S1 & S1R SNR vs. Tonal Range vs. Color Sensitivity question

[Chaemono]

Advertisement (gone after registration)

DxOMark doesn't have the sensor measurements for these cameras, yet, but I suspect the SNR charts won't look good. I compared the S1 to the SL and the SL is about one stop brighter. Another user measured the Q2 vs. the Q1 and found that the Q1 is about 2/3 to one stop brighter. The Q2 sensor has the exact same effective number of pixels as the S1R, 8368 x 5584, so it's probably the same sensor. And we know that the SL uses the Q1 sensor design modified for use with M lenses. As result, one has to bump up the ISO on the S1 and the S1R by about one stop to get pretty much the same exposure as with the SL at a given shutter speed and aperture. Bumping up the ISO leads to more noise. It's safe to assume, therefore, that the DxOMark SNR Panasonic graphs will look relatively poor, particularly for the S1R. What I also noticed with the S1, though, is that it shows nice tonal range, i.e. smooth transitions from darker to lighter shades, and good color sensitivity, i.e. it captures distinct color hues well, also at high ISO like 12800 (basically ISO 6400 on the SL). DxOMark measures these effects, too. See here for the α7R III, for example: https://www.dxomark.com/Cameras/Sony/A7R-III---Measurements. I suspect there the Panasonic/Q2 sensors will come out looking really good.

Here is my question. Shouldn't more noise degrade tonal range and color sensitivity as well? But this is not what I'm seeing with the S1, at least. If I take an ISO 12800 file and I apply NR in LR, then the result will be really usable in terms of nice tones and rich, smooth colors. They won't look washed out. Again, remember this is more like ISO 6400 on the SL, but the colors and the tones start to degrade in the SL file at this point whereas they still hold up well for the S1. Maybe Panasonic uses some really clever algorithms to ensure this is the case (I suspect that's what Sony does to show such great high ISO noise and color performance for a FF 43 MPx sensor). 

BTW, the S1 files are very malleable, not only in terms of lifting Shadows and Exposure but also in terms of bringing back details in clipped Highlight areas without creating an overly artificial effect. Still, the DxOMark SNR charts will look horrible because of the ISO measurement approach that Panasonic took (I suspect for marketing reasons). See Bill Claff's noise chart for the Q2 (same sensor as in the S1R), for example, here: http://www.photonstophotos.net/Charts/RN_ADU.htm (choose Q1 and Q2 to ensure same lens). Panasonic will have a lot of explaining to do because these sensors aren't as bad as they appear. Just look at the discussion about the Q2 ISO measurement. 

Edited March 24, 2019 by Chaemono

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[ron777]

The suggested chart renders interesting results:

Between ISO 200-800 the SL shows less noise, but from thereon the chart suggests that the noise level for the S1R, SL and Q2 are roughly the same, or slightly greater for the Q2.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited March 24, 2019 by ron777

[Chaemono]

The S1 will be even better in terms of noise. Using one stop higher ISO to get 'equal' exposure to the SL for a given shutter speed and aperture can have some advantages in certain situations, I discover. As long as the noise levels are acceptable, which they are for the S1, and Highlight detail can be recovered in post, which it can in the S1 files, the sensor will capture whatever available light is there 'better' without increasing the overall exposure of an image, if this makes sense. Look at the two in the link below. I kind of like the effect of the S1 picture better at ISO 3200 vs. ISO 1600 for the SL. These are processed to match. I'm not sure if I should make the RAWs available.  

https://www.smugmug.com/gallery/n-SZrJhT/

[ron777]

I agree, the S1 image at ISO 3200 is the more appealing of the two.

I am anxiously awaiting the April 4th USA release of the S1R, as the increased resolution—S1R as opposed to the S1—should accommodate the kind of work that I do.

Edited March 24, 2019 by ron777

[jrp]

The S1R is noisier than the S1, although if you downsample to S1 size the results are probably comparable. 

[ron777]

Hi,

I couldn't find the S1 to display on that chart, only the S1R which is depicted above. However, given that the S1R and the Q2 probably employ the same sensor, the S1R is showing somewhat less noise, possibly related to in camera processing. That said, I added the Sony A7RIII (which I own) to the chart and it turns out that it's the best of the bunch in terms of noise.

.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

[thighslapper]

Advertisement (gone after registration)

Well I really don’t care.

With IBIS I personally can never envisage me using even iso 1600 except for astrophotography

Just spent 12hrs out today with the S1R, 16-35 and 75/2 and even with night and house/museum indoor shots my maximum iso was a single at 1250, a handful at 640 and the rest below 400 (out of 300 images). Half second exposures with the 16-35 are a doddle.

Colour rendition and auto wb seems spot on ...... and with most of the auto power save functions on I’ve only used half a battery .... which is just as well as there is no sign of any spares ..... So far I’m pleased and impressed with the image quality. 

Edited March 24, 2019 by thighslapper

[ron777]

Glad to hear that you're pleased.  

The Panasonic S1/S1R will not be released in the USA until April 4th, but when it does ship the seller has promised to include an extra battery and battery grip.  As for noise, I'm not the least bit concerned either.

