Maximizing IQ: In-Camera Pixel Shift or Software Upscaling?

The 20MP cameras with pixel shift won't quite hold up to my 100MP X2D but then for 1/8 the price I can live with it. :-D
I can live with that as well :-) particularly as I prefer 4:3 ratio.

How do you find tonal gradations subtle transitions of Hassy X2D compared to OM1 Tripod Pixel Shift HR and OM1 Handheld Stacked HHHR in landscapes.
2f1d0e9f805540e59f08ea02f5e36b5e.jpg

For landscapes I use the OM1 (or OM3) with HR and HHHR more than my 50MP A1, 60MP SL3, 60MP M11 or 100MP X2D.
That's quite revealing your choice I'm now looking at OM1 as a possible future purchase.

In which types of scenarios would you choose X2D for landscapes.

As someone linked to above Handheld Stacked HHHR is around 32MP, whereas Pixel Shift Tripod HR is under 60MP probably around 50MP. Also other things need to fall in place for a Tripod Pixel Shift HR to be a successful shot.

--
Photography after all is interplay of light alongside perspective.
 
Last edited:
The 20MP cameras with pixel shift won't quite hold up to my 100MP X2D but then for 1/8 the price I can live with it. :-D
I can live with that as well :-) particularly as I prefer 4:3 ratio.

How do you find tonal gradations subtle transitions of Hassy X2D compared to OM1 Tripod Pixel Shift HR and OM1 Handheld Stacked HHHR in landscapes.
2f1d0e9f805540e59f08ea02f5e36b5e.jpg

For landscapes I use the OM1 (or OM3) with HR and HHHR more than my 50MP A1, 60MP SL3, 60MP M11 or 100MP X2D.
That's quite revealing your choice I'm now looking at OM1 as a possible future purchase.

In which types of scenarios would you choose X2D for landscapes.

As someone linked to above Handheld Stacked HHHR is around 32MP, whereas Pixel Shift Tripod HR is under 60MP probably around 50MP. Also other things need to fall in place for a Tripod Pixel Shift HR to be a successful shot.

--
Photography after all is interplay of light alongside perspective.
You want an OM1 mk ii or OM3 because of the mk I bug that forces you to use UniWB to get the full benefits of HHHR.

As far as OM vs A7CR, my answer would be that it depends on your range of use cases which is best. Within FE mount, the A1 ii does everything well, apart from weight and cost.

Andrew

--
Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post
 
As far as OM vs A7CR, my answer would be that it depends on your range of use cases which is best. Within FE mount, the A1 ii does everything well, apart from weight and cost.

Andrew.
For just landscapes would be less faff for me than M4/3 a 61MP A7R4 or A7CR which I could readily crop to 4:3 still get around 50MP.

Or Just landscapes a Sigma Merrill with warranty from photographic retailer, which you have photographed landscapes with even though it is quite limited (I had original DP1 in 2010-2012 for landscapes). On Sigma forum someone recently wrote they preferred Sigma to their Hasselblad P45 for their use case.

--
Photography after all is interplay of light alongside perspective.
 
Last edited:
As far as OM vs A7CR, my answer would be that it depends on your range of use cases which is best. Within FE mount, the A1 ii does everything well, apart from weight and cost.

Andrew.
For just landscapes would be less faff for me than m4/3 a 61MP A7R4 or A7CR which I could readily crop to 4:3 still get around 50MP.

Or Just landscapes a Sigma Merrill DP1 with warranty from photographic retailer, which you have photographed landscapes with. On Sigma forum recently someone wrote they preferred Sigma to their Hasselblad P45 for their use case.
 
ee399a58b64249cc8360329650ab509b.jpg

That said, I rarely carry a "high MP" camera anymore since the advent of HR and HHHR in small/affordable bodies. The output is fantastic straight out of camera and I don't have to sit and fiddle with computers. I feel time "processing" images is time wasted.
LOL, looks like i dont need the X2D afterall, because you made it look like the OM1 HR 80MP offers better IQ.

Surely you didnt use different FOV/focal lengths or made it so the X2D is at any disadvantage, like not add any post sharpening, unlike the OM1 where sharpening was turned to 11. Like you couldnt use HR on the A1 or apply intense software enhancement on the FF and MF body too...

Honestly, what is the purpose this comparison? Its irrelevant to anyone considering A1 or X2D and it will just confuse people who might need hi-resolution but cannt use HH/HR
 
Last edited:
Or Just landscapes a Sigma Merrill DP1 with warranty from photographic retailer, which you have photographed landscapes with.
The DP Merrills have a certain charm, but I could get better results in most cases with an A7R2. It's complicated because highlights render differently with a Foveon sensor and the images overall have a different look.
My OG DP1 I had 2010-2012 was my favourite landscape camera for types of landscapes I'm drawn to. A7R2 42MP been looking out for one with cracked screen and or some other issue around £200-£250. They are around £600 with warranty.
When people say X is all you need, they really mean X is all they need.

Andrew
Lol. There is that. I have photographed certain landscapes with 1/2.5" 1/1.8" cameras released 2004 2005 amongst my favourites.
 
These features (in-camera) are certainly fun to use and the results are quite nice. There's really no reason not to use HHHR if you are shooting still life or landscapes. Worst case, the HHHR file isn't helpful and you still have the single 20MP image to enjoy.

ee399a58b64249cc8360329650ab509b.jpg

Any camera will benefit from a solid/sturdy shooting platform. My 100MP X2D benefits from a tripod even at higher shutter speeds. Of course a tripod is a huge inconvenience at times and for what I am using an image for, handheld is more than capable.

That said, I rarely carry a "high MP" camera anymore since the advent of HR and HHHR in small/affordable bodies. The output is fantastic straight out of camera and I don't have to sit and fiddle with computers. I feel time "processing" images is time wasted.

