S1R vs A7CR - IQ comparison

[pdk42]

I got the chance to shoot a number of shots side by side with my S1R and a Sony A7CR. The Sony has the 60Mp BSI sensor that I think is the same/very close to the one used in the Leica Q3 and SL3, and probably what will be used in any subsequent S1R mark ii. Of course, no one will officially confirm this, but I think it's unlikely that more than one 60Mp FF sensor is made by anyone other than Sony Semiconductor Solutions. This is the same view that DPR came to in their comparison article between the A7RV and the SL3 that you can find here => https://www.dpreview.com/articles/4324053304/leica-sl3-vs-sony-a7rv?slide=2.

I only got chance to shoot at base ISO, so these are not high ISO comparisons. However, the shots were taken in very contrasty conditions, so the exposures were made to preserve highlights, necessitating a lot of shadow pushing, which is a realistic situation for landscape and similar photography types.

I'll use this shot to illustrate the points that follow. Exposure was 1/50 at f8. ISO was 100 on the S1R and 125 on the A7CR. I screwed up not using 100 on the Sony - it is as it is! It's only 1/4 stop difference so I don't think it totally invalidates the test.

1) Raw file rendered by LR using defaults and the "Adobe Neutral" profile:

S1R

• Panasonic - DC-S1R
• LUMIX S 20-60/F3.5-5.6
• 20.0 mm
• ƒ/8
• 1/50 sec
• Pattern
• Auto exposure
• -1.7
• ISO 100

A7CR:

• SONY - ILCE-7CR
• FE 20mm F1.8 G
• 20.0 mm
• ƒ/8
• 1/50 sec
• Pattern
• Auto exposure
• -1.3
• ISO 125

2) After edits to my taste (basic tone tweaking only, no masking):
S1R:

• Panasonic - DC-S1R
• LUMIX S 20-60/F3.5-5.6
• 20.0 mm
• ƒ/8
• 1/50 sec
• Pattern
• Auto exposure
• -1.7
• ISO 100

A7CR

• SONY - ILCE-7CR
• FE 20mm F1.8 G
• 20.0 mm
• ƒ/8
• 1/50 sec
• Pattern
• Auto exposure
• -1.3
• ISO 125

My observations are this:

3) Noise

There is really zero practical difference in noise between the two cameras at base ISO, even when pushing the exposure/shadows a lot. Both show some mild noise under heavy pushing (worse than on the 24Mp sensor of the S1/S5/S5ii/S9), but it's easily cleaned up and only really visible when pixel peeking. Here is a 100% crop of a dark area of the processed images when exported at 47Mp:



3) Highlight recovery

The S1R is marginally better at highlight recovery in my processed example. Now, both shots have had positive EV correction applied - almost 2 stops for the S1R and 1 stop for the A7CR. I've also pulled highlights back -100%. Despite the more aggressive exposure correction to the S1R, the Sony does worse. This could all be explained by ISO calibration and/or LR's processing of the raws. For sure, if I remove the +1EV correction on the Sony, then the hot highlights come under control, but then I can't get the overall scene brightness right using the shadow slider alone. Of course, selective EV correction (by using masks) would fix this. Bottom line for me is that the Sony seems to be brighter than ISO 100/125 difference can account for.



4) Sharpness

You can see from the above images that there is really nothing between them in terms of sharpness. The little 20-60 at 20mm is punching at the same level as the Sony G 20mm f1.8, when both are at f8.


5) Strange light bleed on A7CR
There is a strange light bleed on the A7CR images on high-contrast edges (it was present on all of them I took, not just this one). This is the edge of the bridge:



This certainly isn't right. I don't know whether it's some lens flare, some sensor "blooming", or perhaps some filter effect. It needs more investigation.


6) Conclusion

Based on these base ISO tests shooting high-contrast, high DR scenes, there is really nothing between the cameras in terms of IQ.

The S1R is really an excellent camera for landscape work. I also find the shooting experience with the S1R to be absolutely superb. Weight aside, there is little to complain about with the S1R.

 

[dirk]

Thanks for this comparison. I think this light bleed is lens flare. It is surprising that a 20-60mm zoom has less lens flare than a 20mm FFL though.

46MP vs. 60MP is not that much/nothing in the final image, depending on the size of the print and the viewing distance.

I would prefer a S1Rii with an improved 46MP sensor and better high ISO up to ISO 12800 than a 60MP sensor.

Only for cropping to replace a lens, the 60MP would make sense for me, if there would be a pixel binning feature available like in the Leica Q3.

The highlight recovery difference in the other image is not marginal in my view and there is lens flare too, by the way...

 

[o2mpx]

Thanks for publishing this. Great to see there’s not much give up in IQ from the Sony. For travel, the size of the Sony A7CR is much better than the S1r. There are some photos however that the Sony colors look overblown compared to the S1r, even when shooting raw. But it’s a trade off worth the extra effort post processing to get the reduced bulk and weight.

 

[GeorgeHudetz]

Thanks for the comparison.

