comparining micro four thirds to full frame

[HSakols]

Too snarky? Not at all. I'm sorry you are unable to bear the weight of a full-frame. How old are you, if you don't mind my asking?

Ray, I'm in my early 50's and still love to backpack in Yosemite's wilderness.  Backpacking doesn't create better photos but it is something I have to do to stay inspired and stay in shape.  I enjoy it!  When I drive the Yosemite Valley loop I take my D800 with a hefty tripod - it's my 8x10 camera!

Here is another from Monday morning.  And yes I use a tripod for every shot just as I would with a larger format. It forces me to compose better. 

 

[Ray]

Ray, I'm in my early 50's and still love to backpack in Yosemite's wilderness.  Backpacking doesn't create better photos but it is something I have to do to stay inspired and stay in shape.  I enjoy it!  When I drive the Yosemite Valley loop I take my D800 with a hefty tripod - it's my 8x10 camera!

Here is another from Monday morning.  And yes I use a tripod for every shot just as I would with a larger format. It forces me to compose better.

Nice shot, Hugh! At least you have some trees amongst the snow. At 17,000 feet in Nepal it's a bit more rugged, as the attached, stitched panorama shows, taken with the Nikon D7100 and 24-120/F4 zoom in 2013.

When zooming in on the panorama, I was surprised to see another couple of trekkers who appear to be emerging out of the snow, waist deep. I don't recall noticing that before. I've made a 100% crop which shows that additional couple in the lower left hand corner of the crop. Perhaps it's a perspective quirk. 

[Ray]

Which would make sense, except that DXO badly mangles the horizontal positioning of the data points, by confusing a measure of "highlight headroom after analog amplification into the raw file" with a measure of how much exposure the sensor is getting: Exposure Index, which is what is needed for low-light handling comparisons. To fix this, from what I recall reading, it is sufficient to use the EI values reported by the camera camera maker. Here are some from https://www.dxomark.com/Cameras/Compare/Side-by-side/Nikon-D850-versus-Olympus-OM-D-E-M1-Mark-II___1177_1136

This is getting complicated, BJL. Perhaps the topic deserves another thread.  If you distrust the DXOMark results, then please do your own real-world comparisons and demonstrate to us how those qualified scientists employed by DXO are confused.

As I understand, the concept of 'highlight headroom' is only relevant for people who use the auto-exposure feature of their camera and don't take the trouble to get a proper ETTR exposure.

For example, consider a case of identical sensors in two different brands of camera. Both sensors could have exatly the same DR, but the different manufacturers who  design the cameras will supply different features to appeal to different customers. Changing the nominated ISO sensitivity, as it relates the camera's automatic metering system, will affect the so-called 'highlight headroom'.

I suspect that Olympus is capitalizing on the increased DR of its new cameras by encouraging underexposures so there will be fewer incidents of blown highlights. Here's a good explanation of the issue that I came across on the internet.

"As tests show, the ISO settings reported by camera manufacturers can differ significantly from measured ISO in RAW. This difference stems from design choices, in particular the choice to keep some “headroom” to avoid saturation in the higher exposures to make it possible to recover from blown highlights."

"Suppose you have a camera with a 12 stop exposure range. You can set the middle gray aim point at Zone 6, which gives you six stops of highlight and six stops of shadow range. Or you could decide that you're more concerned about highlights blowing out than anything else and set the aim point at Zone 5, which gives you an extra stop of highlight range but also a little more noise. Or you could go the other way and set the aim point at Zone 7, which gives you more shadow detail and lower noise but increases the risk of blown out highlights."

As I understand, the correct, or precise way to compare the DR figures on the DXO graphs is to draw a vertical line which intersects the two graphs. When the manufacturers' claimed ISO sensitivities are different, the vertical line will pass through only one DR reading. However, the value of the other reading can be easily estimated, and in practice, if you are doing comparisons with the actual cameras and lenses, the lower ISO should be increased, as far as possible, by 1/3rd ISO intervals, or whatever increments are available, till it reaches the higher actual ISO of the other camera, or if you prefer, the higher ISO should be reduced to match the lower.

For example, lets consider the DR reading for the E-M1 MkII at the manufacturer's claimed ISO of 1600, which is actually ISO 688 according to DXO. The DR reading is 10.37 EV. Drawing a vertical line from the ISO 1600 point, the line intersects with the D850 curve at approximately 1/3rd of an ISO beyond the Nikon nominated ISO of 800 (which is actually ISO 563). Increasing the D850's ISO by 1/3rd would make the camera's ISO reading ISO 1000, or actually ISO 703, which is close enough to ISO 688.

