[MJKoski]

It was set to average. I did nothing but multiexposures that night and actually messed up the other way around - started star trail capture with average blending mode in the end

Anyways my unit leaves to service tomorrow. I will keep you posted what happens.

[johnc]

It was set to average. I did nothing but multiexposures that night and actually messed up the other way around - started star trail capture with average blending mode in the end

Anyways my unit leaves to service tomorrow. I will keep you posted what happens.

Been there, done that as well!

Just double checking mine (lenscap on) - and tbh on initial test I am not totally satisfied with the level of chroma noise on longer exposures (resulting in white speckles thanks to LR colour NR). While these only appear with significant (unusual!) use of the clarity or highlights sliders, the noise characteristic here is very different to K-3 at ISO 100. I'm pretty sure (on my body) that this observation is not field-relevant - significant (excessive) boosts to highlights and clarity are needed to reveal them.

I would say there is less general 'baseline' chroma noise than on my K-3 but the dots that there are there are more intense and prominent on file manipulation - and this is apparent on any file manipulation. Few more checks today - but I am going to keep an eye on the situation.

John

[beholder3]

30 seconds is just on the edge where the problems start. Or, when you do any minor adjustments to the file in a converter like LR. Adding minor shadow boost or anything else like clarity explodes the file.
This is 10x30sec frame multiexposure @ ISO100 using in-camera average blending. Averaging 10 frames of ISO100 shots at 30 seconds should eradicate all noise on a modern camera body. Heck, the file should be clean with only one frame.

Well. You should think again.

a) Clarity
That is meant to boost microcontrast. That is the worst you can do to anything with noise. Boost clarity and noise goes up.
What is worse in this case is the fact that microcontrast boosting does automatically turn somewhat brighter pixel into glaring bright pixels, the ones you call "white".
So good way to create an issue where there was none.

b) Stacking images from warmed up sensors
How does the oldfashioned dark frame subtraction work? Night shooters do know that long exposures heat up the sensor in a very reproducable way. Some hot pixels will show up and very specific places. Because those places do not change at all, you can take a dark frame and subtract it. It will not kill any other detail, because it has the exact same hot pixels at the exact same locations.
Now imagine what happens if you stack this scenario. All other random noise because of higher ISO will be averaged away. But the long exposure heating noise doesnt go away. It will remain in full strength, because those pixels will shine in each of those shots.
That is actually the best way to exaggerate the heat noise and make it appear as an issue, while it is none.

So bravo to consolidate all ways to get to a maximum of long exposure noise:

1. warm sensor
2. very long exposures
3. high ISO
4. sharpening
5. clarity boosting
6. stacking
7. avoiding darkframe subtraction prior to stacking

The only thing I would add to maximise noise is to chose a raw converter software which is worse than others at removing the available noise. Usually LR is a good bet here, as it is really not very capable in noise reduction. Rawtherapee would have reduced our valuable noise a lot more.


I don't think there is any better noise way to maximise the effects and visibility of heat noise. You seem to love it.

[awscreo]

I will try to upload few LE raws today. I did some quick testing with cap on, and with no noise reduction I did notice few white pixels, although not a whole bunch. With NR on they were not present. So far I've shot few high iso shots and haven't noticed white pixels, just normal noise at iso 3200 and 6400
(those were not LE shots, just normal low light stuff)

[johnc]

Well. You should think again.

So bravo to consolidate all ways to get to a maximum of long exposure noise:

1. warm sensor
2. very long exposures
3. high ISO
4. sharpening
5. clarity boosting
6. stacking
7. avoiding darkframe subtraction prior to stacking

Er ..... except that 100ISO is not 'high ISO'.

There is no doubt that this effect shows only under specific usage scenarios, that can be worked around. And LR's clarity is an absolute killer, for the reasons you have explained.

But are you really OK that there is more intense chroma noise on some K-1's at 100 ISO, 30 seconds than the same shot with a K-3? That doesn't seem right to me, and I would like to understand why, if only to get the best work-arounds.

John

[beholder3]

Er ..... except that 100ISO is not 'high ISO'.

Yes, the 100 ISO is not related to the last shot. The previous shots in this thread are at a lot higher ISO. That is what I was referring to.

But are you really OK that there is more intense chroma noise on some K-1's at 100 ISO, 30 seconds than the same shot with a K-3? That doesn't seem right to me, and I would like to understand why, if only to get the best work-arounds.

If they were shot at the exact same conditions and if they are viewed at the same size, the should be about same. If you compare at 100% view, the K-1 should show x36/24 as much chroma noise. But that assumes the sensor heat-up is the same, which can be a totally different one due to construction.
As already noted: We currently can't see the actual sensor temperature, like we can on a K-3.

[MJKoski]

What the ****?

Averaging exposures drops noise levels. It does not multiply them. Averaging base ISO images produces silky smooth exposure with ages old Olympus Pen 1 m4/3 system. How come it should not work with K-1?

I do not give a **** about all this theoretical nonsense anymore. What works with a bog standard CaSoNikon device is somehow out of reach with K-1 because I try to do things the same way. If the service will not fix things it is game over.

[johnc]

Yes, the 100 ISO is not related to the last shot. The previous shots in this thread are at a lot higher ISO. That is what I was referring to.

If they were shot at the exact same conditions and if they are viewed at the same size, the should be about same. If you compare at 100% view, the K-1 should show x36/24 as much chroma noise. But that assumes the sensor heat-up is the same, which can be a totally different one due to construction.
As already noted: We currently can't see the actual sensor temperature, like we can on a K-3.

Cheers, I get you with some of the earlier examples.

