I explained above. And long exposure sensor heat up does not create random noise. Stacking without DFS just removes random noise and keeps the heat induced static noise pretty much exactly as it is. The hot pixels do stay where they are on each shot. The average of ten shots which all show 255 or 16000 on one pixel is what? Zero?
Only you will see the heat noise much better because everything around it has been flattened by stacking but they remain shining as they did before.
That is why the Astro pros use dark frame subtraction.

It is extremely easy to understand if you think about what you are doing.

Ii is, the “new” thing for Pentax users is this new noise that is some form of semi stuck pixels that are not 255 or 0 (if we pretend the picture is 8 bit) they are somewhere in-between. I say semi-stuck as they are also heat dependent, or so it seems, and thus not always there.

And subtracting dark frames are NOT the super solution you seem to think. A hot pixel at 255 which are subtracted by a darkframe with the same value of 255 results in 0.

So the bright pixel is instead now a black pixel. That is fine if the surrounding pixels also are black, but that is not always the case. In camera dark frame subtraction patches this hole, but in camera dfs isn’t always practical.

[MJKoski]

I use multiexposure with d-fa 15-30 to get long exposure effects like smooth water and cloud trails. ND solutions are not exactly fun to use with it. Adding in-camera LENR screws up those use cases. I have tried to extract dark frame in post which does not remove all white dots. They are somewhat random.

[banep]

It is, that is why you stack images in astro photography for instance. Random noise is averaged out and only true signals that are present in enough frames remains. But if the noise isn’t random it to will remain.

Sorry my mistake, mixed average with additive.
Yes average multi-exposure cancels noise without any additional in-camera noise reductions turned on.

[JBox]

FastRawViewer displays both a converted RAW image (uses dcraw source via LibRaw) and and embedded JPEGs. Be sure you are viewing the RAW conversion and not the in-camera JPEG.

I double checked and I was viewing the raw conversion. The embedded JPEG shows the noise as white, like it does on the camera preview.

[banep]

Average multiexposure in-camera (K-1) drops noise levels.

Yes sorry mixed average with additive.

[MJKoski]

Okay...This is rather interesting

Here, the unit which left to service, 122 seconds, ISO3200, all NR disabled in-camera: http://mjkoski.1g.fi/kuvat/Miscellaneous/trash/_IMG1151.jpg/_full.jpg
And now, brand new loaner unit: 120 seconds (yeaaaah sorry about those 2 seconds), ISO3200: http://mjkoski.1g.fi/kuvat/Miscellaneous/trash/_IMG0001.jpg/_full.jpg

What the...there you go. No question my old unit was a DUD.

[Gimbal]

Huge difference.
However in the exif there is tag named Exposure2012 (according to exiftoolGUI) that is +2.0 on the "old" picture and 0.0 on the new. I'm not sure what that means, but first guess is of course that the old picture is pushed two steps and the new one isn't. Maybe something to look into.

[stevebrot]

What the...there you go. No question my old unit was a DUD.

That is a relief to know.

Now the larger question might be whether there are a significant number of cameras with similar problems.


Steve

[stevebrot]

However in the exif there is tag named Exposure2012 (according to exiftoolGUI) that is +2.0 on the "old" picture and 0.0 on the new.

That is XMP from Lightroom.


Steve

[MJKoski]

Old picture was 122 seconds, new one 120 seconds. They were exported with zeroed settings from LR. Just checked it again. How come that exif can contain something like that? I used +2 EV compensation with Leitax converted Zeiss before when testing things out, is it writing that info somewhere in the file?

--

There is something else as well. This is the old unit @ ISO 25600, 30 secs, f/8, pushed +3EV in LR to show dark bar on top of the image:



New unit, same settings:



Very weird indeed. Another finnish user also has that same dark area on top of his dark frames.

[awscreo]

Mm, not sure if I'm doing everything correctly, but here are my samples. Basically just a cap on test - 80s at 3200 iso, exposure boosted to +1 in lightroom (at base exposure it was too dark, my mistake for not running it longer), with and without NR, color noise reduction at 0.
I mean, I can see that without NR there's definitely more, but it's definitely way cleaner than what I saw posted here. I purchased my camera few days ago, seller claims they got this particular batch August 26th.

100% jpeg crops in attachment.

Link to original RAWs and full size jpegs:
Update your browser to use Google Drive - Drive Help

[stevebrot]

How come that exif can contain something like that?

Not EXIF, per se. By default, Lightroom packages the process metadata (essentially the same as the catalog entries) into the file as XMP. Exposure 2012 indicates exposure slider adjustments using the 2012 (most recent) protocol.


Steve