Here is an example of how I use HR and HHHR: High-Res Sample Gallery
LOL, looks like i dont need the X2D afterall, because you made it look like the OM1 HR 80MP offers better IQ.

Surely you didnt use different FOV/focal lengths or made it so the X2D is at any disadvantage, like not add any post sharpening, unlike the OM1 where sharpening was turned to 11...

Honestly, what is the purpose this comparison? Its irrelevant to anyone considering A1 or X2D and it will just confuse people who might need hi-resolution but cannt use HH/HR
It's all physics and processing. Have a read of the paper Interceptor121 posted. Fast sensor readout enables super resolution and tools like Parsek maximise information recovery.

The differentiator is lenses, although shooting at typical landscape DoF is a great leveller.

A

--
Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post
 
Or Just landscapes a Sigma Merrill DP1 with warranty from photographic retailer, which you have photographed landscapes with.
The DP Merrills have a certain charm, but I could get better results in most cases with an A7R2. It's complicated because highlights render differently with a Foveon sensor and the images overall have a different look.
My OG DP1 I had 2010-2012 was my favourite landscape camera for types of landscapes I'm drawn to. A7R2 42MP been looking out for one with cracked screen and or some other issue around £200-£250. They are around £600 with warranty.
The price of used A7R2s has gone up over the last 2 years. They used to be a real value proposition.

Still got my DP1M and DP3M, bought on Ebay about when you had yours.
When people say X is all you need, they really mean X is all they need.

Andrew
Lol. There is that. I have photographed certain landscapes with 1/2.5" 1/1.8" cameras released 2004 2005 amongst my favourites.
 
I have not performed exhaustive testing. However, I have shot HHHRES and THRES, and shot HHHRES off a tripod. I can say with certainty that I do get more detail with HHHRES than the standard 20 MP ORF mage, and it's better from a tripod, too. Obviously focus is critical, along with minimization of atmospheric distortions (mirage, wind) and movement in the photo. I use multi-row panos to get the resolution I need, and a 3x3 image with 50 MP HHHFES off of a tripod I get 20K horizontal pixels when I crop slightly, giving me easy 40" x 60" images at 300 dpi. Yes, landscapes in still conditions off of a tripod. I will often carry another body with the Oly 12-40/2.8 (or mount that on my E.M1.III) and use it to determine my desired field of view/focal length and then shoot at twice that focal length (or close to that) with a prime or the 40-150/2.8 Pro set to the specified focal length, and stop down to f/5.6 to f/8 at ISO 200. This gives me a very sharp high res image at large sizes.

I know, the camera is making up pixels, but I find it does a better job than any external software. Honestly, the next step up would be a 100 MP camera... and I can't justify the 5-figure cost of canera and lenses.
Someone has actually tested the pixel shift resolution increase on the OM-1

https://www.optyczne.pl/477.4-Test_aparatu-OM_System_OM-1_Rozdzielczość.html

In tripod mode it produces a resolution inferior to a current 60 megapixels full frame camera

in handheld mode it falls below a 33 megapixel full frame camera

the maximum resolution increase is around 1.6x as if the camera was circa 50 megapixels and in handheld around 38 megapixels if you consider the improvement on its own resolution those are not comparable to full frame resolution because the yield per pixel drops

In conclusion you get a current high resolution camera from sony like the a7cr and this with its 60 megapixels will look better than your tripod hr shot and you can take any shot you like plus it also has pixel shift

The photo will also have more dynamic range off camera too and you will not have restrictions of multi pixel shooting

There are many useful features and benefit of a cropped format but resolution is not one
It's nice to see actual resolution tests from optzczne, thanks for this. It also supports increased resolution in both HHHR and HR. Interesting the OM-1 (single 20MP mode) outperformed the XT4 and a6600 in resolution despite those cameras having more MP.

According to their results, HHHR improved resolution by 38%. HR (pixel shift) improved resolution 63%.

422968426d6b4a079a564668cf331247.jpg

Compared to something like the 60MP Sony bodies, the MTF50 are definitely less but not at all bad when you consider these values are achievable in a $1000 OM5 size body.

A7CR: 5200 lw/ph

OM5 HR: 4400 lw/ph

OM5 HHHR: 3700 lw/ph


A1mkII: 3600 lw/ph

R5II: 3100 lw/ph

A7IV: 3100 lw/ph

OM5 Single: 2900 lw/ph

R6II: 2600 lw/ph

R7: 1800 lw/ph

Of course a lot of this depends on the lens used. For example the A7CR was tested with the 50mm F1.4 GM whereas the A1II was tested with the new 24-70 F2.8 GM II. The R7 was tested with a craptacular STM lens which might explain the low values.
The LW/PH measures are using jpegs and OM systems jpegs are heavily sharpened same applies to the HR shots processed with their software

The results for the Hi Res Shot mode are naturally higher, and quite a bit. Since we used the OM Workspace app, it is worth taking a closer look at the issue of sharpening. Unfortunately, this is applied quite generously, despite setting the appropriate parameter to the lowest available value. This is proven by the graphs from Imatest, which we present below (for the 50 and 80 Mpix modes, respectively).

Eliminate that effect and you are comparing to full frame camera far away from the 80 and 50 megapixels of the shots and remember those sony cameras at 50 60 megapixels also have pixel shift but present a more modest resolution increase of 13-15% which confirms this feature is not really going to change the life of anyone

If you want a high resolution camera you need to buy one and you will have that benefit in any shots you take including moving ones.
Every JPG I use has sharpening so I don't see how I could eliminate that effect. It is always part of the end product. The other cameras are using JPG's too for these tests.
But as you can see when they turn the sharpening to the lowest setting which is their default there is nothing less

OM system like olympus has aggressive sharpening even at the lowest setting as shown by the MTF curves
I have no issue with Oly JPG sharpening in SR, HR or HHHR as it is what I use most - straight out of camera. Really nice balance of sharpness and noise in my opinion. Every camera maker bakes in some sharpening with JPG. The HR and HHHR raw files have little to no sharpening and there are many threads in these forums about how best to approach sharpening them.

