[Dan Paris]

Hi,
I find that my K-1 Mark II has a lot of ampglow, maybe more than it should be.

Ampglow manifests itself as a magenta glow at the bottom of the image when the exposure of an underexposed image is pushed during RAW developpement.


On my copy, at ISO 800 for instance, it starts becoming annoying at only +2EV so it affects no only astro pictures but also ordinary ones.

Here is a dark frame (ISO 800, 20 sec) pushed +3 EV with +30% contrast/brightness :





Here is what it does to the image in the same parameters (800 ISO +3EV push). This time the exposure is only 3 sec:





there's a workaround (substracting dark frames) but it is not very convenient and the noise it not removed.

I don't see this in the DPREVIEW review, see Pentax K-1 II Review: A worthy upgrade?: Digital Photography Review, ISO-invariance test

It may be that my copy is defective, and it would be very helpful for me if someone could test his K-1 II in similar conditions (cap on, ISO 800, +3EV push, raise luminosity/contrast) to check whether I should send mine for repair (still under warranty fortunately)

Dan

[MossyRocks]

That seems like it is showing up awfully soon. Amp glow is part of the read noise so shouldn't be affected by exposure length and I decided a while back to do a deep dive into with my K-3 and there I found that it was just barely detectable at about 4 stops below the shot noise at ISO 1600 and started to become noticable at 5 stops. This was found by stacking and averaging piles of bias frames to drive down the shot noise from lens cap images. Here is the thread over in the astro group where a couple of us wandered off into the esoteric world of systematic errors and true random noise. I would say given what you are seeing you have a defective sensor since with a single shot at ISO 800 even pushed should 3 stops the shot noise should absolutly still overwhelm the amp glow as you have a newer sensor, lower ISO by 1 stop, and I was finding the amp glow barely noticable 4 stops below shot noise at ISO 1600.

[Dan Paris]

Thanks a lot for your message, which confirms what I suspect


Actually before I got the K-1 Mark II I had a K-3 Mark II which does not show this issue. So it is either :

- a problem with the sensor on the K-1
- a specific problem on the K-1 Mark II (related to the accelerator unit ?)
- a specific problem on my copy of the K-1 Mark II

It may be the later but I want to be sure before asking for a repair...

[photoptimist]

Interesting issue.... I'm curious about whether usage patterns affect the amp glow.

Starting each test with the camera having been off for perhaps 1 hour at room temperature, take one of the following test shots:

Test 1. Turn on the camera on and immediately fire an ISO 800, 20 sec shot.

Tests 2a-2e, Turn on the camera, use LV for 5, 10, 20, 40, or 80 seconds and then fire an ISO 800, 20 sec shot.

My hypothesis is that the amp-glow would be different in each of these tests and likely worse for LV usage runs.

[Ivan]

I do not have K-1. But there are four other cameras. I take it as reject.Do not wait, take advantage of the guarantee.

[MossyRocks]

My hypothesis is that the amp-glow would be different in each of these tests and likely worse for LV usage runs.

There I think you would be looking at the effects of dark current which is different than amp glow.

[biz-engineer]

Amp-glow issue on my K-1 Mk II ?

I have a K1 and K1 II. Something I noticed is that my K1 II in LV mode shows a lot more noisy frame than my K1, but I guess it is related to how the firmware manages LV display, not much incidence on actual photos taken. Also , the firmware of the K1 II is not the same as the K1, they not only changed the part of firmware for the accelerator chip, there are other changes in K1II firmware that aren't obvious at first glance, I find the changes logical but not useful in any way. Now if I consider the image frames you posted here.. I'd say, my frames would never be as bad as yours at ISO800 +3 stops. We can see that you have significantly more noise at the bottom of the frame, which isn't normal. And the other thing I want to say is that I would never use ISO800 +3 stops, I'd used ISO6400 with 0 stop of ev compensation instead.

[photoptimist]

There I think you would be looking at the effects of dark current which is different than amp glow.

