New Panasonic S1R - best landscape camera?

[Ancient Tiger]

The new S1R's pixel shift mode produces 187MP files and movement does not seem to be an issue. I have downloaded RAW files fro Dpreview and then converted them with Adobe DNG converter (11.2.1). I then opened them in Adobe Camera Raw and performed a file enhancement. I ended up with a gigabit file! After a little sharpening, I converted it into a TIFF. Unbelievable is the only thing I can say.

Panasonic have created a Medium format killer camera and the only thing holding it back a bit is lens selection. However, with the MC21 adapter from Sigma, you can use all Sigma EF lenses and Sigma SA lenses with full functionality.

I can only imagine what their second gen mirrorless FF camera will be like.

Well done Panasonic. This could be a landscape photographer's dream camera.

[kers]

Pixelshift seems to be the new IBIS- a feature every camera needs to have in future.
The use for landscape is limited to non moving landscapes- rocks or vegetation far away.
If not you have really ugly details, instead of more detail.
For architecture, and table-top it will be very nice.

[petermfiore]

All true...and for Photographing Art work. Amazing for paintings!

Peter

[Paul2660]

I agree, excellent solution, glad to see ACR is doing a better job on the files than it does on pixel shift Pentax files. 

Nikon should have invested some time into a similar solution with the Z7 and Z6, as their IBIS setup is excellent.

Paul C

[BernardLanguillier]

I agree, excellent solution, glad to see ACR is doing a better job on the files than it does on pixel shift Pentax files. 

Nikon should have invested some time into a similar solution with the Z7 and Z6, as their IBIS setup is excellent.

Indeed, not sure why they haven't done this yet. Considering what they are doing with eye-AF in firmware, I am pretty sure we will get the feature in the coming months.

The Z lenses are certainly up to the task also.

The Pana is an interesting addition, but IMHO it doesn’t leverage at all the compactness potential of mirrorless mounts.

Cheers,
Bernard

[shadowblade]

Pixel shift to increase resolution is near-useless for landscapes. Landscapes have grass and leaves which flutter, waves which move, low level mist and smoke which move, clouds which move with longer exposures. Cityscapes have moving vehicles, flags, people and reflections of all three.

Half-pixel shifts (the kind which increase resolution) are mostly useful for things like product photography, microscopy and reproduction, where you're dealing with subjects that don't move at all.

Full-pixel shifts, which don't give you more overall pixels but give you true RGB values for each pixel, without interpolation, are much better for landscapea and other subjects with subtle movement or small areas of movement. You can use the combined exposure for most of the image, gaining its benefits there, while painting over the moving areas with a single exposure (or blending them the traditional, layers-opacity way to emulate the motion blur of a long exposure) to eliminate artifacts. Much harder to do that with a half-pixel-shift setup, where the final image isn't the same resolution as the source images.

[Ancient Tiger]

Pixel shift to increase resolution is near-useless for landscapes. Landscapes have grass and leaves which flutter, waves which move, low level mist and smoke which move, clouds which move with longer exposures. Cityscapes have moving vehicles, flags, people and reflections of all three.

Half-pixel shifts (the kind which increase resolution) are mostly useful for things like product photography, microscopy and reproduction, where you're dealing with subjects that don't move at all.

Full-pixel shifts, which don't give you more overall pixels but give you true RGB values for each pixel, without interpolation, are much better for landscapea and other subjects with subtle movement or small areas of movement. You can use the combined exposure for most of the image, gaining its benefits there, while painting over the moving areas with a single exposure (or blending them the traditional, layers-opacity way to emulate the motion blur of a long exposure) to eliminate artifacts. Much harder to do that with a half-pixel-shift setup, where the final image isn't the same resolution as the source images.

Have a look at Dpreview,s pixel shift examples. I was amazed to see that moving objects WERE NOT blurred. People in the frame have no movement artifact. Using Adobe Camera RAW's new enhance details feature seems to remove jaggies too. It's amazing. I think most here are assuming the pixel shift is the same as before but it seems to be the next generation version and at least lowere moving objects do not seem to be an issue.

Try developing the RAW file on this page: https://www.dpreview.com/sample-galleries/9627273872/panasonic-lumix-s1r-sample-gallery/4357417550
Use Adobe DNG converter 11.2.1 to make a DNG file. Then use ACR to open it. Use the new enhance detail feature in the top left hand area. This creates a new DNG file and it should be about a gigabyte in size. Open that up and apply a bit of sharpening (I actually used Capture One to open the new file).

[shadowblade]

Have a look at Dpreview,s pixel shift examples. I was amazed to see that moving objects WERE NOT blurred. People in the frame have no movement artifact. Using Adobe Camera RAW's new enhance details feature seems to remove jaggies too. It's amazing. I think most here are assuming the pixel shift is the same as before but it seems to be the next generation version and at least lowere moving objects do not seem to be an issue.