Optzczne cites an 11% increase in Sony's 4-shot method and a 33% increase in the 16 shot method so even high res sensors will benefit from pixel shift. The problem with Sony's implementation is it can't do it in camera, which is a deal breaker for me. Panasonic and Olympus have a huge advantage here although Panasonic has some other quirky limitations.

Back to my shots above. The 20MP cameras with pixel shift won't quite hold up to my 100MP X2D but then for 1/8 the price I can live with it. :-D

2f1d0e9f805540e59f08ea02f5e36b5e.jpg

For landscapes I use the OM1 (or OM3) with HR and HHHR more than my 50MP A1, 60MP SL3, 60MP M11 or 100MP X2D.

Incidentally, Hasselblad finally added 4-shot pixel shift to the X2D via FW update but it requires it to be tethered to a computer running Phocus :-|
The point is there are many situations where a single shot is jsut the best thing you can do because there are external factors. Let alone the fact it takes time to take multiple shots

20 megapixels are not an issue in camera try assembling 50 60 is a bit different though panasonic can do 45 easily and really sony implementation is pathetic

In real terms with a very sharp lens there is abundance of resolution with a full frame camera more than you need without having to do anything special, of course other than having a good lens
They are using OMW to convert the raws. That is not a requirement of course. ACR and every major raw converter will give you an unsharpened image if that is desired. That said, the JPG's out of camera with (-1 sharpen) applied to HR and HHHR files look excellent. Nothing aggressive about it.

You're right, sometimes a single shot is the best you can do because of external factors. At the very least you still get a 20MP still frame which as we can see offers good resolution.

A1mkII: 3600 lw/ph

R5II: 3100 lw/ph

A7IV: 3100 lw/ph

OM5 Single: 2900 lw/ph

R6II: 2600 lw/ph

R7: 1800 lw/ph

For landscape, seascape, architecture, urban scenes, still life, product photography, macro, archiving, real estate, and long exposure scenes, HR and HHHR are quite nice!

The S1R and SL2 were assembling 47MP into 187MP pixel shift images with motion correction, in camera back in 2019 so I don't see size as a barrier. Agree Sony's implementation is poor.

In real terms resolution depends on a host of factors and every shooter and situation will yield various results. All that is necessary is to reach the point of sufficient for the photographs intended use. Given the vast majority of photographs shared around the world every second have significantly less resolution than the products we are discussing here, I think we are spoiled by the options available.
 
The 20MP cameras with pixel shift won't quite hold up to my 100MP X2D but then for 1/8 the price I can live with it. :-D
I can live with that as well :-) particularly as I prefer 4:3 ratio.
I call MF cameras "Mega Four-Thirds" :)
How do you find tonal gradations subtle transitions of Hassy X2D compared to OM1 Tripod Pixel Shift HR and OM1 Handheld Stacked HHHR in landscapes.
Quite good! The Hassy makes it easy to achieve high IQ but tonal graduations with HHHR and HR are quite nice. Probably more in the realm of a good FF camera than MF though.
2f1d0e9f805540e59f08ea02f5e36b5e.jpg

For landscapes I use the OM1 (or OM3) with HR and HHHR more than my 50MP A1, 60MP SL3, 60MP M11 or 100MP X2D.
That's quite revealing your choice I'm now looking at OM1 as a possible future purchase.
The OM1 is a "do everything" camera for me. The FF version would be the A1 (although missing so many great features). Before heading out the door I typically start with the OM-1 then rationalize why a different camera or an additional camera is warranted. In the end, the OM-1 gets carried most.

That doesn't detract from the other cameras, they are just less versatile for what I shoot.
In which types of scenarios would you choose X2D for landscapes.
Enjoyment! Hard to express how pleasing the shooting experience is with the Hassy. Just sublime in hand as are the X1D twins. Much like a Leica M - just makes you want to touch it. The bodies are relatively slim, massive screens and the best GUI of any cameras. Base ISO of 64 and gobs of dynamic range. Each camera is custom calibrated with some 2GB of calibration data stored inside.

9de86312e30945419ed2880f35f1e82d.jpg

cccd4a52edc14b558cc893f9304796d6.jpg
As someone linked to above Handheld Stacked HHHR is around 32MP, whereas Pixel Shift Tripod HR is under 60MP probably around 50MP. Also other things need to fall in place for a Tripod Pixel Shift HR to be a successful shot.
You're preaching to the choir! I talk about when, and when not to use HR and HHHR in the guides I posted years ago. These modes don't work for everything but they are much more useful than most people appreciate.

Pixel Shift: A Primer for Beginners

Hand-Held High Res: A Practical Guide


I am glad other manufacturers are starting to offer these. It will help dispel some of the misconceptions that surround them.
 
Last edited:
I have not performed exhaustive testing. However, I have shot HHHRES and THRES, and shot HHHRES off a tripod. I can say with certainty that I do get more detail with HHHRES than the standard 20 MP ORF mage, and it's better from a tripod, too. Obviously focus is critical, along with minimization of atmospheric distortions (mirage, wind) and movement in the photo. I use multi-row panos to get the resolution I need, and a 3x3 image with 50 MP HHHFES off of a tripod I get 20K horizontal pixels when I crop slightly, giving me easy 40" x 60" images at 300 dpi. Yes, landscapes in still conditions off of a tripod. I will often carry another body with the Oly 12-40/2.8 (or mount that on my E.M1.III) and use it to determine my desired field of view/focal length and then shoot at twice that focal length (or close to that) with a prime or the 40-150/2.8 Pro set to the specified focal length, and stop down to f/5.6 to f/8 at ISO 200. This gives me a very sharp high res image at large sizes.