To compare the two cameras at this ISO setting, the E-M1 should be set at ISO 1600 and the D850 at ISO 1000. If the different lenses used at the same F/stop and same equivalent ISO seem to require different shutter speeds for an ETTR exposure of both images, then the likely cause is that the transmission of the lenses, or T-stop, is different.

A crucial point when attempting to do your own comparisons, is to make sure you are starting from a correct ETTR exposure which fully saturates, but not over-saturates, the pixels that are recording the brightest part of the scene. One then successively reduces the exposure by increasing the shutter speed by the same percentage for both cameras, keeping the ISO constant, and comparing the image quality at each reduction of exposure.

At ISO 1000 on the D850, which is approximately the same nominal sensitivity as ISO 1600 on the E-M1, the DR of the D850 pixel is at about 2/3rds of a stop better. The line intersects just above the 11 EV point.

If we compare the full sensors, downsized to a common standard, the DR advantage of the D850, at the actual ISO of 703 is about 1.3 EV better. Refer attached images of the two scenarios.

If you think this approach is wrong, then please provide real-world tests to demonstrate it is wrong. My acceptance of the accuracy of the DXOMark results is based upon my own testing.

[BJL]

Ray, as I said in my update, there is probably another correction needed, which moves the comparison in favour of the D850, and the new “screen” graph will help.No time to post it all now; hopefully later today!

[HSakols]

Nice shot, Hugh! At least you have some trees amongst the snow. At 17,000 feet in Nepal it's a bit more rugged, as the attached, stitched panorama shows, taken with the Nikon D7100 and 24-120/F4 zoom in 2013.

When zooming in on the panorama, I was surprised to see another couple of trekkers who appear to be emerging out of the snow, waist deep. I don't recall noticing that before. I've made a 100% crop which shows that additional couple in the lower left hand corner of the crop. Perhaps it's a perspective quirk. 

Wow! that second crop is strange. 
My wife and I just went to a presentation on trekking through Nepal by author Jon Bock. I look forward to spending some time there when I stop working full time, not only for the landscape, but for the culture and people. 

[BJL]

Ray, as I said in my update, there is probably another correction needed, which moves the comparison in favour of the D850, and the new “screen” graph will help.No time to post it all now; hopefully later today!

To follow up briefly: in addition to moving points right to their stated EI values 200, 400, etc., the extra highlight headroom given by those lower  raw level placements are probably counted in DXO’s total DR, but give nothing in terms of shadow handling, so to measure that (“shadow DR”) the points should then be lowered by as many stops as they are moved rightward. These roughly cancel, putting the curves in about the same places, so for now I will just use the curves as they are.

For per pixel comparisons (e.g. cropping to 20MP with the D850) the focal length ratio needed is 1.3, so about 3/4 stop higher EI on the D850: EM1Mk2 at 800 vs D850 at 1360 and so on. In other words, slide the D850 “screen” curve 3/4 stop left, or measure the horizontal spacing and see how the gap compares to 3/4 stop. Eye-balling it, it is close, with the EM1 edging ahead at extreme high EI; higher than I ever use.

If instead the full sensor is used and then downsampling to equal “print resolution” is assumed, then the “print” graphs can be used, and the focal length factor is two, so the D850 is compared at four times the EI: a two stop horizontal shift. The comparison looks about the same as above.

[BJL]

P. S. Those DXO measures of saturation based ISO speed simply show that everyone, Olympus more than some others, under-amplify when producing high EI raw files, in the sense of not bringing the nominal mid-tone level for that EI up to the minimum headroom level allowed by the ISO standard. Presumably, extra digital amplification in conversion from raw to JPEG is then indicated. It is nothing to do with exposure levels received by the sensor.

For example, some older CCD cameras like the Olympus E-1 simply capped analog gain at EI 1600: going to EI 3200 and halving exposure time halved the raw levels, so in DXO terms it was 1600 at all the higher levels. Newer near “ISO-less” sensors can benefit from something similar, reducing or eliminating analog gain, and thus adding highlight headroom with no significant damage to noise levels.

[BJL]

This pic (attached) showed me again that m43 has the best size/performance tradeoff for my wants & needs when it comes to handholdable reach. ...