But isn't amp-glow just a form of dark-current?

Amp-glow seems to be caused by non-uniform heating of the sensor caused by power-consuming (and heat generating) activities in part of the chip (e.g., the amp circuits) or in the camera (e.g., maybe the accelerator). The hotter parts of the sensor substrate have a higher dark current. It's not a literal glowing light source. At the pixel electron level, all of the hotter pixels have an anomalously higher readings but the fact that the red and blue channels have more gain (to compensate for their lower light sensitivity) causes the anomalous values to look magenta.

P.S. Testing "amp-glow" in grip-up and grip-down portrait orientation as well as right-side-up landscape and upside-down landscape orientation could help determine if the problem is on the sensor or is a more general thermal problem inside the camera.

[MossyRocks]

My understanding was that amp glow is basically something that comes about during readout from the signal amp before the AtoD conversion. Thus it is exposure time independent unlike traditional dark current that comes about from heating and current leakage on the sensor during extended exposure times. In my experience what most would call dark current shows up looking kind of like a grid pattern on long exposure images. Granted some sensors have a corner that is either hotter or not cooled as well as the others where there is a bit more of a reddish orange glow but that always seemed to be camera model specific. You are right in that the cause of what would be called dark current and amp glow are the same, heat and current leakage, but they happen in different parts of the system with varying effects. It would make sense that using live view before taking a show would increase amp glow in the shot since the amp would be warm from the live view signal but I would guess that regular dark current would dominate. I'm not sure and now want to maybe play with that some, again because I seem to think I have infinite time.

[stevebrot]

Ampglow manifests itself as a magenta glow at the bottom of the image when the exposure of an underexposed image is pushed during RAW developpement.

What are you using for RAW development? The schemes used for "pushing" vary by vendor.


Steve

[jbinpg]

Turning on Dark Frame Subtraction during your tests should help determine if the glow disappears. It should. If it doesn't, your sensor->raw image pipeline is wonky and you should get it repaired. Making master bias and dark frames for astro work to remove this glow is critical but the stretching is much more extreme than what you need for daylight shooting. Here is what a stretched, uncalibrated masterdark (no bias subtraction) looks like from my K-1 (at -15 C). The amp glow pattern at the bottom of the frame resembles yours.

[TedH42]

Just ran a test on my K-1ii. Ran a series of about 7 exposures with lens cap on and taped, viewfinder capped, inside a black bag, with LCD on Outdoor Exposure +2, no noise reductions, for a series of ISO 800, 20 second exposures in quick succession. Followed with same setup, but set to LV, for about 7 exposures of 20 seconds with 20 seconds in between the exposures. So more or less following the suggestion by photoptimist. A total of 14 exposures. Used a conversion from DNG to TIF, no image modifications, then edited with Photoshop CC2019. Set Image > Exposure to +3.00, then Brightness/Contrast to +30/+30. I had no visible "amp glow" throughout this test.

Going back to the original unmodified DNG files, I opened the last image in Adobe Camera Raw (ACR, which has essentially the same image modifications as Lightroom), and set Exposure to +3.00, Contrast to +30, Highlights to +30, Whites to +30. No amp glow visible.
Going back to the original unmodified DNG files, I opened the last image in ACR, and set Exposure to +4.00, Contrast to +30, Highlights to +30, Whites to +30. Barely visible amp glow.
Going back to the original unmodified DNG files, I opened the last image in ACR, and set Exposure to +5.00, Contrast to +30, Highlights to +30, Whites to +30. Amp glow visible, about like yours.

Still using ACR, several images in the sequence were similarly checked. The first image had a very, very slight amount of amp glow at the +3.00 Exposure value. The amp glow then actually decreased to not visible from the beginning to end of the sequence at the +3.00 Exposure setting. ??