I know, the camera is making up pixels, but I find it does a better job than any external software. Honestly, the next step up would be a 100 MP camera... and I can't justify the 5-figure cost of canera and lenses.
Someone has actually tested the pixel shift resolution increase on the OM-1

https://www.optyczne.pl/477.4-Test_aparatu-OM_System_OM-1_Rozdzielczość.html

In tripod mode it produces a resolution inferior to a current 60 megapixels full frame camera

in handheld mode it falls below a 33 megapixel full frame camera

the maximum resolution increase is around 1.6x as if the camera was circa 50 megapixels and in handheld around 38 megapixels if you consider the improvement on its own resolution those are not comparable to full frame resolution because the yield per pixel drops

In conclusion you get a current high resolution camera from sony like the a7cr and this with its 60 megapixels will look better than your tripod hr shot and you can take any shot you like plus it also has pixel shift

The photo will also have more dynamic range off camera too and you will not have restrictions of multi pixel shooting

There are many useful features and benefit of a cropped format but resolution is not one
It's nice to see actual resolution tests from optzczne, thanks for this. It also supports increased resolution in both HHHR and HR. Interesting the OM-1 (single 20MP mode) outperformed the XT4 and a6600 in resolution despite those cameras having more MP.

According to their results, HHHR improved resolution by 38%. HR (pixel shift) improved resolution 63%.

422968426d6b4a079a564668cf331247.jpg

Compared to something like the 60MP Sony bodies, the MTF50 are definitely less but not at all bad when you consider these values are achievable in a $1000 OM5 size body.

A7CR: 5200 lw/ph

OM5 HR: 4400 lw/ph

OM5 HHHR: 3700 lw/ph


A1mkII: 3600 lw/ph

R5II: 3100 lw/ph

A7IV: 3100 lw/ph

OM5 Single: 2900 lw/ph

R6II: 2600 lw/ph

R7: 1800 lw/ph

Of course a lot of this depends on the lens used. For example the A7CR was tested with the 50mm F1.4 GM whereas the A1II was tested with the new 24-70 F2.8 GM II. The R7 was tested with a craptacular STM lens which might explain the low values.
The LW/PH measures are using jpegs and OM systems jpegs are heavily sharpened same applies to the HR shots processed with their software

The results for the Hi Res Shot mode are naturally higher, and quite a bit. Since we used the OM Workspace app, it is worth taking a closer look at the issue of sharpening. Unfortunately, this is applied quite generously, despite setting the appropriate parameter to the lowest available value. This is proven by the graphs from Imatest, which we present below (for the 50 and 80 Mpix modes, respectively).

Eliminate that effect and you are comparing to full frame camera far away from the 80 and 50 megapixels of the shots and remember those sony cameras at 50 60 megapixels also have pixel shift but present a more modest resolution increase of 13-15% which confirms this feature is not really going to change the life of anyone

If you want a high resolution camera you need to buy one and you will have that benefit in any shots you take including moving ones.
Every JPG I use has sharpening so I don't see how I could eliminate that effect. It is always part of the end product. The other cameras are using JPG's too for these tests.
But as you can see when they turn the sharpening to the lowest setting which is their default there is nothing less

OM system like olympus has aggressive sharpening even at the lowest setting as shown by the MTF curves
I have no issue with Oly JPG sharpening in SR, HR or HHHR as it is what I use most - straight out of camera. Really nice balance of sharpness and noise in my opinion. Every camera maker bakes in some sharpening with JPG. The HR and HHHR raw files have little to no sharpening and there are many threads in these forums about how best to approach sharpening them.

Optzczne cites an 11% increase in Sony's 4-shot method and a 33% increase in the 16 shot method so even high res sensors will benefit from pixel shift. The problem with Sony's implementation is it can't do it in camera, which is a deal breaker for me. Panasonic and Olympus have a huge advantage here although Panasonic has some other quirky limitations.

Back to my shots above. The 20MP cameras with pixel shift won't quite hold up to my 100MP X2D but then for 1/8 the price I can live with it. :-D

2f1d0e9f805540e59f08ea02f5e36b5e.jpg

For landscapes I use the OM1 (or OM3) with HR and HHHR more than my 50MP A1, 60MP SL3, 60MP M11 or 100MP X2D.

Incidentally, Hasselblad finally added 4-shot pixel shift to the X2D via FW update but it requires it to be tethered to a computer running Phocus :-|
The point is there are many situations where a single shot is jsut the best thing you can do because there are external factors. Let alone the fact it takes time to take multiple shots

20 megapixels are not an issue in camera try assembling 50 60 is a bit different though panasonic can do 45 easily and really sony implementation is pathetic

In real terms with a very sharp lens there is abundance of resolution with a full frame camera more than you need without having to do anything special, of course other than having a good lens
They are using OMW to convert the raws. That is not a requirement of course. ACR and every major raw converter will give you an unsharpened image if that is desired. That said, the JPG's out of camera with (-1 sharpen) applied to HR and HHHR files look excellent. Nothing aggressive about it.

You're right, sometimes a single shot is the best you can do because of external factors. At the very least you still get a 20MP still frame which as we can see offers good resolution.

A1mkII: 3600 lw/ph

R5II: 3100 lw/ph

A7IV: 3100 lw/ph

OM5 Single: 2900 lw/ph

R6II: 2600 lw/ph

R7: 1800 lw/ph

For landscape, seascape, architecture, urban scenes, still life, product photography, macro, archiving, real estate, and long exposure scenes, HR and HHHR are quite nice!

The S1R and SL2 were assembling 47MP into 187MP pixel shift images with motion correction, in camera back in 2019 so I don't see size as a barrier. Agree Sony's implementation is poor.

In real terms resolution depends on a host of factors and every shooter and situation will yield various results. All that is necessary is to reach the point of sufficient for the photographs intended use. Given the vast majority of photographs shared around the world every second have significantly less resolution than the products we are discussing here, I think we are spoiled by the options available.
Isaac

Apologies but I missed the details. Are those resolution MTF50s wide open, at lens/sensor peak, or typical landscape DoF?