Try developing the RAW file on this page: https://www.dpreview.com/sample-galleries/9627273872/panasonic-lumix-s1r-sample-gallery/4357417550
Use Adobe DNG converter 11.2.1 to make a DNG file. Then use ACR to open it. Use the new enhance detail feature in the top left hand area. This creates a new DNG file and it should be about a gigabyte in size. Open that up and apply a bit of sharpening (I actually used Capture One to open the new file).

If you're not getting movement artifacts, it means that the software is disregarding three out of the four exposures in that area and you aren't getting the added resolution there either. If the overall image dimensions are increased, as is the case with half-pixel shifts, it means that the software must be interpolating the extra pixels there to fill in the gaps.

It would be interesting to shoot a moving test chart to see what the software does, and what sort of resolution you're actually getting in areas which should be motion-blurred.

The other issue is that, with source images and output images of differing dimensions, you're entirely reliant on the software to get it right. If the software makes a mistake, you can't go back and blend over the affected area with a single source image, since the image would be a different size.

[Paul2660]

Pixel shift to increase resolution is near-useless for landscapes. Landscapes have grass and leaves which flutter, waves which move, low level mist and smoke which move, clouds which move with longer exposures. Cityscapes have moving vehicles, flags, people and reflections of all three.

Half-pixel shifts (the kind which increase resolution) are mostly useful for things like product photography, microscopy and reproduction, where you're dealing with subjects that don't move at all.

Full-pixel shifts, which don't give you more overall pixels but give you true RGB values for each pixel, without interpolation, are much better for landscapea and other subjects with subtle movement or small areas of movement. You can use the combined exposure for most of the image, gaining its benefits there, while painting over the moving areas with a single exposure (or blending them the traditional, layers-opacity way to emulate the motion blur of a long exposure) to eliminate artifacts. Much harder to do that with a half-pixel-shift setup, where the final image isn't the same resolution as the source images.

Near useless, I would disagree.  The images I took with the K1, handled movement very well depending on the raw converter in use.  Adobe could not handle any motion but other converters did OK. 

IMO still not a bad option to have with a modern camera. 

Paul C

[shadowblade]

Near useless, I would disagree.  The images I took with the K1, handled movement very well depending on the raw converter in use.  Adobe could not handle any motion but other converters did OK. 

IMO still not a bad option to have with a modern camera. 

Paul C

But were you getting any benefit out of it? All you're really saying there is that you didn't get motion artifact (was that only when viewed in full, or also when viewed at the pixel level?). But the flip side to that is that is that it means the software wasn't combining the four frames there either, and was just using one of them - if all four frames were used, you would have gotten motion artifact, while, if it only used one frame in motion-affected areas and just blew it up to the same resolution as the rest of the combined image, you didn't get any improvement in resolution in those areas out of using pixel-shift either. And, since you're completely reliant on the software to get it right, with no way to effectively 'paint in' a single frame in areas where the software gets it wrong, it's leaving a lot up to chance.

[Paul2660]

It worked for me. Cleaner files. More details.  What stopped me from using it was software however. Adobe did a once and done conversion which was terrible. Capture One never supported it. And the rest of the converters did not offer tools I wanted.

But everyone should try it and see if they see any gain.

K1 did it differently as you did not gain output resolution like other systems allow.

Paul C

[shadowblade]

It worked for me. Cleaner files. More details.  What stopped me from using it was software however. Adobe did a once and done conversion which was terrible. Capture One never supported it. And the rest of the converters did not offer tools I wanted.

But everyone should try it and see if they see any gain.

K1 did it differently as you did not gain output resolution like other systems allow.

Paul C

Different kind of pixel shift - it's similar to the Sony system, and actually useful for landscapes.

Pentax and Sony use full-pixel shifts, leaving overall dimensions unchanged but giving true RGB values for every pixel. This allows pixels from one of the frames to be painted over the blended image to cover any parts affected by motion blur. You gain the benefit of pixel shift for all the nonmoving parts of the image, while avoiding motion artifacts in the motion-blurred parts of the image (which, in many cases, wouldn't benefit from the increased detail of pixel shift anyway, since they're diffuse and moving, e.g. flowing water or clouds).

Other systems, such as Panasonic's, use half pixel shifts (essentially taking measurements 'in between' the normal pixels) and double the pixel count on each side compared to a single exposure (for 4x the overall image size). You can't 'paint over' motion-blurred pixels with a single exposure in this case, since the final image is four times the size of each single exposure. This makes it much more difficult to use for things like landscapes, where there are often significant elements with subtle movement (e.g. leaves, grass, clouds, water, etc.)

[Paul2660]

Thanks for the clarification.  That may explain why Nikon choose to not invest in the technology.

I was curious how the Panasonic system would work. 

I always felt Pentax was a bit lax in not pushing Adobe to make a better conversion of their system.  Other raw converters like Raw Therape, or Iridient can make good conversions, but just don't have the tools I have gotten used to with LR or C1.  Sure wish C1 had taken the time to work up a solution for the K1 pixel shift files.