That’s impressive! I’m sure you would love to have been this guy:
https://arstechnica.com/science/2018/02/amateur-astronomer-tries-out-new-camera-catches-supernova-at-its-start/

[Telecaster]

That’s impressive! I’m sure you would love to have been this guy:
https://arstechnica.com/science/2018/02/amateur-astronomer-tries-out-new-camera-catches-supernova-at-its-start/

Yeah, I heard about that! I'd need a little more aperture than what I've got to get in on that action. 

-Dave-

[Ray]

P. S. Those DXO measures of saturation based ISO speed simply show that everyone, Olympus more than some others, under-amplify when producing high EI raw files, in the sense of not bringing the nominal mid-tone level for that EI up to the minimum headroom level allowed by the ISO standard. Presumably, extra digital amplification in conversion from raw to JPEG is then indicated. It is nothing to do with exposure levels received by the sensor.

It's everything to do with the exposure levels received by the sensor. Underexposure at any ISO results in more noise and less DR. Regardless of how accurate the manufacturer's nominated ISO's are in relation to the ISO standard, the same principle applies. A valid comparison between two cameras must always be at the same DXO-measured ISO sensitivity, which means raising or lowering the ISO of one of the cameras to match the other.

If Olympus is under-amplifying when producing RAW files at high ISO, that should be no problem for those who shoot in RAW mode and who understand the need for an ETTR exposure to achieve the maximum DR and the lowest noise. I always shoot in RAW mode.

I consider myself to be a practical person. I'm concerned with actual, real-world, results. In an imaginary world where all camera designs gave one the choices of only Field of View and Depth of Field, and the selection of shutter speed and F/stop was done automatically in-camera, according to the camera's light meter and according to a DoF index which one could select by turning a wheel with one's thumb, then the E-M1 MkII would have a consistent advantage over the D850, except for shallow DoF requirements.

In the real world, we use tripods when slow shutter speeds are needed. The lighting conditions tend to be so variable that quite often the shutter speed selected is far faster than it need be for a sharp, hand-held shot, and is selected only to avoid overexposure.

Nevertheless, I admit that a 20mp E-M1 MkII, in certain circumstances, should have a clear advantage in image quality, which is why I'd like to see some real-world comparisons. I'd be particularly interested in seeing a comparison between the E-M1 with 200mm lens, and both the Nikon D850 and D7200 with 400mm lens, because I already own a Nikkor 80-400, which I'm very pleased with, and am currently undecided whether to upgrade to a D850, and/or a D7200. My D5300 is seriously disadvantaged because it doesn't have an AF fine tuning feature, which is why I'm considering a D7200 upgrade. The D7200 also has approximately 2/3rds of a stop better DR than the D5300, from base ISO up to ISO 800 (nominal).

[Ray]

Wow! that second crop is strange. 
My wife and I just went to a presentation on trekking through Nepal by author Jon Bock. I look forward to spending some time there when I stop working full time, not only for the landscape, but for the culture and people.

My association with Nepal goes back to 1964 when I was a backpacking hippie with a Pentax 35mm SLR and a couple of prime lenses. As a result of spending many hours in the 1990's scanning my old slides of shots I'd taken many years ago in Nepal, I was eventually motivated to revisit the country in 2005 with my new Canon 5D DSLR. I really enjoyed the experience and have revisited the country a number of times. The last time in 2013.

I haven't been to the country since the recent, devastating earthquake, because I think it would be too upsetting for me to witness the increased poverty and desperation of the common people. The only purpose would have been to help with the reconstruction process by contributing my own labour.

However, I expect they have now mostly recovered from the devastation, and tourism is on the rise again, especially Chinese tourism. 

Pokhara is my favourite city destination. Unfortunately, Pokhara does not have an international airport yet, but they are planning or considering one. Kathmandu is a bit of a congested mess, but the mess is so extreme it becomes interesting because it's so amazing how people can live like that.

In some respects, visiting Nepal is like travelling in a time-capsule back to the Middle Ages in Europe. In the countryside, one often sees women sitting in front of a mound of freshly-harvested rice or corn, pounding and thrashing the rice with a long stick in their hands, just as one imagines people did in the Middle Ages.

Then suddenly the woman will stop thrashing the rice, lay her stick down, and reach into her pocket to grab an iPhone, have a conversation for the next few minutes, then resume her thrashing. (The correct term is threshing, but that's usually done with machinery, so 'thrashing' seems the better word in these circumstances.  )

The first attached image was taken in 2006 when iPhones in the countryside were rare. The second image was taken in 2011.