A

--
Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post
 
ee399a58b64249cc8360329650ab509b.jpg

That said, I rarely carry a "high MP" camera anymore since the advent of HR and HHHR in small/affordable bodies. The output is fantastic straight out of camera and I don't have to sit and fiddle with computers. I feel time "processing" images is time wasted.
LOL, looks like i dont need the X2D afterall, because you made it look like the OM1 HR 80MP offers better IQ.

Surely you didnt use different FOV/focal lengths or made it so the X2D is at any disadvantage, like not add any post sharpening, unlike the OM1 where sharpening was turned to 11. Like you couldnt use HR on the A1 or apply intense software enhancement on the FF and MF body too...

Honestly, what is the purpose this comparison? Its irrelevant to anyone considering A1 or X2D and it will just confuse people who might need hi-resolution but cannt use HH/HR
Actually it doesn't, and I never said that it did. The X2D offers more detail than any of the other cameras.

All of my shots were taken at ~24mm equivalent and matched for physical aperture except for the iPhone which is 24mm and fixed aperture.
  • iPhone: 6.8mm F1.8
  • OM1: 12mm F2.8
  • A1: 24mm F5.6
  • X2D: 30mm F6.7
Sharpening was the same for all cameras:
  • Sharpening: 40
  • Radius: 1.0
  • Detail: 25
  • Masking: 0
The images you see above were all viewed at the same size as the X2D which means the OM1 is at 205%, A1 at 130%, etc. If you prefer to match images to say the 50MP A1 we can look at that also:



Center frame
Center frame



Corner frame
Corner frame

In every case the X2D offers more detail.

The point of this was to see how my most commonly used cameras render the same scene at the same FoV and DoF. Sure we could tweak color, WB, sharpening and noise reduction individually and get some improvement but that was beyond my interest at the time.

I have been pretty clear about when HR and HHHR can and cannot be used. I have shared examples of how I use it: High-Res Gallery

I'm sure I am missing some uses - others are much more creative than I. Just my experience.
 
I have not performed exhaustive testing. However, I have shot HHHRES and THRES, and shot HHHRES off a tripod. I can say with certainty that I do get more detail with HHHRES than the standard 20 MP ORF mage, and it's better from a tripod, too. Obviously focus is critical, along with minimization of atmospheric distortions (mirage, wind) and movement in the photo. I use multi-row panos to get the resolution I need, and a 3x3 image with 50 MP HHHFES off of a tripod I get 20K horizontal pixels when I crop slightly, giving me easy 40" x 60" images at 300 dpi. Yes, landscapes in still conditions off of a tripod. I will often carry another body with the Oly 12-40/2.8 (or mount that on my E.M1.III) and use it to determine my desired field of view/focal length and then shoot at twice that focal length (or close to that) with a prime or the 40-150/2.8 Pro set to the specified focal length, and stop down to f/5.6 to f/8 at ISO 200. This gives me a very sharp high res image at large sizes.

I know, the camera is making up pixels, but I find it does a better job than any external software. Honestly, the next step up would be a 100 MP camera... and I can't justify the 5-figure cost of canera and lenses.
Someone has actually tested the pixel shift resolution increase on the OM-1

https://www.optyczne.pl/477.4-Test_aparatu-OM_System_OM-1_Rozdzielczość.html

In tripod mode it produces a resolution inferior to a current 60 megapixels full frame camera

in handheld mode it falls below a 33 megapixel full frame camera

the maximum resolution increase is around 1.6x as if the camera was circa 50 megapixels and in handheld around 38 megapixels if you consider the improvement on its own resolution those are not comparable to full frame resolution because the yield per pixel drops

In conclusion you get a current high resolution camera from sony like the a7cr and this with its 60 megapixels will look better than your tripod hr shot and you can take any shot you like plus it also has pixel shift

The photo will also have more dynamic range off camera too and you will not have restrictions of multi pixel shooting

There are many useful features and benefit of a cropped format but resolution is not one
It's nice to see actual resolution tests from optzczne, thanks for this. It also supports increased resolution in both HHHR and HR. Interesting the OM-1 (single 20MP mode) outperformed the XT4 and a6600 in resolution despite those cameras having more MP.

According to their results, HHHR improved resolution by 38%. HR (pixel shift) improved resolution 63%.

422968426d6b4a079a564668cf331247.jpg

Compared to something like the 60MP Sony bodies, the MTF50 are definitely less but not at all bad when you consider these values are achievable in a $1000 OM5 size body.

A7CR: 5200 lw/ph

OM5 HR: 4400 lw/ph

OM5 HHHR: 3700 lw/ph


A1mkII: 3600 lw/ph

R5II: 3100 lw/ph

A7IV: 3100 lw/ph

OM5 Single: 2900 lw/ph

R6II: 2600 lw/ph

R7: 1800 lw/ph

Of course a lot of this depends on the lens used. For example the A7CR was tested with the 50mm F1.4 GM whereas the A1II was tested with the new 24-70 F2.8 GM II. The R7 was tested with a craptacular STM lens which might explain the low values.
The LW/PH measures are using jpegs and OM systems jpegs are heavily sharpened same applies to the HR shots processed with their software

The results for the Hi Res Shot mode are naturally higher, and quite a bit. Since we used the OM Workspace app, it is worth taking a closer look at the issue of sharpening. Unfortunately, this is applied quite generously, despite setting the appropriate parameter to the lowest available value. This is proven by the graphs from Imatest, which we present below (for the 50 and 80 Mpix modes, respectively).

