Does the S1 produce sharper images than the S1R?

[Chaemono]

Advertisement (gone after registration)

Hey guys, I know we all here are enamored of cameras with lots of megapixels because we either need to crop a lot/print large or we like to imagine that the higher the pixel count the sharper a picture is perceived to be, even when viewed on a computer monitor. In the former case, the S1R should be preferred over the S1 as it resolves more detail. But which of the two cameras do you think produces a crispier looking image, let’s say at ISO 250 and below? I actually don’t the answer, yet, but am asking because according to Bill Claff, the S1R pre-treats its RAW files with mandatory NR at ISO 250 and below. Sort of like LENR, I guess, except always. 😂

[Jared]

You may have trouble getting answers to your question for a couple reasons.  First, there aren’t all that many people in a position to compare the two cameras in question.  Second, it’s not at all clear what you mean by “sharper”.  At what viewing distance?  With what screen size? With sharpening optimized for that particular screen resolution and viewing distance? What lens?  With the S1R down-samples to the same pixel count as the S1 or left at native resolution?

In general, I’d be shocked if whatever smoothing is applied to the S1R were so aggressive as to completely remove the additional contrast available with the higher pixel count.  That would just be silly.  But once you get above 20 megapixels or so it becomes really challenging to actually capture all the resolution a camera is truly capable of.  For example, my SL with the 35mm Summicron SL lens shot at f/2 will routinely out-resolve my Hasselblad X1D II with a 45mm XCD lens at f/3.2 when shot at close focus.  But that’s at close focus.  And it still doesn’t have the dynamic range of the X1D.  Still, that lens (among the best Leica makes for 135 format) can, at close focus, more than make up for what the SL gives up in megapixels.  What lens would you use for your comparison?  What aperture?  Diffraction will become the dominant factor in resolution at a lower f-stop on the S1R than on the S1.  Would you be comparing at f/8?  F/5.6?  F/2?  

Personally, I am quite happy with 24 megapixels for everything I do as long as I don’t have to crop much.  The main advantage I see in the higher pixel count of the S1R (or my Q2, for that matter, which likely has the same sensor) is additional flexibility in cropping.  Even with the Hasselblad, I enjoy the camera’s output more because of the dynamic range improvements and because the camera manages to hold color saturation in very fine detail much better than the SL rather than because of any additional resolution.  Some people need every megapixel they can get.  I don’t.  They don’t really help me in my photography any more.  Cropping is really about it.

[Stuart Richardson]

I have not used the S1R, but I can say that the S1 has some of the sharpest pixel level detail I have seen on a camera. It is very good, and consequently it should enlarge very well for printing.  I would suspect, however, given how good the L mount lenses are, that the S1R will be the sharper camera. The question of sharpness will depend more on the exact sizing you print. Below A2 or so (perhaps even A1), I suspect they will be largely indistinguishable. The S1R has roughly double the megapixels...even if they are not quite so crisp as they are on the S1 (which is not something I am saying), a slight downsampling should make up for any difference. Above A1, the extra resolution in the S1R will almost certainly help minimize artefacts and increase fine detail. These days, there is a lot of signal processing one can do to increase microcontrast...the texture function in Lightroom is quite remarkable when used sparingly, and a careful blend of unsharp mask and deconvolution sharpening can make up for some slight fuzziness when used properly. A little goes a long way with both tools. In general, I would choose the camera based more on your usage pattern and expected print size. In my opinion, the S1R makes sense primarily if you like to crop (for example, shooting in a 4x5 or 4x3 ratio), or if you print rather large. If you are primarily interested in moderate enlargements, low light performance and video, the S1 is likely to be just as useful, if not moreso. 

P.S. The idea of using noise reduction on low ISO images fills me with dread. I work day in and day out as a printer, and luminance noise reduction for me is a tool of last resort. It almost never improves an image...grain is grain and softening it means softening all your fine detail. The algorithms might get better, but they still cannot adequately differentiate between detail and noise. Color noise reduction is more useful, but still has a tendency to deaden all saturated colors and kill your color depth. The best method of noise reduction is to use a tripod or a flash. Raising your ISO significantly over base ISO is always a compromise for image quality...I don't care if you are using a Sony A7SII or the latest, greatest low light beast...they are all better at base ISO. 

