Q2/S1R Dynamic range measurements out - nothing too special

[Dan Wells]

Photons to Photos has dynamic range charts out for the new sensor in the S1R and Q2... It underperforms the current best on the market (A7rIII/Z7/D850) significantly. It performs more like a Canon 5Ds R at low ISO, and worse as ISO rises. The Q2 interpretation of the sensor has a base ISO of 50, which offers more DR (although it's still close to a stop behind the Sony sensors) - current specifications suggest that the S1R won't support that (50 will be an extended ISO, which generally reduce DR). All of these cameras are within a small enough resolution range that a straight-out DR comparison is pretty fair, if not perfect.

Why are Canon bothering with their own sensors, and why are Panasonic/Leica bothering with TowerJazz (or whomever that sensor turns out to be made by)? Sony is more than happy to sell their class-leading sensors to anyone who wants one... Nikon even gets custom versions, which suggests that Canon could too, while Panasonic/Leica volume might be small enough that they can only get off the shelf models.

I haven't seen anything to suggest that these cameras are going to have really special image quality in some other way (fantastic color???)... The video specs are good, but not mind-blowing - offering very little that a Z7 or an A7r III can't do. I personally see only niche use cases (adapters for Leica lenses?) where an S1R makes sense over a Nikon or a Sony.

I saw an initial production number for the S1R in the hundreds per month in some interview somewhere (the Z7 is 10,000/month, and the Sony has to be at least that many). If that number is true, Panasonic knows they have an extreme niche product (why bother?).

[faberryman]

I haven't seen anything to suggest that these cameras are going to have really special image quality in some other way (fantastic color???)... The video specs are good, but not mind-blowing - offering very little that a Z7 or an A7r III can't do. I personally see only niche use cases (adapters for Leica lenses?) where an S1R makes sense over a Nikon or a Sony.

I haven't seen any information that would indicate that the SR1 has the special Leica microlens arrangement in the OFA that benefits manual focus M lenses.

[BJL]

I haven't seen any information that would indicate that the SR1 has the special Leica microlens arrangement in the OFA that benefits manual focus M lenses.

Do any of Leica's L mount cameras use those special "inward turned" micro-lenses? I would think that they help lenses of low exit pupil heights (like traditional M mount wide-angle lenses) but at the cost of harming edge and corner performance with lenses having high exit pupils ("highly telecentric") that send light to the edges and corners close to square-on, as with longer SLR and MILC lenses. If so, that micro-lens strategy only works with the somewhat limited focal length range of a range-finder, not with a flexible modern ILC lens system.

Back on topic: given the somewhat higher pixel count and the potential for advantages in other areas (like lenses), I would not sweat too much about this roughly half-stop difference.
But I admit I still do not fully understand the photographic significance of dynamic range (or PDR) measured at elevated ISO speed unless one is using default conversion to JPEG. What happens if one changes to a lower ISO speed setting on the camera for its higher PDR but then uses the same exposure settings (shutter speed and aperture ratio) and then compensate for this "underexposure" (actually just under-amplification, since the sensor receives the same amount of exposure) by pushing levels up a bit in conversion?

[Dan Wells]

I certainly wouldn't trade over 1.5 stops of low-ISO DR for less than two megapixels, especially seeing that everything on that chart is already above 45 megapixels. It's not just the 2/3 stop that shows up on the Photons to Photos chart at base ISO (10.82 stops for the Q2 as compared to 11.56 for the Z7)  - that's ISO 50 on a Q2, which the S1R lacks. You have to look at the ISO 100 number for the Q2 (9.98 stops) to get an estimate of the base ISO performance of the S1R (unless they enable a true ISO 50 in firmware somehow). There are few if any cameras where an extended ISO offers the best DR of any setting.

Is there any evidence that the L-mount lenses have any advantage over the Z-mount line, or over Sony G-Masters (there are some weak Sony lenses, but the G-Masters are uniformly excellent, and the L-mount lenses are as expensive as G-Masters if not pricier)? Is there any evidence that they are better than Canon's EF-R lenses, which are just waiting for a body that will show them off?

I'm just not seeing the S1R's rationale...

Admittedly, I shoot a Z7, so I'm biased (I love the light, sharp 24-70 and rugged body for long hikes and can't wait for the 14-30 to join the system), but I can certainly see why many people would prefer Sony. First of all, a lot of people had Sony before any of the other systems came out, but they also have a much more mature lens lineup. Canon also makes some sense to me - that glass is built for the future! I don't like either of Canon's current bodies for what I do (landscape, often a long way from the car), but they'll build one to go with the great lenses.

If anyone either already has a S1R from a very early shipment or is awaiting one, what drew you that way? Having seen, but not handled them, maybe I'm missing something?

