[beholder3]

If detrimental effects are only to be expected in fringe conditions then I don't think it would be fair to proclaim a general problem.

I bet you understand that the subjective bit here is, what is "fringe conditions".
I could argue that the vast majority of photographers uses smartphones and considers all scenarios where a ILC is preferable "fringe conditions". And that would be a sound quantitatively based assumption.

Within the minority group ILC usage class again it is open to debate which items are "fringe conditions". I could argue that all shooting scenarios where the dynamic range differences between a K-7 and a K-1 are important are "fringe conditions". And just the same any shooting scenarios where today noobs complain about improved automation helpers (autofocus, green mode etc) are "fringe conditions" as for vast majority user groups they do not pose any problem.
Since all caneras today are so well rounded I guess all negative findings are for "fringe conditions".

I see only two scientifically sound ways to approach this:

1. we describe the issues as any general problem plus we describe the caveat : the exact parameters under which they appear and where it will not. Here the burden of evaluation is still on the reader to judge relevancy.
   It is like all "resolution" discussions on forums. I guess only a tiny minority really understands how seldom they'll ever recognize differences between real life images.
2. (preferred for most levels of readers) we first define a realistic usage pattern in detail (some users who shoot wildlife day in and day out could discuss and describe their usage) and the "test" / "measurement" then tries to emulate the scenario. The findings obviously then only are valid for this usage pattern. Wildlife sensor heatup is not the same as sensor heatup for users who already complain that they have to push the button again after 30 min of non-stop video recording.

[house]

Photographic Dynamic Range versus ISO Setting doesn't indicate noise reduction until iso 32180. What gives? Flawed methodology from Photons to Photos or mistaken description of the Sony problem?

Based on the tests nr is not in widespread use for raw files despite frequent comments on pentaxforums suggesting it is.

[beholder3]

Flawed methodology from Photons to Photos

^^This.

If you follow the links already provided in above threads you'll see that Jim Kasson has clearly identified Sony noise reduction massaging raw files: The Sony a7RIII eats stars
Sony manipulates raw files at least from ISO 1000 (probably even at lower ISO) by "spatial filtering".
And Jim Kasson is rather a Sony fan than somebody being too critical.


Bill Claff's page simply turns a blind eye (as discussed elsewhere, his whole approach is highly questionable in many other ways as well).

[Rondec]

Photographic Dynamic Range versus ISO Setting doesn't indicate noise reduction until iso 32180. What gives? Flawed methodology from Photons to Photos or mistaken description of the Sony problem?

Based on the tests nr is not in widespread use for raw files despite frequent comments on pentaxforums suggesting it is.

If you look at beholder's link, Sony is doing minimal filtering for exposures under 3 to 4 seconds, but doing a bunch of work on images with longer exposure times. Most of this stuff is over my head, but I guess the sensors are just too noisy for long exposures without such noise reduction behind the scenes.

Bill Claff is not looking at such images, choosing rather to look at images shot with shorter exposure times and so he doesn't pick this up in his analysis. I suppose it goes back to Class A's "fringe shooting conditions" question.

[biz-engineer]

This is a fundamental drawback of an EVF -- the sensor runs full time in "video" mode. That 15° C higher temperature equates to 4X the dark current noise in the sensor.

They will do fine on test charts, depending how the camera is used. When I used the A7III for the first time, I thought the sensor would be switched off as soon as I take my eye off the EVF, but no, by default the sensor is running all the time and that drains the battery. By default the A7III switches between EVF and back LCD, so it's running all the time as long as the camera is switched on. I asked the vendor if this was normal, he answered "yes, that's the way it works, but you can go in the menus and change it", I changed is, but then there was a little lag when looking through the evf again as the camera needs to switch the sensor on again, and acquire a number of frames to readjust the exposure. For me, trying the A7III was an interesting experience.

[Rondec]

Interesting idea, but I'm not sure what is "best"; I guess it depends on the target audience.

I feel the CIPA conditions reflect serious photography better than a "many shots per on-time" approach does. A "on-time" could all too easily be converted into many images that wouldn't materialise in the context of serious photography.
As beholder3 put it, a snapshooter may get the so-called "real life" figures from a MILC, but a serious photographer might be closer to getting the CIPA figures.

I'm assuming that high-end cameras are more often used for serious photography and that testers do not always succeed in emulating the respective circumstances (to put it mildly) .

So comparing DSLRs with MILCS in terms of "minutes of on-time" would yield figures that would be useful for snapshooters (or technicians interested in efficiency) but wouldn't be as useful as an approach that considers that one can use a DSLR during "off-time" as well.

