leica sl2 Jono Slack: Leica SL2 Review

[rsmphoto]

Advertisement (gone after registration)

1 hour ago, thighslapper said:

From my S1R experience Multi Shot High Def. mode is more of a novelty than a useful feature.

Although 180+ mpx images sounds wonderful the increased resolution does not appear to be as good as you would expect from a 16700x1100 sized sensor ....... and there are issues with subject movement (although the Lumix has a mode that mitigates some of this). 

I've messed about with it and even with a 5k iMac it's difficult to appreciate the difference, let alone print large enough to gain much benefit from it. As Paul mentions above, we are well into the territory of technical specifications that are well beyond the needs of most of us. 

Interesting.  It's a given that motion is an issue with multi-shot, and this is indeed territory not necessary for most people's needs, so we'll just have to see how they do implementing this when and if it comes. As I suspect you know, Hasselblad has been at this for quite some time now, their multi-shot is impressive and does have it's useful applications.

Edited November 11, 2019 by rsmphoto

[jonoslack]

1 hour ago, scott kirkpatrick said:

While it is not trivial, it is comparable to the effort to produce a 24 MP jpeg, which also has to create a virtual lattice of pixels for the JPEG image, with a spacing bigger than that of the sensor.  But jpegs have the advantage that all three colors are available at each "pixel" but with only 8 bits for each.  A 24 bit raw has to figure out where to put the 3x14 bits of color information without causing artifacts. 

Thank you Scott 

I'm beginning to realise why nobody offers this option! 

[Advertisement]

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

[jonoslack]

1 hour ago, LocalHero1953 said:

Every time I think I could be tempted by IBIS and an improved interface and other performance, I remember all those pixels I really do not need or want. I wonder how many sales are lost for this reason? I know you extol the virtues of Leica's 3-in-1 approach to the SL2 compared to Panasonic's 3 models, but a 24 mp SL2 would get my attention, even if a 24mp down-sampling option is technically too difficult at the moment.

Hi Paul

It's a difficult conundrum - nothing to do with the file size (a pain in the ass maybe). The problem is the processing time involved with a wedding or an event. 

But to be fair, I don't think there are going to be that many lost sales (dammit, I was certain I didn't want one, but I do!). There aren't so many people I've spoken to who really object - and lots of people seem really keen.

All the best

[jaapv]

1 hour ago, Jeff S said:

Puts (first ‘post-Leica’ post) apparently isn’t impressed with the practice either (English version at end)...

https://photo.imx.nl/blog/files/940fc0b7473e258690189fcc1c4d68ff-144.html

If course he couldn’t avoid another dig at Leica, and the SL2 in particular.

Jeff

I cannot say that I am overly impressed by the test images he based his conclusion on. As the structures on the 20 MP image are not beyond the resolution of the sensor, it is to be expected that the 50 MP image won't show any difference. It would have been better to use a Siemens Star or similar, which shows the precise loss of resolution as structures get smaller.

[jonoslack]

45 minutes ago, jaapv said:

I cannot say that I am overly impressed by the test images he based his conclusion on. As the structures on the 20 MP image are not beyond the resolution of the sensor, it is to be expected that the 50 MP image won't show any difference. It would have been better to use a Siemens Star or similar, which shows the precise loss of resolution as structures get smaller.

It seems to be in German

which, inexcusably I cannot read, and I don't think that Google translate is up to Erwin's subtleties. So I'll have to give it a miss

Oh damn. It's at the bottom!

Edited November 11, 2019 by jonoslack

[IkarusJohn]

6 minutes ago, jonoslack said:

It seems to be in German

which, inexcusably I cannot read, and I don't think that Google translate is up to Erwin's subtleties. So I'll have to give it a miss

Scroll down

[scott kirkpatrick]

Advertisement (gone after registration)

8 minutes ago, jonoslack said:

It seems to be in German

Dutch.

[Chaemono]

vor 8 Minuten schrieb IkarusJohn:

Scroll down

Don’t spoil it.  My phlegm was just starting to clear. 😁

[jaapv]

20 minutes ago, jonoslack said:

It seems to be in German

which, inexcusably I cannot read, and I don't think that Google translate is up to Erwin's subtleties. So I'll have to give it a miss

Oh damn. It's at the bottom!

And it is Dutch...😡

[IkarusJohn]

I found the article interesting and informative.  If I understand his conclusions correctly, "pixel shift" on the SL2 involves moving the sensor 1/2 a pixel size in 4 different directions to increase the light captured by each photo site, thereby improving colour rendition (a flaw inherent in the Bayer colour filter array); the end result is a reduction in noise, improvement in colour (and dynamic range?), but no change in resolution.

Presumably the resulting increase in file size reflects the increased quality of each photo site capture, rather than any improvement in resolution ... Doesn't sound that useful, unless you want to take really splendid landscape images and you have the patience to deal with huge files.  I think I'll stick with my X1DII 50c for landscape!

1 hour ago, rsmphoto said:

As I suspect you know, Hasselblad has been at this for quite some time now, their multi-shot is impressive and does have it's useful applications.

I assume this is on the H series cameras?  I haven't seen that function on the X1DII.

[rsmphoto]

Indeed. Currently H6D-400c, but started way back with H3D-39 Multi-shot

https://www.hasselblad.com/h6d-multishot/

[jonoslack]

29 minutes ago, jaapv said:

And it is Dutch...😡

Sorry Jaap - I didn't look very carefully - but I did read the English carefully. I spend a lot of time in Holland and I'm always impressed by the direct and intelligent approach. 