Edited March 24, 2019 by ron777

[Chaemono]

I find it interesting that when the DxOMark noise graphs come out for these Panasonic sensors they will show mediocre performance. And people will keep referring to them like some do with the Q2 now. That's just the way the ISO works with these sensors. Their real strength lies in how malleable the files are and it's not just recovery of Shadow details, it's bringing back Highlight detail without creating an overly artificial look. I took a similar one with the α7R III but the focus was off. So, I have to redo the comparisons, possibly throw in the Z7 as well. But I probably won't get to them in a while because I'm also traveling. What I've seen so far, with the S1 sensor the Highlights slider in LR can be pulled back to -100 to bring back details with little regrets. I'm seeing some moire then on the jacket in this one here, though. Still, the S1 files look impressive.

S1 RAW file: https://cc2032.zenfolio.com/img/g612080077-o750076470.dat?dl=2&tk=sJ2NKhuvOVVTX5iVust-ux9r-xHDZSi6wrI-DTSLKWo=

S1 + 50 Noctilux, AWB by the camera, LR default settings.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

ISO 100 f/2.8 @1/80 sec.

Edited March 24, 2019 by Chaemono

[Stuart Richardson]

Personally, I have never found much value in DxO measurements. The way I see them used seems to mostly be for bragging rights among amateur photographers who want to say their Sony is better than your Nikon etc. As far as I can tell, Leica cameras have never fared well with DxO, and I have consistently found them to have the most pleasing image quality among the manufacturers. I am sure there are real, valuable measurements that they are doing, but I think this kind of measurement is less useful than it might appear. After lens quality and handling characteristics, the best judge is your perception of the results. As you noted, if the Panasonic looks better but measures worse, I am much more inclined to trust my eyes than trust the numbers. For example, in my work as a printer, I see a lot of different files from different kinds of cameras. I am often surprised by Canon...the files just look...bad. I do not like the color, the way of rendering detail at 100%, and just the way the files look in general. Sony, Nikon, and Fuji seem better to varying degrees. The only camera that I have seen that looks significantly better to my Leica S006 is the Phase One IQ3 100, I am sure there are others in the MF field as well, but I have not seen them at the studio. The GFX was better in some ways, worse in others, for example. 

[scott kirkpatrick]

The noise measurements from photonstophotos.com are only part of the story, because the black point can change with ISO.  The full dynamic range data still hasn't appeared there yet.

So if someone is willing to put in an hour or so following Bill Claff's instruction, and get their hands on one of the April 4 new models, please give him a full data set to work with.  His data is much more meaningful than DxO's, because he can tell  you what it means, while DxO produces aggregated data with little explanation.

[SrMi]

On 3/24/2019 at 7:54 PM, ron777 said:

Hi,

I couldn't find the S1 to display on that chart, only the S1R which is depicted above. However, given that the S1R and the Q2 probably employ the same sensor, the S1R is showing somewhat less noise, possibly related to in camera processing. That said, I added the Sony A7RIII (which I own) to the chart and it turns out that it's the best of the bunch in terms of noise.

.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

The caption for this graph says:

"These raw values are not appropriate for comparing camera models because they are not adjusted for gain or area."

[ron777]

Missed that disclaimer.  

And with that in mind, the chart is of limited value. That said, I have rarely placed any emphasis on those so-called studies, as i put personal experience ahead of the pundit's pontifications.

[SrMi]

1 hour ago, ron777 said:

Missed that disclaimer.  

And with that in mind, the chart is of limited value. That said, I have rarely placed any emphasis on those so-called studies, as i put personal experience ahead of the pundit's pontifications.

You are not the first one to miss it:-). All the rumor sites misuse that chart to compare the cameras, as those charts are the first ones to be generated for a new camera.

The main chart to be used for comparisons is PDR vs ISO: http://www.photonstophotos.net/Charts/PDR.htm. Occasionally it also has only limited value.

In "Read Noise... " chart you can compare the forms of the curves, but not the values. Still very useful, I think

 

[ron777]

Thanks, I'll have a look, although, as I've mentioned before, I'm more concerned with how a given camera/lens works for me.

[SrMi]

The PDR vs ISO chart for S1R is live on Photonstophotos.net:

http://www.photonstophotos.net/Charts/PDR.htm#Leica Q2,Panasonic Lumix DC-S1R,Sony ILCE-7M3

 

[thighslapper]

8 hours ago, SrMi said:

The PDR vs ISO chart for S1R is live on Photonstophotos.net:

http://www.photonstophotos.net/Charts/PDR.htm#Leica Q2,Panasonic Lumix DC-S1R,Sony ILCE-7M3

 

Well ...... plotting comparisons with all the current FF mirrorless with high pixel counts (Sony, Canon, Nikon, Panasonic) shows a selection of lines that are so close to each other that I doubt there would be any real world difference in performance.... even with the usual caveats about graphs not telling you everthing ....

The S1R and Q2 sensors seem to behave differently ...... but whether this means the sensors are completely different (despite the identical pixel dimensions) or just that the signal is processed differently I don't know. If the Q2 sensor appears in the SL2 with this performance I suspect there will be a few grumbles ....

Edited April 1, 2019 by thighslapper