[BJL]

Ray, read DXO’s description of how they make those so-called ISO speed measurements and get back to me. You will see that the “lower than stated” values have nothing to do with the sensor being given less exposure (which would involve the camera using a higher shutter speed than selected!). It is instead based on measurements of raw file data after any gain (analog or digital) applied in pruducing those raw files. So you (and perhaps DXO) are ignoring the issue of how much gain is applied when at a high EI setting. Clearly, and for good reasons, most or all camera makers choose a gain level that places metered midtones below the highest level permitted by the ISO standard, for the sake of reducing the risk of amplifying highlights into clipping that were not blown out in the photo sites. Conversely, it would be folly for a camera to give less exposure than indicated by the EI setting, because it make the camera look worse in testing of noise levels.

[BJL]

Ray, “in the real world”, many of us have good reasons to prefer working hand-held most or all of the time so as not to carry a tripod—including places where tripods are not allowed. Indeed, this tends to be more the style of smaller format users, while tripod use tends to be more common with larger formats. I note also that hand-holding is vastly more doable with five stops or more of IS.

I completely agree with you that when using a tripod while photographing subjects that are stationary (or slowly enough moving and well enough lit), so that a sufficiently low shutter speed is always usable, the IQ advantages offered by a larger format are clear, at least when images are displayed large enough and scrutinised closely enough.

But wait: weren’t you comparing performance at elevated EI, which we agreed usually indicates a minimum shutter speed constraint based on subject motion, which would then be independent of format or tripod usage?

Indeed I get the feeling that one big factor in format size preferences is preferences and constraints on hand-holding vs tripod usage, and also the related preferences and constraints on kit weight.

[Ray]

Ray, read DXO’s description of how they make those so-called ISO speed measurements and get back to me. You will see that the “lower than stated” values have nothing to do with the sensor being given less exposure (which would involve the camera using a higher shutter speed than selected!). It is instead based on measurements of raw file data after any gain (analog or digital) applied in pruducing those raw files. So you (and perhaps DXO) are ignoring the issue of how much gain is applied when at a high EI setting. Clearly, and for good reasons, most or all camera makers choose a gain level that places metered midtones below the highest level permitted by the ISO standard, for the sake of reducing the risk of amplifying highlights into clipping that were not blown out in the photo sites. Conversely, it would be folly for a camera to give less exposure than indicated by the EI setting, because it make the camera look worse in testing of noise levels.

BJL,
One of us seems very confused here. Perhaps it's me. As I understand, it's the internal processes of the camera that apply the gain at higher ISOs, in accordance with the manufacturer's design and decisions. We don't have a choice in the matter. Nor does DXO.

Pixel saturation, in at least a part of the sensor, is a required starting point to measure DR. At higher than base ISO, pixel saturation is achieved through in-camera amplification, not through the absorption of too many photons. This is why ISO-less, or ISO invariant cameras have an advantage. At fast shutter speeds, when one might normally raise the ISO setting, one can avoid using the camera's in-built amplification system, and do one's own amplification in Photoshop, thus removing the risk of blowing highlights, yet still maintain the same image quality as a correctly exposed shot at the usual higher ISO setting.

Unfortunately, Nikon DSLRs are not exactly ISO-invariant. There's usually a slight advantage in using the camera's in-built amplification, especially between base ISO and ISO 800. For example, if a camera is truly ISO-invariant then the drop in DR should be 3 EV when one uses ISO 800 instead of underexposing 3 stops at ISO 100.

However, with the D850, using the camera's amplification by choosing ISO 800, instead of using the same exposure at ISO 100, results in a drop in DR of only 2 stops. One gains a full stop of DR by using the ISO 800 setting on the D850. I find this type of information one can glean from the DXOMark graphs very useful in a practical sense.

Whilst many manufacturers over state their ISO settings, in relation to the ISO standards, most of them over state them by only 1/3rd to 2/3rds of a stop. Olympus over states the ISO sensitivity of the E-M1 MkII by over one full stop, approximately 1 & 1/4th of a stop. I see an element of marketing salesmanship here. By doing this the manufacturer is basically stating that their camera has an elevated ISO range at both the high end and low end, from ISO 200 to ISO 25,600. The actual range, according to the ISO standards, is from ISO 83 to ISO 10,916, approximately equal to ISO 100 to 12,800.