[BJL]

I certainly wouldn't trade over 1.5 stops of low-ISO DR for less than two megapixels, especially seeing that everything on that chart is already above 45 megapixels. It's not just the 2/3 stop that shows up on the Photons to Photos chart at base ISO (10.82 stops for the Q2 as compared to 11.56 for the Z7)  - that's ISO 50 on a Q2, which the S1R lacks

Firstly, this has to be said yet again: the lowest "normal" Exposure Index setting (often called simply the "ISO") offered by a camera — as opposed to extended low ones — is not its saturation-based "base ISO speed" Ssat as in ISO standard 12232; nor it the same distance above that Ssat measure for all cameras and all camera makers; instead it is based on each camera maker's decision about how much highlight headroom to guarantee, and for all modern ILCs, that amount is significantly more than the amount specified by the ISO Ssat standard, because that latter standard is a guideline for a bare _minimum_ tolerable amount of highlight headroom, not the "ideal" or "correct" amount. What is more, that part of the ISO standard 12232 is from the late 1990's and was aimed mainly at the needs of default in-camera JPEG output from compact digicams with small CCD sensors of that era, so relative to modern high performance sensors, that headroom requirement is very minimal indeed!

Anyway, the PDR measure is not affected by the "normal" vs "extended" distinction: it measures directly what matters, which is the range from some acceptable SNR level (a bit above 1:1) up to saturation; if an "extended low ISO setting" has an impaired DR, the PDR measurement will see that and quantify it. So as far as I can tell, the roughly 2/3 stop difference is the valid measure.

[SrMi]

Photons to Photos has dynamic range charts out for the new sensor in the S1R and Q2... It underperforms the current best on the market (A7rIII/Z7/D850) significantly. It performs more like a Canon 5Ds R at low ISO, and worse as ISO rises. The Q2 interpretation of the sensor has a base ISO of 50, which offers more DR (although it's still close to a stop behind the Sony sensors) - current specifications suggest that the S1R won't support that (50 will be an extended ISO, which generally reduce DR). All of these cameras are within a small enough resolution range that a straight-out DR comparison is pretty fair, if not perfect.

Why are Canon bothering with their own sensors, and why are Panasonic/Leica bothering with TowerJazz (or whomever that sensor turns out to be made by)? Sony is more than happy to sell their class-leading sensors to anyone who wants one... Nikon even gets custom versions, which suggests that Canon could too, while Panasonic/Leica volume might be small enough that they can only get off the shelf models.

I haven't seen anything to suggest that these cameras are going to have really special image quality in some other way (fantastic color???)... The video specs are good, but not mind-blowing - offering very little that a Z7 or an A7r III can't do. I personally see only niche use cases (adapters for Leica lenses?) where an S1R makes sense over a Nikon or a Sony.

I saw an initial production number for the S1R in the hundreds per month in some interview somewhere (the Z7 is 10,000/month, and the Sony has to be at least that many). If that number is true, Panasonic knows they have an extreme niche product (why bother?).

Nothing indicates that S1R and Q2 have the same sensor. If you look at the raw files with something like RawDigger, you will notice that they differ slightly in the number of pixels. And as you mention, the native ISO is different in S1R and Q2.

It would be sad if all the camera manufacturers would use the same sensors. The progress would be significantly hampered even if, currently, Sony sensors may produce the best measurement results.

When I decide which of my cameras to pick, the dynamic range is definitely not a criterion.

[scooby70]

When I decide which of my cameras to pick, the dynamic range is definitely not a criterion.

Interesting.

I shoot hand held single exposures often in difficult lighting. Protecting the highlights and lifting the shadows with a relatively low DR sensor gives predictably less good results, as predictable as exposing for the shadows and not protecting the highlights will blow large areas.

Any modern camera will be better than my old Canon 20D and 5D but I would hate to go back now and I'd have a bit of a mental block to overcome if any potential new purchase significantly lagged the competition for DR.

[Dan Wells]

I've never seen a camera in Photons to Photos where an "extended" ISO setting added DR - the manufacturers tend to pick a "base" ISO setting that is the closest round number to the sensor's best overall performance. Maybe there are exceptions - I can't claim to have combed through the whole database - I've pretty much only looked at what I've shot over the years, its direct competitors, plus a few student cameras whose performance has confused me (I taught photography for years, and would often get the "why did my camera do that" question).