Paul C

[Dan Wells]

It sounds like the Panasonic system is essentially similar to what Hasselblad has been doing for years. That has proven enormously useful for art reproduction (where the camera is on a very heavy stand which makes even a large tripod look light, weildy and convenient by comparison). As far as I know, art reproduction is the major focus of Hasselblad's MS (MultiShot) backs, perhaps along with the very high end of still life advertising photography (most ads don't need 400 MP!).

The S1R may prove a boon to local or regional museums that can't afford a $50,000 Hasselblad H6D 400c MS...

[Ancient Tiger]

I told you so....

https://www.dpreview.com/articles/3528119869/panasonic-lumix-s1-s1r-studio-scenes-with-high-res-mode-published

[Dan Wells]

This is a non-moving scene... The Hasselblad multi-shot back (which is known to not handle motion AT ALL - it's made specifically for art reproduction) would handle it just fine (with about twice the resolution of the Panasonic).

Lloyd Chambers is seeing some promising results from the Panasonic even with subject movement - I don't subscribe to his site, although I'd like to, it's too expensive. I anxiously await a result that isn't behind an expensive paywall - someone other than Lloyd has got to try shooting a landscape with this monster.

It would be fascinating to see what happened if you moved an entire test chart slightly during the multishot sequence (the test that shadowblade proposed). You'd want to move the test chart slowly enough that the shutter speed dealt with the motion within a single exposure, but the motion was significant between exposures. There are three possible results:

1.) A blurred mess, which would happen with many previous systems, including Hasselblad's.

2.) A nice ~50 MP shot (the camera detected the motion and used one of the images). In this case, the follow-up test would be to move part of the test chart, in order to find out how big an area it replaces with a single image (it could be anything from "only pixels that move, analyzing every darn pixel" on up to " the whole image reverts to 50 MP if anything moves")

3.) A shot that significantly exceeds ~50 MP (it's doing some sort of computational photography that deals with the motion).

If what it's doing is either a well thought out version of option 2 (replacing only moving parts of scenes) or option 3, it's a very interesting landscape camera. If it's giving 187 MP in 90% of a scene, with 47 MP in the areas with moving leaves or a river, that's great.  It's probably tripod-bound (well, it reverts to being a lot like a Z7 off a tripod, but a Z7 is a superb landscape camera itself), but Phase Ones are pretty much tripod-bound as well, with their size, weight and mirror slap. It could be a $4000 Phase One for many purposes.

If what it's doing is either making a mess with any motion or reverting to a single shot if there's motion, it's still useful for limited landscape  scenarios, and of course for art reproduction and similar jobs.

The other thing we need to think about is how to display 187mp? The 50mp class cameras make stunning 24x36" prints and beg for more printer. For the first time, I ran into an exhibition space that made me wish I owned a Canon Pro-4000 (the 44" version) instead of my 24" Pro-2000. My landscapes are on display in a large Unitarian church that has enormous expanses of white wall. I really could use 40x60" prints in there, and I suspect my Z7 is up to the job (I don't think my older 24 MP files will go that big). How many spaces like that are there? And, if I'm right that the 50 MP class cameras make those prints, how many spaces are there that want even larger prints?

Dan

[petermfiore]

And, if I'm right that the 50 MP class cameras make those prints, how many spaces are there that want even larger prints?

Dan

I make paintings. 48x60 and up...For almost all homes that size is huge and near impossible. Once your above 40 inches in width your audience becomes limited. Corporate and institutional spaces are the obvious choice.

Peter

[armand]

I guess it depends on what you need but as stated above the ~ 45MP cameras cover most reasonable print sizes for homes, unless you want to get a pano out of a single shot.

For me I realized after getting the Z7 that 47MP is too much for most cases, 24MP is good enough (A reason that I'm tempted to stick mostly with Fuji APS-C for general shooting, depending on how good the new 16-80 is, but this is a different discussion). Focus should start moving elsewhere, such as lens options and rendering, dynamic range, functionality, speed, etc.

[hogloff]

I make paintings. 48x60 and up...For almost all homes that size is huge and near impossible. Once your above 40 inches in width your audience becomes limited. Corporate and institutional spaces are the obvious choice.

Peter

I sell into the high end home market through home designers and I sell way more large prints than smaller prints. The trend is towards a large eye grabbing print than multiple smaller prints. One big print leaves the space uncluttered whereas numerous small ones result in a clutter look.

[sarrasani]

do you want to avoid bayer mosaic with pixel shift (this in panasonic, equal in sony, pentax, other ones)? Good but unuseful for 70-80% of (moving) images.
More simple and useful to NOT use mosaic, and to obtain in this way an evidently superior sharpness.....
Told by a lover and user of both mosaic and foveon (and more of large format film, the choice for strong men, stronger than me....), if you try foveon you'll continue to use it. Or a 10-20 times expensive MF back.
All the best,
Sandro