Do you see the deception? I've used cameras that have a base ISO of 200, such as the Nikon D700 full-frame, although the DXO-measured ISO was 158 which is approximately 1/3rd of a stop down, as is the case with most cameras (except Olympus, apparently).

I've found such cameras to be very advantageous for hand-held shots because at a real ISO of 158 (as opposed to a real ISO of, say 83, (which the E-M1 MkII has at base ISO), there is far less need to mess around changing the ISO settings back and forth. ISO 200 is useful for more conditions.

To advertise a camera as having a base ISO of 200, when in practice it is ISO 83, seems deceptive to me.
Now this leads on to the nub of the issue. At base ISO I presume that the camera's in-built software does not amplify the signal to compensate for the underexposure of the sensor that takes place as a matter of course at higher ISOs. Right?

If we were to compare a Nikon D700 and an Olympus E-M1 MkII, shooting the same scene, at the same nominated ISO of 200, with equivalent focal length lenses to give the same FoV, using the lenses at the same F/stop, and ensuring that the lenses had the same transmission or T-stop, would the shutter speeds be approximately the same for both cameras, if we were shooting in RAW mode and attempting to achieve an ETTR shot for maximum DR and the lowest noise?

That's the $64,000 question. Can you answer it JBL and dispel my confusion. 

I am assuming that the E-M1 MkII, in those circumstance, at base ISO, which should involve no analogue amplification of the signal, will require approximately double the exposure of the D700 to get an ETTR shot in RAW mode.

I won't answer the other points you've made until I am clear about this issue. It would be really useful if someone reading this thread happens to own both a D700 and E-M1 Mk II and is prepared to do the comparison under the conditions I've described. However, an awareness of the T-stop factor is also essential for accurate and meaningful results.

[BJL]

BJL,
One of us seems very confused here. Perhaps it's me.

Yes, I am fairly sure it is you, gin the number of flatly incorrect statements in the rest of your post. I will try to dig up and refresh an old post that I made, explaining what DXO actually measures and what the ISO 12232:2006 standard actually says and does not say, but in summary:

- The ISO standard defines several measures of completely different aspects of camera performance (SSNR10 and SSNR40, SSat, SOS and REI); these are not just several ways of measuring the same thing, even though they use the same units of measurement. I will use the analogy that the focal length and the physical length of a lens are both measured in mm, but are not the same thing: their values are often similar, but can be wildly different as with some slow telephoto lenses and with extreme wide angle lenses.

- The "ISO" setting on a camera is for Exposure Index, relating to shutter speed at given film/sensor illumination conditions, as roughly described by "sunny 16" film speed description: "the shutter speed used to get correct exposure with standard development if the subject is an 18% gray card illuminated by bright sunshine and photographed at f/16".

- The SSat measure of highlight headroom is not in any sense a measure of Exposure Index, even if the values are typically similar; you (and maybe DXO) seems to confuse these two distinct concepts. Saying that a company is lying when the Exposure Index ("ISO" setting) differs significantly from the measured SSat is as totally wrong as saying that a lens maker is lying if the stated focal length of a lens is significantly different than its measured physical length.

- Likewise, there is no reason why the minimum Exposure Index setting should equal the sensor base-ISO speed. Instead, a guideline is that the SSat measure at a given EI setting should be at least a bit less than the EI value, to reduce the risk of blown/clipped highlights. Any EI setting less than the base ISO speed should be flagged as an "extended low" EI, to warn of its sub-standard highlight headroom and greater proneness to blown highlights. This is what Nikon does with its EI=32 setting, which according to DXO has SSat=44, the sensor's base ISO speed.

But you will have to wait for me to dig out my justification of these statements.

[Ray]

BJL,
Please don't misinterpret my use of the word 'deception' as lying. Deception is a normal part of every day life, amongst all creatures, not just humans. In Australia, lying in advertisements for products is illegal. Deception is not. There is a subtle difference which I'm sure you must be able to appreciate.

If in fact the E-M1 MkII produces an ETTR shot in RAW mode at ISO 200, at approximately the same shutter speed and F/stop as most other DSLRs at the ISO 200 setting, give or take 1/3rd of a stop, then I would consider DXOMark's rating of ISO sensitivity as being deceptive.