Usually, the answer to why a shot got screwed up was either a missed setting (scene mode on by mistake, ISO 1600 in broad daylight, shutter speed accidentally knocked to 1/8000 or 30 seconds, etc.) or occasionally a broken camera. Once in a while, I'd find a performance limit on a camera that I wasn't expecting, and sensor DR was a relatively common culprit. It was often an impossible shot - not even a D850 will handle the noonday sun in the middle of the frame, which doesn't stop students from trying "expose for the sun", despite warnings about not pointing glass objects at the sun, with interesting effects on the rest of the image 20 stops down! Occasionally (old Micro 4/3 cameras and compacts were frequent suspects), we'd run across shots that "should have worked" and worked on my camera or another student's camera, but didn't work on a particular camera.

At least for me, $3500 cameras are valuable primarily because they extend the range of images one can capture. If an expensive camera does the same things as a cheaper camera, why use the expensive one? That could be that the expensive camera is very fast (D5, IDx mk II, A9). It could be that it is especially rugged. It could be that it works well with special lenses (Leica). It could be that the expensive camera offers qualities in its files, whether resolution, DR, color or whatever, that other cameras don't (D850, Z7, A7r series, medium format). Right now, the top-end Sony sensors seem to be offering the very best in file quality, and I'd wonder why a new, expensive system aimed at image quality didn't use one (Canon is a somewhat different story because of the huge amount of great glass).

[faberryman]

Right now, the top-end Sony sensors seem to be offering the very best in file quality, and I'd wonder why a new, expensive system aimed at image quality didn't use one (Canon is a somewhat different story because of the huge amount of great glass).

Unfortunately, all the great Canon glass doesn't do anything for the second rate sensor.

[SrMi]

Interesting.

I shoot hand held single exposures often in difficult lighting. Protecting the highlights and lifting the shadows with a relatively low DR sensor gives predictably less good results, as predictable as exposing for the shadows and not protecting the highlights will blow large areas.

Any modern camera will be better than my old Canon 20D and 5D but I would hate to go back now and I'd have a bit of a mental block to overcome if any potential new purchase significantly lagged the competition for DR.

To be clear, all my cameras have good enough DR. I stopped using my 5D loooong time ago ;-). E.g., when I decide to pick between X1D or Z 7, a9 or a7rIII, best DR will not be the criteria.

[Dan Wells]

Nice fleet of cameras... All have excellent DR - aren't they all within half a stop, with the possible exception of the A9 (which offers speed instead - it's for a different shot)?

[OmerV]

With the actual Panasonic S1R:

http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205DS%20R,Leica%20Q2,Nikon%20Z%207

[faberryman]

With the actual Panasonic S1R:

http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205DS%20R,Leica%20Q2,Nikon%20Z%207

I don't see the Panasonic S1R in the graph or on the list. Are you assuming the performance of the Leica Q2 and the Panasonic S1R are the same?

[OmerV]

I don't see the Panasonic S1R in the graph or on the list. Are you assuming the performance of the Leica Q2 and the Panasonic S1R are the same?

Sorry, I failed to notice the need to use the "Link" button to get the correct URL. Try this one:

http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205DS%20R,Leica%20Q2,Nikon%20Z%207,Panasonic%20Lumix%20DC-S1R

[faberryman]

Sorry, I failed to notice the need to use the "Link" button to get the correct URL. Try this one:

http://www.photonstophotos.net/Charts/PDR.htm#Canon%20EOS%205DS%20R,Leica%20Q2,Nikon%20Z%207,Panasonic%20Lumix%20DC-S1R

Surprisingly, the Q2 looks as if it has a stop or so less dynamic range. Different sensor or different implementation?

[OmerV]

Surprisingly, the Q2 looks as if it has a stop or so less dynamic range. Different sensor or different implementation?

I would think it's the same sensor, so yeah, who knows? Maybe the DNG format Leica uses?

[SrMi]

I would think it's the same sensor, so yeah, who knows? Maybe the DNG format Leica uses?

Why would you think they have the same sensor? After all, their resolution differs slightly and PDR measurements differ as well. The raw format should not matter at all.

[OmerV]

Why would you think they have the same sensor? After all, their resolution differs slightly and PDR measurements differ as well. The raw format should not matter at all.

Both Panasonic and Leica list, respectively, the S1R and Q2 sensor as being 47.3megs. Same sensor? Well, given the years of cooperation between the two we can assume something, right?

Beyond the sensor, only Leica knows.

[SrMi]

Both Panasonic and Leica list, respectively, the S1R and Q2 sensor as being 47.3megs. Same sensor? Well, given the years of cooperation between the two we can assume something, right?

Beyond the sensor, only Leica knows.

As discussed in this thread, the unprocessed raws have different resolutions.

[OmerV]

As discussed in this thread, the unprocessed raws have different resolutions.

Actual full rez, unprocessed raw files from each camera viewed in Adobe Camera Raw measure exactly the same:
  8368 x 5584

We're getting a bit pedantic here so I'm leaving it at this.

Peace.