Clearly, the processing software makes a huge difference. I suspect that Photoshop's Exposure control is not in units of EV, whereas ACR's Exposure is (certainly acts that way IMO). I also can clearly see that ACR's Highlights slider is rather weak, while the Whites slider is immensely strong, with an Exposure setting ranging from -5.00 to +5.00 making no difference against a Whites setting of +100 (image is a light, grainy pink, and really ramps up after Whites +75). I also noticed that increasing any (Photoshop or ACR) Contrast setting actually reduced the showing of amp glow, while decreasing it increased the amp glow visibility.

Based on this testing, I'd suggest that your K-1ii is performing about 1.5 to 2EV less than mine and could be a candidate for service.

Oh, do the Dark Frame Subtraction setting as suggested above. If it fixes the noise, then your K-1ii is operating better than what my test seems to show. My testing was, of course, with that setting OFF.

[TedH42]

Turning on Dark Frame Subtraction during your tests should help determine if the glow disappears. It should. If it doesn't, your sensor->raw image pipeline is wonky and you should get it repaired. Making master bias and dark frames for astro work to remove this glow is critical but the stretching is much more extreme than what you need for daylight shooting. Here is what a stretched, uncalibrated masterdark (no bias subtraction) looks like from my K-1 (at -15 C). The amp glow pattern at the bottom of the frame resembles yours.

For a single long exposure image, usually the camera's Slow Shutter Speed NR (Noise Reduction) setting is set to ON in order to have the camera take the subject exposure, then immediately take the dark exposure that the camera uses to subtract from (cancel out) the camera's dark noise. As the dark exposure is thus at essentially the same temperature as the subject exposure, the noise cancellation is near optimal. This takes twice the total exposure time, of course.

In astrophotography, when many Light frames (subject images) are stacked to improve the final image's signal to noise ratio, the camera's Slow Shutter Speed NR (Noise Reduction) setting is set to OFF (also the High-ISO NR), and a Master Dark frame (multiple stacked dark frames) is subtracted from each of the multiple Light frames before stacking them together. The Light frames are taken one after another with no time used for any dark exposures. The Dark frames are frequently taken right after the total set of Light frames, while the astro-imager is doing other things (sleeping?). This is the more efficient use of the limited dark sky time, as Dark frames require only that the light be blocked from entering the lens/telescope and camera, so nighttime is not required. (abbreviated description)

My normal rule for a given subject:
If taking a single long exposure, the Slow Shutter Speed NR is ON, even for astro images.

If taking many long exposures, the Slow Shutter Speed NR is OFF, and separately-created Dark frames are used for noise reduction.

If taking some long exposures, depends on how many.

[Dan Paris]

Thank you all for your insights.

I'd say, my frames would never be as bad as yours at ISO800 +3 stops.

Compared to ISO800 + 3 stops, shooting at ISO6400 gives you 3 stops less dynamic range. Since the Sony sensor is supposed to be (almost) ISO-invariant, in high contrast situations (moonscapes, concerts,...) I believe that, would the amp-glow be absent, ISO800 + 3 stops push is a better strategy.







I did some more inquiry and check with K-1 and K-1 II files from friends. Here are my findings so far :

- it does occur both on K-1 and K-1 II so it has nothing to do with the accelerator unit


- the K-3 II that I had is much more noisy but does not show the problem. Here is a picture with the K-3 II with similar kind of exposure push :


Lot of noise but no glow


- My latest find : amp-glow varies a lot on my K-1 Mark II from shot to shot :



Crops from two pictures taken 1 minute appart in manual mode with the exact same settings, and using the exact same RAW conversion settings in Rawtherapee (copy and paste). Between the two images I did not take any picture and the liveview was turned off so it excludes sensor overheating issues. How could you explain that ???

[Dan Paris]

Here is a new piece of data. Another night shot with more extreme push, +4.5 EV this time, and no trace of amp glow. Such amount of underexposure was necessary to keep details on the moon disk.




I used pixel shift on this one which is why this picture is very clean but no glow on any of the sub-frames.

So it appears that the amp-glow is really shot-dependent which is really puzzling. In which circumstances could I trust my camera for this kind of low-light work ???