I think it would be interesting to know both things. How much battery life does it eat if you use the viewfinder a lot but don't necessarily snap photos? I don't really know for a professional shooting, say, a wedding, what it would mean. Typically for a wedding you shoot a quick burst, then move to another spot, shoot another quick burst and so on through an eight hour day. How many batteries would you burn through on an A7 III versus a K-1? My wife typically has a grip on and goes through most of two batteries per camera in a day of shooting, but doesn't have to change batteries, but she is only shooting still images.

[Class A]

I could argue that all shooting scenarios where the dynamic range differences between a K-7 and a K-1 are important are "fringe conditions".

You could argue that but I believe many would disagree, as running into the limitations of the K-7's sensor does not require extreme situations or outrageous expectations towards dynamic range.

For sure, what constitutes "fringe scenarios" is a judgement call. However, I think we can all agree that shooting at -30°C, for instance, is not a situation that many would encounter. Likewise, we could say that taking a very high ISO image after 30min of shooting video isn't a common scenario either. I'm not trying to make the case that there is a clear cut line between "normal" and "fringe scenarios", but I think that the concept of "rather uncommon scenarios" is relatively well-defined.

You state later

we first define a realistic usage pattern in detail

(emphasis is mine)
I meant "non-fringe" to be similar in spirit to your "realistic".

Since all ca,eras today are so well rounded I guess all negative findings are for "fringe conditions".

I'd disagree with such an assumption.

For instance, the striping of some MILCs can be excited by regular backlit scenes. These stripes may not occur very often, but backlit scenes are not uncommon and it is to be expected that a regular photographer (as opposed to deep sea diving photographer or extreme arctic explorer photographer) will encounter the issue at some point in time. I think "striping" is not comparable to an issue that would only occur when pursuing rather obscure photography applications.

I'd also say that once a certain DR has been reached across the board, it is fair to call out cameras that do not reach that common level. While pushing an exposure by five stops could have been considered unreasonable (or "fringe") in the past, I don't think that still applies today, for instance because HDR images have become quite popular and it is an advantage if one can simply push the shadows of a single exposure instead of having to combine several exposures.

Again, I'm not trying to establish absolutes, just arguing that it isn't impossible to define reasonable expectations towards camera performance in today's context.

we describe the issues as any general problem plus we describe the caveat : the exact parameters under which they appear and where it will not. Here the burden of evaluation is still on the reader to judge relevancy.

Yes, that is certainly a good approach and I don't see any conflict to the other approach you are describing.

we first define a realistic usage pattern in detail

Yes, or we identify the affected realistic usage patterns after we have established in what range of situations issues should be expected.

So as soon as it is known when issues occur in what intensity, one can derive in what situations issues should become relevant and then the reader still can decide whether these situations are relevant to them or not.

[Rondec]

You could argue that but I believe many would disagree, as running into the limitations of the K-7's sensor does not require extreme situations or outrageous expectations towards dynamic range.

For sure, what constitutes "fringe scenarios" is a judgement call. However, I think we can all agree that shooting at -30°C, for instance, is not a situation that many would encounter. Likewise, we could say that taking a very high ISO image after 30min of shooting video isn't a common scenario either. I'm not trying to make the case that there is a clear cut line between "normal" and "fringe scenarios", but I think that the concept of "rather uncommon scenarios" is relatively well-defined.

You state later

(emphasis is mine)
I meant "non-fringe" to be similar in spirit to your "realistic".


I'd disagree with such an assumption.

For instance, the striping of some MILCs can be excited by regular backlit scenes. These stripes may not occur very often, but backlit scenes are not uncommon and it is to be expected that a regular photographer (as opposed to deep sea diving photographer or extreme arctic explorer photographer) will encounter the issue at some point in time. I think "striping" is not comparable to an issue that would only occur when pursuing rather obscure photography applications.

I'd also say that once a certain DR has been reached across the board, it is fair to call out cameras that do not reach that common level. While pushing an exposure by five stops could have been considered unreasonable (or "fringe") in the past, I don't think that still applies today, for instance because HDR images have become quite popular and it is an advantage if one can simply push the shadows of a single exposure instead of having to combine several exposures.

Again, I'm not trying to establish absolutes, just arguing that it isn't impossible to define reasonable expectations towards camera performance in today's context.


Yes, that is certainly a good approach and I don't see any conflict to the other approach you are describing.


Yes, or we identify the affected realistic usage patterns after we have established in what range of situations issues should be expected.

So as soon as it is known when issues occur in what intensity, one can derive in what situations issues should become relevant and then the reader still can decide whether these situations are relevant to them or not.