As for the 20-60 comparison to the G 20mm, one of the things I have realized - after staring at far too many Lenstip resolution graphs & my own tests - is that at F8, most lenses tend to converge to a similar resolution level on a 60 MP sensor at the center - about 60 lp/mm, give-or-take. All of the "record" resolution values of 80+ lp/mm that Lenstip reports tend to happen around F2.8 to F4 (for fast primes). Which is of course of limited value when shooting landscape, unless you are willing to focus stack. I will say that zooms & slower primes do tend to fall off in the corners more than a fast prime, even at F8, however.

So it may be that you would have seen more of a resolution difference with the 60 MP sensor had you used a top-tier lens, and then kept it in the F2.8 to F4 range. But, I do think F8 is a good baseline test for landscape work and I can see why you used that aperture. So, not a criticism - just an observation.

Also, as much as I agree with you about "nothing to complain about with the S1R" at base ISO, I do think that had you elevated the ISO, you'd see some differences with the Sony. Particularly if you also strongly pushed the high ISO files. Certainly, the S1R files fall apart for astro but the Sony 60 MP sensor works quite well in that context.

 

[pdk42]

Thanks for the comparison.

As for the 20-60 comparison to the G 20mm, one of the things I have realized - after staring at far too many Lenstip resolution graphs & my own tests - is that at F8, most lenses tend to converge to a similar resolution level on a 60 MP sensor at the center - about 60 lp/mm, give-or-take. All of the "record" resolution values of 80+ lp/mm that Lenstip reports tend to happen around F2.8 to F4 (for fast primes). Which is of course of limited value when shooting landscape, unless you are willing to focus stack. I will say that zooms & slower primes do tend to fall off in the corners more than a fast prime, even at F8, however.

Of course, corner performance is always the challenge for any lens. But my little 20-60 is really excellent even in that department. I reckon it's slightly sharper than my 24-105 and 14-28 zooms, although the 70-200 f4 beats it. None of the zooms though gets close to my two Sigma primes - the 35 f2 i-Series and the 105mm f2.8 Art Macro. On the S1R, both deliver jaw-dropping detail, even wide open.

So it may be that you would have seen more of a resolution difference with the 60 MP sensor had you used a top-tier lens,

Reading Dustin Abbott's review of the Sony G 20mm f1.8, he reckons its sharpness is right up there with the GM series lenses. But it didn't seem to deliver anything more than the 20-60, at least at f8.

and then kept it in the F2.8 to F4 range.

Maybe at f4 it would have been better. I took this with it at f4 - what do you think?


Who Put that Tree There ? by Paul Kaye, on Flickr

But, I do think F8 is a good baseline test for landscape work and I can see why you used that aperture. So, not a criticism - just an observation.

Yes, f8 is the sweet spot for many FF lenses. By f11 there is a small but detectable loss of sharpness due to diffraction.

Also, as much as I agree with you about "nothing to complain about with the S1R" at base ISO, I do think that had you elevated the ISO, you'd see some differences with the Sony. Particularly if you also strongly pushed the high ISO files. Certainly, the S1R files fall apart for astro but the Sony 60 MP sensor works quite well in that context.

Yes, I think that's probably true. I try to avoid going past ISO 800 on the S1R, but I'm happy to take the S5 up to 6400 without too many qualms. But modern AI NR can do wonders!

 

[GeorgeHudetz]

Of course, corner performance is always the challenge for any lens. But my little 20-60 is really excellent even in that department.

That's the "frowny" field curvature talking. Great FG corners but weak edges. Your 24-105 should give sharper left and right edges (at the focus distance) than the 20-60 when focused at hyperfocal distances. But the 20-60 gives sharper FG corners.

Of course, there are folks who like this type of field curvature for landscape, and it's not hard to understand why. But I'll always take a flat field for "serious" landscape work.

Reading Dustin Abbott's review of the Sony G 20mm f1.8, he reckons its sharpness is right up there with the GM series lenses. But it didn't seem to deliver anything more than the 20-60, at least at f8.

Again, at F8, most lenses deliver very similar performance.

And yes, I just checked the Lenstip review, and indeed that Sony 20mm is super impressive at F2.8 and F4.

Maybe at f4 it would have been better. I took this with it at f4 - what do you think?

Not sure that one could tell much by looking at a single image.

Having said that - it's an interesting question. How obvious is the difference between 60 lp/mm and 80 lp/mm at various resolutions?

 

[pdk42]

That's the "frowny" field curvature talking. Great FG corners but weak edges. Your 24-105 should give sharper left and right edges (at the focus distance) than the 20-60 when focused at hyperfocal distances. But the 20-60 gives sharper FG corners.

Of course, there are folks who like this type of field curvature for landscape, and it's not hard to understand why. But I'll always take a flat field for "serious" landscape work.

But take a look at this shot with the 12-60, admittedly at f8, that shows its sharp right across the frame.