Edited October 5, 2019 by Stuart Richardson

[scott kirkpatrick]

22 hours ago, Chaemono said:

.... according to Bill Claff, the S1R pre-treats its RAW files with mandatory NR at ISO 250 and below. Sort of like LENR, I guess, except always. 😂

This must be a private conversation.  Could you tell us more about what measurements the comment is based on?

[scott kirkpatrick]

8 hours ago, Stuart Richardson said:

...  In my opinion, the S1R makes sense primarily if you like to crop (for example, shooting in a 4x5 or 4x3 ratio), or if you print rather large. If you are primarily interested in moderate enlargements, low light performance and video, the S1 is likely to be just as useful, if not moreso. 

Besides cropping because you don't have a long enough lens or to change format, there are always applications in which there is a lot happening in the details and resolution of the details is important.  (Bruegel-photos.) 

The most common example would be class photos, in which each parent, who is paying for the photos, wants to see their kid's face clearly.  I remember seeing this advantage clearly when I switched from my M2 to an old Rolleicord for pictures of research groups in their labs.  I was in a conference group of over 100 people recently where the photographer was using a bog-standard old Canon to get us all.  I'm not expecting much of the picture, even though I was near the front of the mob.  That used to be a job for a 4x5 on a tall tripod and still you can't tell who's who in historic Solvay Group style pictures.

[Chaemono]

vor 15 Minuten schrieb scott kirkpatrick:

This must be a private conversation.  Could you tell us more about what measurements the comment is based on?

https://www.dpreview.com/forums/post/62516666

[scott kirkpatrick]

Advertisement (gone after registration)

I have both SL and S1R, which seems like a meaningful comparison.  I would think a text-rich relatively flat image taken at 20 meters or greater distance and exposed for the midrange, would be a fair test.  Using the SL35 on both cameras.  Rendering the SL at 100% and the S1R at 70% to produce the same number of pixels.  No postprocessing.  In my case with Capture One defaults.

Any other proposals?

 

[thighslapper]

15 minutes ago, scott kirkpatrick said:

I have both SL and S1R, which seems like a meaningful comparison.  I would think a text-rich relatively flat image taken at 20 meters or greater distance and exposed for the midrange, would be a fair test.  Using the SL35 on both cameras.  Rendering the SL at 100% and the S1R at 70% to produce the same number of pixels.  No postprocessing.  In my case with Capture One defaults.

Any other proposals?

 

.... and which fool is going to attempt it ..... ? 

there are many other potentially confounding factors so the results will end up be in interpreted in whatever way suits your fancy ......

I've done all this multiple times before, and an exercise to see if a 24mpx camera is 'as sharp' as a 47mpx one strikes me as completely pointless. 

[Chaemono]

I agree with thighslapper.  Jared and Stuart make very valid points, too.  Exact focus point, whether a tripod is used, difference in DR between the two cameras, and even WB are all such confounding factors.  The Panasonic likely applies sharpening to the S1R files as well.  At ISO 100, therefore, the perceived sharpness between pictures from the two cameras will be indistinguishable.  The question is, does this change at ISO 250?

I’d say, Scott, go ahead and do it.

vor 9 Stunden schrieb Stuart Richardson:

I have not used the S1R, but I can say that the S1 has some of the sharpest pixel level detail I have seen on a camera.

Totally agree and I have both.

vor 9 Stunden schrieb Stuart Richardson:

The S1R has roughly double the megapixels...even if they are not quite so crisp as they are on the S1 (which is not something I am saying), a slight downsampling should make up for any difference.

That’s what my tests showed yesterday but I wasn’t 100 percent happy with them.  So, I went back and redid them today. I haven’t gone through the files, yet.

[nicci78]

The more pixel. The sharper the camera will be. 
I know this is crazy maths. 😉

Q2 is way sharper than Q. It’s a fact. And yes you can see it with your own eyes.
 

Of course side effects of twice the number of pixels will be crazy enlargement, massive crops or excellent down sampling. 