Do you think that Sony should be identified by review sites as instituting non-defeatable noise reduction in raw images of over 4 seconds duration?

[Class A]

Photographic Dynamic Range versus ISO Setting doesn't indicate noise reduction until iso 32180.

NR (in the sense as assumed by Bill Claff) is not present for the Sony 7RIII until ISO 32180 but already starts at ISO 636 for the K-1 II.

As Rondec has pointed out, there are good reasons as to why these results need not be distrusted. I don't know much about what Sony's approach to removing unwanted artefacts is, but if their processing is limited to eliminating certain (in their view pathological) hot-pixel constellations then such isolated processing would not necessarily show up in Bill Claff's analysis.

This is not to say that the Sony processing is OK, for sure it sometimes eliminates real signal that it just mistakes for hot-pixel constellations. I'm against such (non-optional) RAW fixing as much as I'm against the K-1 II's RAW massaging. My only point is that there is no need to suspect incompetence or malice on behalf of Bill Claff.

If you follow the links already provided in above threads you'll see that Jim Kasson has clearly identified Sony noise reduction massaging raw files:

Yes, but it isn't clear that the respective Sony processing would show in the conditions Bill Claff set for his tests. For one or another reason, the processing (if it takes place under the conditions he assumed/set) was not measurable using the method he applied.

Bill Claff's page simply turns a blind eye (as discussed elsewhere, his whole approach is highly questionable in many other ways as well).

I don't think that's a fair statement.

I rather assume he only looked at a limited range of shooting conditions he thought to be realistic. I don't think he claims to do a comprehensive sensor analysis.

From our perspective, it may seem regrettable that the K-1 II's processing is caught by his analysis whereas the Sony A7RIII isn't (despite having other flaws), but I don't think that proves any ill-intent on behalf of Bill Claff.

Based on the tests nr is not in widespread use for raw files despite frequent comments on pentaxforums suggesting it is.

Well, as you can see both cameras process RAW files at some point in time. Virtually all cameras do at very high ISO settings.

The main discussion on pentaxforums, as I see it, was about the K-1 II starting to apply a pretty heavy-handed NR already at moderate ISO settings (starting at ISO 636).

[Class A]

Do you think that Sony should be identified by review sites as instituting non-defeatable noise reduction in raw images of over 4 seconds duration?

Of course, why wouldn't I?

While exposures over 4s may not be commonplace, they are certainly not extreme, IMO.

In any event, a review site should be examining all camera properties even if certain phenomena are only expected in rare situations. It should be up to the reader to decide how relevant a certain limitation is to them.

May I also add that there are many different review sites and some target specific readers only. I don't think it would be fair/appropriate to have the same expectations regarding scrutiny level towards all review sites. However, for a given review site, it is of utmost important to be consistent in their evaluation. A given site cannot perform evaluations at an microscopic level and make a huge deal out of a shortcoming for one brand and turn a blind eye to arguably more invasive issues for another brand. So, for instance, if a review site has a very principled view against the manipulation of RAW data, it should hold that view on all cameras it evaluates.

[photoptimist]

Interesting idea, but I'm not sure what is "best"; I guess it depends on the target audience.

I feel the CIPA conditions reflect serious photography better than a "many shots per on-time" approach does. A "on-time" could all too easily be converted into many images that wouldn't materialise in the context of serious photography.
As beholder3 put it, a snapshooter may get the so-called "real life" figures from a MILC, but a serious photographer might be closer to getting the CIPA figures.

I'm assuming that high-end cameras are more often used for serious photography and that testers do not always succeed in emulating the respective circumstances (to put it mildly) .

So comparing DSLRs with MILCS in terms of "minutes of on-time" would yield figures that would be useful for snapshooters (or technicians interested in efficiency) but wouldn't be as useful as an approach that considers that one can use a DSLR during "off-time" as well.

That CIPA figure of 310 shots for the Z6 suggests the Z6 has about 155 minutes of on-time battery life (about 2 and half hours).

As for "serious photography," the CIPA number only accurately reflects one specific type of serious photography: people who average 30 seconds of composition/setup time and then take a single shot. Super-serious photographers who take 60 seconds for composition/setup time would get half the shot count during the roughly 155 minute battery life. Photographers who use bracketing (i.e., 3 to 5 shots after a 30-second setup) would get more shots than 310 during the roughly 155 minute battery life. Wedding/sports shooters who use lots of bursts might get many more shots. On the other hand, very serious wildlife shooters who patiently watch their quarry through the EVF and telephoto lens while waiting for the perfect positioning or behavior of the creature might get no shots at all during the 155 minute battery life.