I'm not in the business of promoting any particular brand of camera. I'm concerned only with the facts.
I did a search for reviews of the Olympus E-M1 MkII, and it certainly has some impressive features, but my confusion about its base ISO remains.

Here are a couple of quotes from: http://www.ayton.id.au/wiki/doku.php?id=photo:olympusomdem1ii
and  http://www.ayton.id.au/wiki/doku.php?id=photo:olympusomdem1ii#fn__2

Issues
"ISO settings are more than 1EV higher than actual measured ISO and the LOW ISO of 64 was measured as being the same ISO as ISO 200 setting - that is, both equate with a true ISO of 83!"

"Strangely, the E-M1 II's ISO is actually around 1 stop higher than actual measured ISO except for the extended low ISO of 64 which is really ISO 83 and comparable to it's ISO 200 (the E-M1 I base ISO of 200 was measured at ISO 122, which was about the same as it's LOW ISO of 100) - so be careful when setting exposures manually based upon other camera's exposure settings!"

Even they seem to be confused.

I also did a search on the ISO Standard 12232:2006, and came across a paywall for the full pdf. However, I also came across the following comments from someone who appears to have read the full document.

"The ISO standard 12232:2006 gives digital still camera manufacturers a choice of five different techniques for determining the exposure index rating at each sensitivity setting provided by a particular camera model. Three of the techniques in ISO 12232:2006 are carried over from the 1998 version of the standard, while two new techniques allowing for measurement of JPEG output files are introduced from CIPA DC-004.

Depending on the technique selected, the exposure index rating can depend on the sensor sensitivity, the sensor noise, and the appearance of the resulting image. The standard specifies the measurement of light sensitivity of the entire digital camera system and not of individual components such as digital sensors, although Kodak has reported using a variation to characterize the sensitivity of two of their sensors in 2001."

"The Standard Output Specification (SOS) technique , also new in the 2006 version of the standard, effectively specifies that the average level in the sRGB image must be 18% gray plus or minus 1/3 stop when exposed per the EI with no exposure compensation . Because the output level is measured in the sRGB output from the camera, it is only applicable to sRGB images—typically JPEG—and not to output files in raw image format. It is not applicable when multi-zone metering is used.

The saturation-based technique is closely related to the SOS technique, with the sRGB output level being measured at 100% white rather than 18% gray. The saturation-based value is effectively 0.704 times the SOS value. Because the output level is measured in the sRGB output from the camera, it is only applicable to sRGB images—typically TIFF—and not to output files in raw image format. It is not applicable when multi-zone metering is used."

I look forward to any clarification you can provide on this issue, BJL.

[BJL]

I have mostly prepared a detailed discuss of what the Exposure Index setting on a camera is and is not about, but meanwhile, just one comment: the ISO standard like 12232:2006 do not define an quantity called the "ISO" or the "true ISO" of a camera; that is just casual, somewhat sloppy language, with "ISO" being used roughly as a synonym for Exposure Index. The ISO standards instead define measurements of several different quantities — and "SSat" is simply not the same thing as Exposure Index, "EI".

Thus when DXO says:

"ISO settings are more than 1EV higher than actual measured ISO and the LOW ISO of 64 was measured as being the same ISO as ISO 200 setting - that is, both equate with a true ISO of 83!"

it is literally nonsense, and reveals a mysterious sloppiness or ignorance, and even a touch of arrogance when it says "true ISO" without knowing the truth of what ISO standards actually say and mean.

More in a while!

[Ray]

I have mostly prepared a detailed discuss of what the Exposure Index setting on a camera is and is not about, but meanwhile, just one comment: the ISO standard like 12232:2006 do not define an quantity called the "ISO" or the "true ISO" of a camera; that is just casual, somewhat sloppy language, with "ISO" being used roughly as a synonym for Exposure Index. The ISO standards instead define measurements of several different quantities — and "SSat" is simply not the same thing as Exposure Index, "EI".

Thus when DXO says:

"ISO settings are more than 1EV higher than actual measured ISO and the LOW ISO of 64 was measured as being the same ISO as ISO 200 setting - that is, both equate with a true ISO of 83!"

it is literally nonsense, and reveals a mysterious sloppiness or ignorance, and even a touch of arrogance when it says "true ISO" without knowing the truth of what ISO standards actually say and mean.

More in a while!