Eliminate that effect and you are comparing to full frame camera far away from the 80 and 50 megapixels of the shots and remember those sony cameras at 50 60 megapixels also have pixel shift but present a more modest resolution increase of 13-15% which confirms this feature is not really going to change the life of anyone

If you want a high resolution camera you need to buy one and you will have that benefit in any shots you take including moving ones.
Every JPG I use has sharpening so I don't see how I could eliminate that effect. It is always part of the end product. The other cameras are using JPG's too for these tests.
But as you can see when they turn the sharpening to the lowest setting which is their default there is nothing less

OM system like olympus has aggressive sharpening even at the lowest setting as shown by the MTF curves
I have no issue with Oly JPG sharpening in SR, HR or HHHR as it is what I use most - straight out of camera. Really nice balance of sharpness and noise in my opinion. Every camera maker bakes in some sharpening with JPG. The HR and HHHR raw files have little to no sharpening and there are many threads in these forums about how best to approach sharpening them.

Optzczne cites an 11% increase in Sony's 4-shot method and a 33% increase in the 16 shot method so even high res sensors will benefit from pixel shift. The problem with Sony's implementation is it can't do it in camera, which is a deal breaker for me. Panasonic and Olympus have a huge advantage here although Panasonic has some other quirky limitations.

Back to my shots above. The 20MP cameras with pixel shift won't quite hold up to my 100MP X2D but then for 1/8 the price I can live with it. :-D

2f1d0e9f805540e59f08ea02f5e36b5e.jpg

For landscapes I use the OM1 (or OM3) with HR and HHHR more than my 50MP A1, 60MP SL3, 60MP M11 or 100MP X2D.

Incidentally, Hasselblad finally added 4-shot pixel shift to the X2D via FW update but it requires it to be tethered to a computer running Phocus :-|
The point is there are many situations where a single shot is jsut the best thing you can do because there are external factors. Let alone the fact it takes time to take multiple shots

20 megapixels are not an issue in camera try assembling 50 60 is a bit different though panasonic can do 45 easily and really sony implementation is pathetic

In real terms with a very sharp lens there is abundance of resolution with a full frame camera more than you need without having to do anything special, of course other than having a good lens
They are using OMW to convert the raws. That is not a requirement of course. ACR and every major raw converter will give you an unsharpened image if that is desired. That said, the JPG's out of camera with (-1 sharpen) applied to HR and HHHR files look excellent. Nothing aggressive about it.

You're right, sometimes a single shot is the best you can do because of external factors. At the very least you still get a 20MP still frame which as we can see offers good resolution.

A1mkII: 3600 lw/ph

R5II: 3100 lw/ph

A7IV: 3100 lw/ph

OM5 Single: 2900 lw/ph

R6II: 2600 lw/ph

R7: 1800 lw/ph

For landscape, seascape, architecture, urban scenes, still life, product photography, macro, archiving, real estate, and long exposure scenes, HR and HHHR are quite nice!

The S1R and SL2 were assembling 47MP into 187MP pixel shift images with motion correction, in camera back in 2019 so I don't see size as a barrier. Agree Sony's implementation is poor.

In real terms resolution depends on a host of factors and every shooter and situation will yield various results. All that is necessary is to reach the point of sufficient for the photographs intended use. Given the vast majority of photographs shared around the world every second have significantly less resolution than the products we are discussing here, I think we are spoiled by the options available.
Isaac

Apologies but I missed the details. Are those resolution MTF50s wide open, at lens/sensor peak, or typical landscape DoF?

A
Those are peak. The site interceptor121 linked to previously is in Polish(?) but maps out MTF's with various cameras and lenses.

For example the new A1 mark II peaks at F5.6 but probably couldn't tell a difference from F2.8-F5.6.

434ed591e58241319ac0675714914b6d.jpg

https://www.optyczne.pl/511.4-Test_aparatu-Sony_A1_II_Rozdzielczość.html
 
I have not performed exhaustive testing. However, I have shot HHHRES and THRES, and shot HHHRES off a tripod. I can say with certainty that I do get more detail with HHHRES than the standard 20 MP ORF mage, and it's better from a tripod, too. Obviously focus is critical, along with minimization of atmospheric distortions (mirage, wind) and movement in the photo. I use multi-row panos to get the resolution I need, and a 3x3 image with 50 MP HHHFES off of a tripod I get 20K horizontal pixels when I crop slightly, giving me easy 40" x 60" images at 300 dpi. Yes, landscapes in still conditions off of a tripod. I will often carry another body with the Oly 12-40/2.8 (or mount that on my E.M1.III) and use it to determine my desired field of view/focal length and then shoot at twice that focal length (or close to that) with a prime or the 40-150/2.8 Pro set to the specified focal length, and stop down to f/5.6 to f/8 at ISO 200. This gives me a very sharp high res image at large sizes.

I know, the camera is making up pixels, but I find it does a better job than any external software. Honestly, the next step up would be a 100 MP camera... and I can't justify the 5-figure cost of canera and lenses.
Someone has actually tested the pixel shift resolution increase on the OM-1

https://www.optyczne.pl/477.4-Test_aparatu-OM_System_OM-1_Rozdzielczość.html

In tripod mode it produces a resolution inferior to a current 60 megapixels full frame camera

in handheld mode it falls below a 33 megapixel full frame camera

the maximum resolution increase is around 1.6x as if the camera was circa 50 megapixels and in handheld around 38 megapixels if you consider the improvement on its own resolution those are not comparable to full frame resolution because the yield per pixel drops

In conclusion you get a current high resolution camera from sony like the a7cr and this with its 60 megapixels will look better than your tripod hr shot and you can take any shot you like plus it also has pixel shift

The photo will also have more dynamic range off camera too and you will not have restrictions of multi pixel shooting

There are many useful features and benefit of a cropped format but resolution is not one
It's nice to see actual resolution tests from optzczne, thanks for this. It also supports increased resolution in both HHHR and HR. Interesting the OM-1 (single 20MP mode) outperformed the XT4 and a6600 in resolution despite those cameras having more MP.