For DSLR shooters, whether they take 5 seconds to grab a snapshot or 5 minutes to carefully capture the decisive moment makes only a minor difference to the number of shots they get per battery. On the other hand, DSLR operating on-times per battery charge might range from a couple of hours to dozens of hours depending on how the user shoots. For MILC shooters, shots per battery would range all over the place, but on-time per battery would be more nearly constant.

[biz-engineer]

The main discussion on pentaxforums, as I see it, was about the K-1 II starting to apply a pretty heavy-handed NR already at moderate ISO settings (starting at ISO 636).

The big advantage of the Pentax K-1II is that the Pentax K-1 now sells for 1400 Euros, new, makes the original K-1 more affordable for a lot of people. You need to see Pentax from a positive angle.

---------- Post added 13-01-19 at 17:46 ----------

Mirrorless cameras and sensors - the good, the bad and the really ugly

Mirrorless cameras are something that customers did not asked for. Mirrorless was an idea from Sony executives when Sony failed to penetrate the DSLR market, they needed something different to develop into a market free of competition. And then, when Canon and Nikon saw their DSLR sales plummet due to general market shrinking and due to satisfied customers not upgrading, they decided it was time to force people to obsolete what they have (DSLR) and changes they bodies and lenses for the new mounts. Mirrorless was and still is business driven, it was never customer driven.

[beholder3]

This here is a quite good example of what we are talking about (even if old news):
Noise comparison between hot and cold sensor (long exposure): Micro Four Thirds Talk Forum: Digital Photography Review
The decrease of image quality in the mirrorless camera after just 10 minutes of video usage and heating the sensor is quite obvious.

Another interesting bit is that Canon even explicitly seems to warn about image degradation after longer liveview shooting:
dslr - Do mirrorless cameras have more sensor noise because they are constantly exposed to light? - Photography Stack Exchange

And here I found a very good graph showing how different the sensor noise for long exposures can grow:
Length of Exposure vs Sensor Temp. - DSLR & Digital Camera Astro Imaging & Processing - Cloudy Nights
At the end of the thread there is a similar measurement to the one I have done, there with a Nikon D5100 doing long exposures.

and here are some interesting evaluations:
https://www.thorlabs.com/newgrouppage9.cfm?objectgroup_id=10773
(old tiny sensors, but still interesting)

[clackers]

I think it would be interesting to know both things. How much battery life does it eat if you use the viewfinder a lot but don't necessarily snap photos? I don't really know for a professional shooting, say, a wedding, what it would mean. Typically for a wedding you shoot a quick burst, then move to another spot, shoot another quick burst and so on through an eight hour day. How many batteries would you burn through on an A7 III versus a K-1? My wife typically has a grip on and goes through most of two batteries per camera in a day of shooting, but doesn't have to change batteries, but she is only shooting still images.

I read a Sony A9 owner complaining to another that after about four hours use whether he took lots of pictures or not the battery was dead.

It's always in video mode, always reading out, and there's extra demands because it's a new kind of expensive EVF that has less lag than other EVFs, almost as good as an OVF.

[Class A]

You need to see Pentax from a positive angle.

I have an overall positive outlook on Pentax.

However, this doesn't stop me from becoming concerned when I perceive Ricoh to be making mistakes.
For instance, I thought the K-01 was ill-conceived and expressed myself accordingly at the time. It wasn't completely without merit, but many mistakes were made in designing the K-01 and its commercial failure did not surprise me in the least.

I think that solely regarding the K-1 II as a positive because it makes the K-1 cheaper is short-sighted. First of all, even a K-1 II without mandatory noise reduction would have made the K-1 a cheaper buy. Second, the K-1 has been discontinued. All arguments based on the availability of the K-1, e.g., "buy a K-1 instead", "the K-1 has become cheaper", ignore that they only work for a limited time. Third, if Ricoh were to continue to use mandatory noise reduction of RAW files at relatively low ISO settings for all its future products then I'd regard that as very concerning. I do not think that this would be sustainable.

Could I kindly ask for this thread not to be used for discussing the K-1 II?
There is a large existing thread for that purpose and this particular thread was meant to focus on DSLRs vs MILCs.

Mirrorless was and still is business driven, it was never customer driven.

I agree that mirrorless cameras are very interesting for companies because they are cheaper to make and could enable "DSLR -> MILC replacement sales" that otherwise wouldn't occur.

I wouldn't go as far as stating that MILCs are of no interest to customers, though. It is a game of trade offs as there are certainly advantages to MILC designs. My problem with a lot of the tech press is that they exaggerate negatives associated with DSLRs and are very forgiving regarding the negatives of MILCs.