[jaapv]

Don't take my smilies too seriously

[caissa]

I am interested in knowing how the algorithm in the SL2 (And probably also S1/R) really works. How the 8 pictures are actually shifted. And the transformations used when joining the images.

He describes that he uses another camera. So simply said his conclusions are worthless, because we do not know how similar the two algorithms are. I am really disappointed if he takes that to criticize the Leica SL2. In German “Knapp daneben ist auch vorbei.”    (A miss is as good as a mile.)

Edited November 11, 2019 by caissa

[scott kirkpatrick]

16 minutes ago, caissa said:

I am interested in knowing how the algorithm in the SL2 (And probably also S1/R) really works. How the 8 pictures are actually shifted. And the transformations used when joining the images.

He describes that he uses another camera. So simply said his conclusions are worthless, because we do not know how similar the two algorithms are. I am really disappointed if he takes that to criticize the Leica SL2. In German “Knapp daneben ist auch vorbei.”    (A miss is as good as a mile.)

Puts took out an Olympus F and shot a rather sloppy example, concluding that the high res gave just as bad a result.  The whole thing should be an embarrassment to him.  He further extends his conclusions to say that this won't work for anyone else.  I think we should just ignore it.  I have used the Olympus multi-shot high res in the past, as have others.  There were some comparisons on the GetDPI site a few years back.  It works, but is no panacea.  On the Panasonic S1R, the verdict is the same.  It works.

I can imagine how the data is gathered and how it is combined, but I have never found any publication in the enthusiast literature describing precisely the process  used by a particular camera.  I have written up my theory and put it in this forum somewhere a few days ago.  Gathering the data is simple.  The hard part is translating it back into a virtual pixel array with more data points and disguising it as a Bayer raw data file that can be processed with normal software.  Opportunities to introduce unwanted artifacts abound, even if everything in the frame holds still.

[Edax]

36 minutes ago, scott kirkpatrick said:

I can imagine how the data is gathered and how it is combined, but I have never found any publication in the enthusiast literature describing precisely the process  used by a particular camera.  I have written up my theory and put it in this forum somewhere a few days ago.  Gathering the data is simple.  The hard part is translating it back into a virtual pixel array with more data points and disguising it as a Bayer raw data file that can be processed with normal software.  Opportunities to introduce unwanted artifacts abound, even if everything in the frame holds still.

 

The way it is most probably done is by outputting a so-called linear .dng (or other format raw) file, e.g. a raw file that has the full RGB data per pixel, and does not have to be de-mosaiced (btw the same type of .dng file is made during the Lightroom /  Photoshop "Enhance Details" procedure).

[IkarusJohn]

54 minutes ago, scott kirkpatrick said:

Puts took out an Olympus F and shot a rather sloppy example, concluding that the high res gave just as bad a result.  The whole thing should be an embarrassment to him.  He further extends his conclusions to say that this won't work for anyone else.  I think we should just ignore it.  I have used the Olympus multi-shot high res in the past, as have others.  There were some comparisons on the GetDPI site a few years back.  It works, but is no panacea.  On the Panasonic S1R, the verdict is the same.  It works.

I can imagine how the data is gathered and how it is combined, but I have never found any publication in the enthusiast literature describing precisely the process  used by a particular camera.  I have written up my theory and put it in this forum somewhere a few days ago.  Gathering the data is simple.  The hard part is translating it back into a virtual pixel array with more data points and disguising it as a Bayer raw data file that can be processed with normal software.  Opportunities to introduce unwanted artifacts abound, even if everything in the frame holds still.

Hi Scott,

Leaving aside the quality of Erwin's example images (which I have no view on one way or the other), aren't you saying the same thing?

Cheers
John

[NRKstudio]

Hi @jonoslack and the rest of the forum gurus: would you say the AFC tracking is a step up from the Q2 tracking in AFC?  Whether it’s Due in part from the Maestro III processor or really any other upgrades... 

I am asking as I currently shoot with an m10-p and when photographing family events or running toddlers, I grab my Sony A7r4.  But now I have a SL2 pre-order and I am concerned the AF-C and face/eye detect may not be sufficiently accurate.  I saw a review that said AF-C, when tracking a face approaching the camera, exhibited a low percentage of properly focused shots.  

My Q2 had a pretty decent hit rate in AF-C with tracking engaged, so if the SL2 is a step forward, my concerns may be very well unfounded. 

thank you for the help! 

 

[scott kirkpatrick]

5 hours ago, Edax said:

 

The way it is most probably done is by outputting a so-called linear .dng (or other format raw) file, e.g. a raw file that has the full RGB data per pixel, and does not have to be de-mosaiced (btw the same type of .dng file is made during the Lightroom /  Photoshop "Enhance Details" procedure).

I guess I have to look at some DNGs and see if that is what they chose to do.  Which means I have to upgrade my Adobe DNG verifier to Catalina.  There is still the problem that aggregated pixels don't form a nice square lattice.  The files produced (at least on the S1R) are 4X the size of the original raw files.  Linear DNG for 4X as many pixels would be even larger by a factor of three.

Edited November 12, 2019 by scott kirkpatrick

[Advertisement]

Hallo,

Look here: Leica L-Mount

[scott kirkpatrick]

5 hours ago, IkarusJohn said:

Hi Scott,

Leaving aside the quality of Erwin's example images (which I have no view on one way or the other), aren't you saying the same thing?

Cheers
John

Not quite.  I have always seen some enhanced resolution in multishot, but not 4X, and not without doing a little sharpening to overcome the blurry pixels.