According to their results, HHHR improved resolution by 38%. HR (pixel shift) improved resolution 63%.

422968426d6b4a079a564668cf331247.jpg

Compared to something like the 60MP Sony bodies, the MTF50 are definitely less but not at all bad when you consider these values are achievable in a $1000 OM5 size body.

A7CR: 5200 lw/ph

OM5 HR: 4400 lw/ph

OM5 HHHR: 3700 lw/ph


A1mkII: 3600 lw/ph

R5II: 3100 lw/ph

A7IV: 3100 lw/ph

OM5 Single: 2900 lw/ph

R6II: 2600 lw/ph

R7: 1800 lw/ph

Of course a lot of this depends on the lens used. For example the A7CR was tested with the 50mm F1.4 GM whereas the A1II was tested with the new 24-70 F2.8 GM II. The R7 was tested with a craptacular STM lens which might explain the low values.
The LW/PH measures are using jpegs and OM systems jpegs are heavily sharpened same applies to the HR shots processed with their software

The results for the Hi Res Shot mode are naturally higher, and quite a bit. Since we used the OM Workspace app, it is worth taking a closer look at the issue of sharpening. Unfortunately, this is applied quite generously, despite setting the appropriate parameter to the lowest available value. This is proven by the graphs from Imatest, which we present below (for the 50 and 80 Mpix modes, respectively).

Eliminate that effect and you are comparing to full frame camera far away from the 80 and 50 megapixels of the shots and remember those sony cameras at 50 60 megapixels also have pixel shift but present a more modest resolution increase of 13-15% which confirms this feature is not really going to change the life of anyone

If you want a high resolution camera you need to buy one and you will have that benefit in any shots you take including moving ones.
Every JPG I use has sharpening so I don't see how I could eliminate that effect. It is always part of the end product. The other cameras are using JPG's too for these tests.
But as you can see when they turn the sharpening to the lowest setting which is their default there is nothing less

OM system like olympus has aggressive sharpening even at the lowest setting as shown by the MTF curves
I have no issue with Oly JPG sharpening in SR, HR or HHHR as it is what I use most - straight out of camera. Really nice balance of sharpness and noise in my opinion. Every camera maker bakes in some sharpening with JPG. The HR and HHHR raw files have little to no sharpening and there are many threads in these forums about how best to approach sharpening them.

Optzczne cites an 11% increase in Sony's 4-shot method and a 33% increase in the 16 shot method so even high res sensors will benefit from pixel shift. The problem with Sony's implementation is it can't do it in camera, which is a deal breaker for me. Panasonic and Olympus have a huge advantage here although Panasonic has some other quirky limitations.

Back to my shots above. The 20MP cameras with pixel shift won't quite hold up to my 100MP X2D but then for 1/8 the price I can live with it. :-D

2f1d0e9f805540e59f08ea02f5e36b5e.jpg

For landscapes I use the OM1 (or OM3) with HR and HHHR more than my 50MP A1, 60MP SL3, 60MP M11 or 100MP X2D.

Incidentally, Hasselblad finally added 4-shot pixel shift to the X2D via FW update but it requires it to be tethered to a computer running Phocus :-|
The point is there are many situations where a single shot is jsut the best thing you can do because there are external factors. Let alone the fact it takes time to take multiple shots

20 megapixels are not an issue in camera try assembling 50 60 is a bit different though panasonic can do 45 easily and really sony implementation is pathetic

In real terms with a very sharp lens there is abundance of resolution with a full frame camera more than you need without having to do anything special, of course other than having a good lens
They are using OMW to convert the raws. That is not a requirement of course. ACR and every major raw converter will give you an unsharpened image if that is desired. That said, the JPG's out of camera with (-1 sharpen) applied to HR and HHHR files look excellent. Nothing aggressive about it.

You're right, sometimes a single shot is the best you can do because of external factors. At the very least you still get a 20MP still frame which as we can see offers good resolution.

A1mkII: 3600 lw/ph

R5II: 3100 lw/ph

A7IV: 3100 lw/ph

OM5 Single: 2900 lw/ph

R6II: 2600 lw/ph

R7: 1800 lw/ph

For landscape, seascape, architecture, urban scenes, still life, product photography, macro, archiving, real estate, and long exposure scenes, HR and HHHR are quite nice!

The S1R and SL2 were assembling 47MP into 187MP pixel shift images with motion correction, in camera back in 2019 so I don't see size as a barrier. Agree Sony's implementation is poor.

In real terms resolution depends on a host of factors and every shooter and situation will yield various results. All that is necessary is to reach the point of sufficient for the photographs intended use. Given the vast majority of photographs shared around the world every second have significantly less resolution than the products we are discussing here, I think we are spoiled by the options available.
Isaac

Apologies but I missed the details. Are those resolution MTF50s wide open, at lens/sensor peak, or typical landscape DoF?

A
Those are peak. The site interceptor121 linked to previously is in Polish(?) but maps out MTF's with various cameras and lenses.

For example the new A1 mark II peaks at F5.6 but probably couldn't tell a difference from F2.8-F5.6.

434ed591e58241319ac0675714914b6d.jpg

https://www.optyczne.pl/511.4-Test_aparatu-Sony_A1_II_Rozdzielczość.html
As my landscape tutor said, shooting at peak resolution is rarely appropriate as a composition aid.

Thanks for the answer. I see you provided details on your images in another post.

A

--
Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post
 
It looks like the Topaz image is the least accurate, with the foliage being finer and more weby than any of the other renditions, and that railing near the center looks double. The AI seems to be adding detail that wasn't there. But there's a lot of adjustment to that program, so might be a matter of playing with settings to get it more accurate?
I used Topaz AI about a year ago, it was doing lots of weird stuff, their forum was full of people having problems, since then l have not used it apart from sharpening on some images.