[Chaemono]

Obvious answer to a naive sounding question, Nicci. But what we are trying to figure out is, how does introducing the pre-treatment-with- NR-variable into the equation alter the “sharpness” result of this “crazy” math.

[scott kirkpatrick]

Incidentally, the Photonstophotos.net data that is graphed on Bill Claff's website shows very strange behavior*, with the strangest occurring between ISO 50 and what looks like ISO 80.  Claff awards those settings a double modified data point, for having both "noise reduction" and "scaling."  The noise reduction is indicated from 100 to 200.  But my S1R, with firmware 1.2, no longer offers the ISOs below 100.

*His curves run from ISO 50 to 51200 (currently I have in firmware 1.2 only 100 to 25600.  There are gain changes indicated at ISO 250 and ISO 6400, and some form of noise reduction is imposed from 100 to 200 and at 6400 and above.  6400 and above doesn't concern me.

[scott kirkpatrick]

BTW, does the S1 retain the ISOs below 100 and above 25600?  I  don't have one and I would not expect the instruction manual to show this.

[gbpost]

"The noise reduction is indicated from 100 to 200.  But my S1R, with firmware 1.2, no longer offers the ISOs below 100."

I can set my S1R with firmware 1.2  to ISO L50, L64 and L80 when I set Extended ISO to "on" in the menu. When I set Extended ISO to "off" the lowest ISO is 100. I wonder if our two S1R work differently.

[Chaemono]

A bit of an in-depth discussion about the NR/DR impact here: https://www.fredmiranda.com/forum/topic/1581322/21

What is confounding is that despite these shenanigans, the S1R’s DR doesn’t seem to be affected. 

[scott kirkpatrick]

1 hour ago, gbpost said:

"The noise reduction is indicated from 100 to 200.  But my S1R, with firmware 1.2, no longer offers the ISOs below 100."

I can set my S1R with firmware 1.2  to ISO L50, L64 and L80 when I set Extended ISO to "on" in the menu. When I set Extended ISO to "off" the lowest ISO is 100. I wonder if our two S1R work differently.

Thanks.  It's on page 334 of the manual, and "menu" --> "gear" --> "image quality"  (one of several pages with that name...) and adds in the LISO and HISO values at the ends of the range.  It doesn't seem that Auto ISO uses the L values, but I'll have to check.  I'd prefer to leave them OFF.

[Richardgb]

As an aside to the discussion, let's not forget that both the S1 and the S1R offer 4x increased resolution by shifting the sensor. Clearly it only works fully for static subject, although test reports that evidence (blurring) caused by of small amounts of movement may be overcome by built-in software.

Thus, for still-life work the S1's 96MP should be outstanding and have the benefits of larger individual pixels on the sensor; the S1R's image would be roughly double that...

[Jon Warwick]

1 hour ago, Richardgb said:

As an aside to the discussion, let's not forget that both the S1 and the S1R offer 4x increased resolution by shifting the sensor. Clearly it only works fully for static subject, although test reports that evidence (blurring) caused by of small amounts of movement may be overcome by built-in software.

Thus, for still-life work the S1's 96MP should be outstanding and have the benefits of larger individual pixels on the sensor; the S1R's image would be roughly double that...

The "high resolution mode" is indeed outstanding, and to my eyes (for large prints) makes a very noticeable difference to image quality - more resolution, no false colour, no moire etc.

Clearly tripod mounted only, it works superbly for landscapes, architecture etc.

I'm surprised more people don't talk about the "high resolution mode" here.  We pay a lot for arguably the best lenses in the world (SL Summicron primes), so why throw away a lot of the advantage away via less than perfect technique and by not using the one feature that noticeably improves image quality!!

[01af]

If you want a sharper picture then buy a better lens.

[Advertisement]

Hallo,

Look here: Leica L-Mount

[Stuart Richardson]

High resolution mode is good, but to be honest, I have seen weird artefacts and textures if you really dive in and pixel peep. I am not talking about moving objects here...more about static ones. I think whether or not those would become visible and cause problems in extremely large prints is up in the air. I know that in my testing, it was no replacement for native resolution. That said, I think that there is no question that it can help in certain situations, and the advantage is that the camera generates a regular image alongside it, so if ever it does cause an issue, you still have a non-pixel shifted version as well.