Don't worry about it, BJL. DXOMark measurements are relevant only for those who shoot in RAW mode and who are concerned about achieving an ETTR exposure most of the time. I believe most people who use DSLRs shoot in Jpeg mode and are quite happy to let the camera do most of the image processing, in accordance with their own selection for the appropriate scene.

One could argue endlessly whether or not DXO's interpretation of the ISO standard is technically correct and strictly conforms to the letter of the standard.

For me, the only issue is the consistency of DXO's measurements and their practical significance.

For example, if at the nominated ISO of 200, Camera A has a DXO-measured ISO of 83, and Camera B has a DXO-measured ISO of 130, then camera A should require about 2/3rds of a stop more exposure than camera B, (ie. slower shutter speed), to achieve  an ETTR exposure with minimum noise.

This is the only issue for me, and this issue can only be resolved with practical experiments comparing ETTR shots at the same f/stop, same T-stop, and same equivalent focal lengths, and by making exposure adjustments if the T-stops differ.

There is also the point to consider that Guillermo made. As DR improves generally, as camera technology progresses, DR becomes less of an issue. It certainly has for Guillermo, but he used to be a fanatic about DR, even promoting a procedure known as UniWB to extract slightly more DR from an image by avoiding the clipping of perhaps just one color that could result from a camera-selected white balance.

I was never that fanatical. 

[NancyP]

Somewhat off topic:

While people are gram-counting, it might be worth thinking about how you carry that equipment. Gram counting may be essential for peak-bagging or through-hiking, less essential for day hikes and weekend hikes. A good pack and carriage system makes a huge difference in how the weight feels. Photo-specific packs don't fit me, even the high quality brands such as F stop. I have a short torso length (15") and need a women's extra-small or small size pack with a properly fitting shoulder harness and properly fitting hip belt. I use an ordinary panel-loading hiker's pack with a camera insert, and may also use the Cotton Carrier vest with it for instant camera access. 20 pounds of gear feels like "nothing" when not dealing with significant elevations.

Anxiously awaiting next weekend, to give my new Mystery Ranch Cairn 32 L (women-specific design) pack a spin with full kit in F stop insert (medium and large inserts fit). Best panel access yet.

[BJL]

DXOMark measurements are relevant only for those who shoot in RAW mode and who are concerned about achieving an ETTR exposure most of the time.

Ray, can you define EXPOSE To The Right [ETTR] and explain why — with your definition — an ETTR exposure is relevant in "low light" situations where the exposure index needed well above the "base ISO speed"? (Which by the way is a nickname for what ISO 12232 defines as SSat, the basis for DXO's "true ISO".)

I ask because my understanding of the virtue of EXPOSING To The Right is about choosing the maximizing sensor EXPOSURE level that avoids blown highlights due to over-filled photosites. That is, pushing the histogram of photo-electron counts in photosites as far to the right as possible. This is desirable if it can be done, because that maximized sensor exposure maximizes SNR levels and so minimizes visible noise in the final image.

If you are instead looking at histograms of raw levels after possibly different levels of analog gain have been applied, the connection between histogram placement and SNR levels is broken. As a hypothetical example, consider two cameras with identical sensors: camera A could record a scene with a half stop more exposure than camera B [1/2 stop lower EI], but then amplify it one stop less (half as much) so placing the raw values half a stop lower (further from the right), so camera A would be worse by some "ETTR" criterion, but so long as the amplification in both cases raises the photon shot noise above the cameras' internally produced noise, camera A would give lower visible noise levels in the final image. (There is also the distinct possibility that the greater amplification on camera B causes highlight clipping that is avoided by camera A.)
Indeed, this could be done with the same camera, by choosing different combination of f-stop, shutter speed and EI setting ("ISO dial" setting): say 1/2 stop lower actual EI and one stop lower "ISO dial" setting.

But anyway, these comparisons with different exposure levels are mostly irrelevant for low light ("high ISO speed") photography with shutter speed constraints and such. In this situation, and assuming manual selection of shutter speed and aperture, the main questions for me are:
Once a minimum shutter speed is required and a maximum aperture size imposed by factors like the limits of available lenses or a minimum DOF requirement, and the result is underexposure of the sensor,

Q1) What "ISO dial" setting is best? Or at a more basic level, what is the best amount of analog gain to apply before ADC?

Q2) With these optimal settings, how are the noise levels?

The way that DXO does its testing, the needed "equal shutter speed at equal sensor illumination" comparison for that situation is given with equal on-camera "ISO dial" settings, not equal "DXO SSat" measurement.