Some people get good results but l gave up with it. Maybe they have sorted some of the issues out now, think they were trying to do too much.
 
It looks like the Topaz image is the least accurate, with the foliage being finer and more weby than any of the other renditions, and that railing near the center looks double. The AI seems to be adding detail that wasn't there. But there's a lot of adjustment to that program, so might be a matter of playing with settings to get it more accurate?
I used Topaz AI about a year ago, it was doing lots of weird stuff, their forum was full of people having problems, since then l have not used it apart from sharpening on some images.

Some people get good results but l gave up with it. Maybe they have sorted some of the issues out now, think they were trying to do too much.
From my experience with Topaz Photo AI (which was similar to yours a year ago), they have made great improvements since then. It’s now a standard tool in my post processing kit.
 
Last edited:
It looks like the Topaz image is the least accurate, with the foliage being finer and more weby than any of the other renditions, and that railing near the center looks double. The AI seems to be adding detail that wasn't there. But there's a lot of adjustment to that program, so might be a matter of playing with settings to get it more accurate?
I used Topaz AI about a year ago, it was doing lots of weird stuff, their forum was full of people having problems, since then l have not used it apart from sharpening on some images.

Some people get good results but l gave up with it. Maybe they have sorted some of the issues out now, think they were trying to do too much.
From my experience with Topaz Photo AI (which was similar to yours a year ago), they have made great improvements since then. It’s now a standard tool in my post processing kit.
OK, l may retry it. I use DxO for RAW conversion, which is good for denoising and lens correction. I get a bit of noise on long exposures, just switched from Canon to Sony, maybe Sony is better?

LR and Photoshop seem to have improved also, l only have CS6. Not keen on monthly subs but l may bite the bullet.
 
It looks like the Topaz image is the least accurate, with the foliage being finer and more weby than any of the other renditions, and that railing near the center looks double. The AI seems to be adding detail that wasn't there. But there's a lot of adjustment to that program, so might be a matter of playing with settings to get it more accurate?
I used Topaz AI about a year ago, it was doing lots of weird stuff, their forum was full of people having problems, since then l have not used it apart from sharpening on some images.

Some people get good results but l gave up with it. Maybe they have sorted some of the issues out now, think they were trying to do too much.
From my experience with Topaz Photo AI (which was similar to yours a year ago), they have made great improvements since then. It’s now a standard tool in my post processing kit.
OK, l may retry it. I use DxO for RAW conversion, which is good for denoising and lens correction. I get a bit of noise on long exposures, just switched from Canon to Sony, maybe Sony is better?

LR and Photoshop seem to have improved also, l only have CS6. Not keen on monthly subs but l may bite the bullet.
The new AI DeNoise in LR now makes taking my ORF files directly into Lightroom a decent option as opposed to doing the raw conversion in Workspace first. A welcome advancement. I am having issues with my OM-3 files with the Super Focus in Topaz Photo AI. No issue with my files from my OM-1 II, just the OM-3, both tiff and JPEG
 
The 20MP cameras with pixel shift won't quite hold up to my 100MP X2D but then for 1/8 the price I can live with it. :-D
I can live with that as well :-) particularly as I prefer 4:3 ratio.

How do you find tonal gradations subtle transitions of Hassy X2D compared to OM1 Tripod Pixel Shift HR and OM1 Handheld Stacked HHHR in landscapes.
2f1d0e9f805540e59f08ea02f5e36b5e.jpg

For landscapes I use the OM1 (or OM3) with HR and HHHR more than my 50MP A1, 60MP SL3, 60MP M11 or 100MP X2D.
That's quite revealing your choice I'm now looking at OM1 as a possible future purchase.

In which types of scenarios would you choose X2D for landscapes.

As someone linked to above Handheld Stacked HHHR is around 32MP, whereas Pixel Shift Tripod HR is under 60MP probably around 50MP. Also other things need to fall in place for a Tripod Pixel Shift HR to be a successful shot.
You want an OM1 mk ii or OM3 because of the mk I bug that forces you to use UniWB to get the full benefits of HHHR.

As far as OM vs A7CR, my answer would be that it depends on your range of use cases which is best. Within FE mount, the A1 ii does everything well, apart from weight and cost.

Andrew
IIRC, I tested OM1 mk II and OM3, and the clipping in HR modes was still there (requiring UniWB or avoiding clipping completely).
 
The 20MP cameras with pixel shift won't quite hold up to my 100MP X2D but then for 1/8 the price I can live with it. :-D
I can live with that as well :-) particularly as I prefer 4:3 ratio.

How do you find tonal gradations subtle transitions of Hassy X2D compared to OM1 Tripod Pixel Shift HR and OM1 Handheld Stacked HHHR in landscapes.
2f1d0e9f805540e59f08ea02f5e36b5e.jpg

For landscapes I use the OM1 (or OM3) with HR and HHHR more than my 50MP A1, 60MP SL3, 60MP M11 or 100MP X2D.
That's quite revealing your choice I'm now looking at OM1 as a possible future purchase.

In which types of scenarios would you choose X2D for landscapes.

As someone linked to above Handheld Stacked HHHR is around 32MP, whereas Pixel Shift Tripod HR is under 60MP probably around 50MP. Also other things need to fall in place for a Tripod Pixel Shift HR to be a successful shot.
You want an OM1 mk ii or OM3 because of the mk I bug that forces you to use UniWB to get the full benefits of HHHR.

As far as OM vs A7CR, my answer would be that it depends on your range of use cases which is best. Within FE mount, the A1 ii does everything well, apart from weight and cost.

Andrew
IIRC, I tested OM1 mk II and OM3, and the clipping in HR modes was still there (requiring UniWB or avoiding clipping completely).
Bummer!

:(

--
Infinite are the arguments of mages. Truth is a jewel with many facets. Ursula K LeGuin
Please feel free to edit any images that I post
 

Keyboard shortcuts

Back
Top