• Panasonic - DC-S5
• LUMIX S 20-60/F3.5-5.6
• 44.0 mm
• ƒ/8
• 1/4 sec
• Pattern
• Auto exposure
• ISO 100

The Art Shop on Regent Street by Paul Kaye, on Flickr

 

[GeorgeHudetz]

It's easy to see the field curvature in the 20-60. The typical test is to set the camera up on a tripod, and point it at a brick wall. Set the focus to MF, and then take a series of shots. The first shot should be with the plane of focus behind the wall, and then slowly pull the plane forward, shot-by-shot, until the plane of focus is in front of the wall. Do this with about 7 or 8 shots. A lens with no field curvature will have one shot where the focus is obviously better than all the others over the entirety of the wall. However, a lens with field curvature will show best focus in different shots over different areas of the wall. Best to do this wide-open, of course.

If you do this with the 20-60, you will see that two or three different images have the best sharpness over different areas of the wall. Ergo, field curvature.

I first noticed it in my backyard test-shot, which has a a very deep depth-of-field of what is basically a flat lawn. When I focus at the hyperfocal distance and compare the 20-60 to my other lenses, the 20-60 has surprisingly sharp FG corners, but the left and right edges at the hyperfocal depth are soft, unlike my other lenses. That's an indication that the 20-60 has "frowning" field curvature, where the peak-sharpness line forms a frowning curve from the focus point back to the FG corners, while a lens with no curvature gives peak sharpness in a straight line left-to-right through the focus point.

Again, it's easy to see using either of these techniques, and others have noted it as well.

As for your shot of the shop, I think that if you had taken this same shot with the 24-105, and then compared them closely, you would be able to detect the field curvature in the 20-60, particularly had you shot it wide open. But it also depends on where you focused.

 

[dirk]

But take a look at this shot with the 12-60, admittedly at f8, that shows its sharp right across the frame.

Reflecting also the other thread with the comparison of the Olympus EM1.3, Lumix S1R and Lumix S5, the EM1.3 with the 12-100/4.0 zoom would be a serious alternative for your S1R. You only need F8.0 anyway. Only dynamic range at lower ISO compared to the S5 (less to the S1R) could be in favor to the S5.

I do not think that you need 60MP. I even do not think anymore that you need 46MP (except for cropping), if you bear in mind that you can use for the rare cases high resolution multiframe shots or upscaling in Photoshop.

I do not have many walls in my house to hang very large prints above 50x70cm.

I love my S5ii, but I also love my G9 and EM1.3. I do not need more than 20/24MP. Really difficult to decide....

 

[Pete_W]

Reflecting also the other thread with the comparison of the Olympus EM1.3, Lumix S1R and Lumix S5, the EM1.3 with the 12-100/4.0 zoom would be a serious alternative for your S1R. You only need F8.0 anyway. Only dynamic range at lower ISO compared to the S5 (less to the S1R) could be in favor to the S5.

I do not think that you need 60MP. I even do not think anymore that you need 46MP (except for cropping), if you bear in mind that you can use for the rare cases high resolution multiframe shots or upscaling in Photoshop.

I do not have many walls in my house to hang very large prints above 50x70cm.

I love my S5ii, but I also love my G9 and EM1.3. I do not need more than 20/24MP. Really difficult to decide....

I was very happy with the 24MP of the original S5 and now the S5II compared with my 20MP G95.

The noticeable difference for me was much more malleable files from the S5/S5II. Shadows can be lifted significantly without penalty and ISO can be much higher without penalty, both of which are weak points for m4/3 cameras especially in high DR scenes.

The IQ from the larger sensor definitely pays off for me. I understand that macro and long telephoto with m4/3 has some benefits, or if you are needing something lighter to carry although on that latter point, full frame cameras are smaller and lighter than they were in the past and lenses are too from wide to moderate focal lengths.

 

[Markuswelder]

I was very happy with the 24MP of the original S5 and now the S5II compared with my 20MP G95.

The noticeable difference for me was much more malleable files from the S5/S5II. Shadows can be lifted significantly without penalty and ISO can be much higher without penalty, both of which are weak points for m4/3 cameras especially in high DR scenes.

The IQ from the larger sensor definitely pays off for me. I understand that macro and long telephoto with m4/3 has some benefits, or if you are needing something lighter to carry although on that latter point, full frame cameras are smaller and lighter than they were in the past and lenses are too from wide to moderate focal lengths.

My thoughts exactly. I don't really care about the high ISO performance of my S5, what I am in love with is the amount of shadow lifting you can do, with no penalty in noise, colour shifts, or other weirdness. You can view the shots at any % you like, they still look magnificent. Most of my shooting is in bright sunlight with harsh shadows. That's Australia for you. Snap a single frame, load the raw into any old raw converter app/program, and tweak to your hearts content. No expensive edition/NR programs, or supercomputer to run it all. No 100+MB files to store. It's a walk in the park for the S5. It just makes it easier and more enjoyable. For me.

 

[dirk]

I do not disagree. Like I said, the DR is better with fullframe.

But if @pdk42 uses always ISO100 for his landscape work, high ISO advantages does not matter.

Dynamic range differences depend on the image and the final output medium, on which you look at the image, assuming you did exposure correctly.

But even DR can be improved with bracketing in some cases.