I'm just looking carefully at my K-1 and K-3 images again here, based on a quick (lens-cap on) test in the house this afternoon. Both cameras adjacent to each other in a dark bag. Chroma is only visible at all where the images have been pushed to specifically exaggerate the problem (e.g. by seriously increasing highlights or clarity on the frame).

That said, while the K-3 has a more significant noise-floor across the frame, there is certainly a higher number of more intense chroma speckles on K-1 than K-3 at 30 seconds ISO 100. (note again - while I say 'more intense', these are only visible at all with very significant processing of the file). By the way, It seems on the K-1 that that long exposure NR doesn't help with these. Also that this is a non-issue for shorter exposures at base ISO. And I didn't notice the problem being worse in the field with higher ISOs (e.g. 1600).

So, on my body it is a subtle effect that can be seriously exaggerated by the 'perfect storm' of extreme processing we have seen from some posters here. I would say for me, at ISO 100, for critical long-exposure low light photography it's something to be aware of, and can be worked around by sensible processing. Not a show stopper!

But is it really right that I should be seeing this? I recognise that there will be differences in sensor performance and noise characteristics between different bodies. That is normal. But I am a little surprised by these findings, I kind of hoped that 30 seconds, ISO 100 on K-1 would be more free from chroma than this. And I wonder if D810 is better, following the 'fix'?

The clear work-around is to be careful with highlight adjustment and clarity on long-exposure images with dark areas in them. Or to employ a processing strategy that doesn't just apply the same boosts across the entire frame. But that is normal for any critical processing of low-light photography I'd say.

John

[beholder3]

But is it really right that I should be seeing this? I recognise that there will be differences in sensor performance and noise characteristics between different bodies. That is normal. But I am a little surprised by these findings, I kind of hoped that 30 seconds, ISO 100 on K-1 would be more free from chroma than this. And I wonder if D810 is better, following the 'fix'?

I fear, it is just normal.
On this page here somebody has compared a load of cameras includig the K-1 and its peers:
The Sensor Noise DB. – Brendan Davey Photography
You can see them in comparison by selecting them sequentially in the drop down field at the end of the page.
It is even giving statistical analysis data on the noise beyond subjective view with MEAN, STANDARD DEVIATION and MEDIAN.

That is the best resource for this topic I could find at least. If you find better ones, go ahead, that would be interesting.

[johnc]

I fear, it is just normal.
On this page here somebody has compared a load of cameras includig the K-1 and its peers:
The Sensor Noise DB. – Brendan Davey Photography
You can see them in comparison by selecting them sequentially in the drop down field at the end of the page.
It is even giving statistical analysis data on the noise beyond subjective view with MEAN, STANDARD DEVIATION and MEDIAN.

That is the best resource for this topic I could find at least. If you find better ones, go ahead, that would be interesting.

Nice find. My inquisitive mind would love to see that comparison at ISO 100 too.

And I do think you are most likely right, I don't see my observations as meaning there is any problem with mine here. Sure there must be some 'anomalous' bodies out there - given the number that have been sold its inevitable that some have real issues. But of course, it's unwise to regard unusual situations guaranteed to result in noise as being evidence of some kind of wider problem.

John

[JBox]

It's a bright sunny morning here and my visibility is a little bit reduced but I can't see white dot problem in your last file named 6400T30.DNG.

Thanks for taking a look. What raw converter/viewer were you using?

Not saying that the level of chroma noise is acceptable - but is this Lighroom factor a cause of some of the confusion here?

I think that software differences were a source of confusion for me at least. Having said that, it sounds like I will need to do some longer exposure tests to really compare to others' examples here.

[stevebrot]

I tried the FastRawViewer program and the noise appears as mostly red dots.

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.

This is 10x30sec frame multiexposure @ ISO100 using in-camera average blending.

Nice shot. Was this done using your loaner K-1 or the defective sensor camera you sent in for service/exchange? Edit: From later comments, the shot was probably taken with the camera that was sent in for service.

It has clarity upped by +40 in LR.

Sounds like a winning strategy if one is trying to minimize processing artifact, but not really.


Steve

[stevebrot]

Averaging exposures drops noise levels. It does not multiply them.

I believe you may be mistaken on this point. There is more than one way to cook a merge and not all reduce noise.

Averaging base ISO images produces silky smooth exposure with ages old Olympus Pen 1 m4/3 system. How come it should not work with K-1?

Perhaps because it is not a Pen E-P1? I guess one might also ask why the Pen E-P1 is not your current camera of choice and what the deficiencies were with the multiple other cameras you used in the past. After all, the K-1 has very few unique features that make it stand out from other cameras you have tried in the past.

BTW, how are things with the K-1 loaner you have on hand while your camera is in for service?


Steve

[banep]

What the ****?

Averaging exposures drops noise levels. It does not multiply them. Averaging base ISO images produces silky smooth exposure with ages old Olympus Pen 1 m4/3 system. How come it should not work with K-1?

Calm down, please.
First I'm not sure what are you talking about.
Second if you are talking about multi-exposure in average mode it does exactly and only what that name suggest: averages exposures.
Third averaging exposures is not related anyhow to reducing noise levels, but could be used for noise reduction if in-camera noise reduction is turned on.
That depends on manufacturer's implementation of that functionality.

[Tas]

It was set to average. I did nothing but multiexposures that night and actually messed up the other way around - started star trail capture with average blending mode in the end

Anyways my unit leaves to service tomorrow. I will keep you posted what happens.

Good luck with getting to the bottom of the issue, and thank you for the feedback on what you've experienced.

It will be interesting to see what they come back with.

Tas