p.8 #1 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
Here's a quick treatment I did of an A7rII image that has soft edges using USM in PS with a gradient mask to progressively mask the sharpening into the center. These are 100% crops that include the center to extreme left edge:
p.8 #2 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
Based on just two days of shooting with the A7RIV, with one lens (Batis 40mm), I see pixel level sharpness similar to that from my A7RIII. Normalized for similar size prints, that's a net gain. In some cases, I think I can see less sharpness in the corners, but if so, not much. (The Batis is very even across the frame to begin with.) I also don't see major diffraction problems with shooting at f8 or f11, with a little extra deconvolution sharpening.
Shadow noise becomes prominent earlier in the ISO range using the A7RIV. But that's not an issue for the kind of photography I usually do. The "Aptina bump" at ISO 320 makes that ISO very useful, with little loss of DR compared to ISO 100.
This is pretty much what I was looking for from the new camera, resolution-wise. Pixel level results close to the A7RIII, but with more pixels available for large prints, perspective control, and cropping.
Last time I tried Gigapixel AI, I found that it sometimes "made up" false details or did other unexpected things to my files. Maybe it's evolved since then?
p.8 #3 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
When appropriate/necessary, what are you using for deconvolution sharpening?
"Based on just two days of shooting with the A7RIV, with one lens (Batis 40mm), I see pixel level sharpness similar to that from my A7RIII. Normalized for similar size prints, that's a net gain. In some cases, I think I can see less sharpness in the corners, but if so, not much. (The Batis is very even across the frame to begin with.) I also don't see major diffraction problems with shooting at f8 or f11, with a little extra deconvolution sharpening."
p.8 #4 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
I use the Detail slider in ACR, set at or near 100. Small radius, easy on the amount. This is not my only sharpening step for my files, but is part of the process when diffraction is an issue.
p.8 #5 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
snapsy wrote:
Thanks. If you or anyone else has the time I'd love to see before/after crops of the center vs extreme edge of this C1 area-dependent lens correction sharpening in action.
In order for it to work, the lens needs to be supported by C1 via a “C1 lens profile”. The “manufacturer profile” (supplied by Sony in the RAW files) does not have sufficient information for this function to work. C1 generally supplies the “C1 profiles” a few months after a new lens is launched. E.g. right now the FE 35/1.8 doesn’t have one, you can use only the “manufacturer profile” which comes from the lens itself via the RAW embedded info.
p.8 #6 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
No free lunch with puny 47 or 60 mP images. I have a friend who makes stitched GigaPixel files. I made 2’ X 10’ print for him only using a portion of the available resolution. No end of detail with your nose right up against it.
The thing is that he can use most any camera with a medium tele for the work. However, you really need a special kind of computer to crunch the files and hours of patience.
In order for it to work, the lens needs to be supported by C1 via a “C1 lens profile”. The “manufacturer profile” (supplied by Sony in the RAW files) does not have sufficient information for this function to work. C1 generally supplies the “C1 profiles” a few months after a new lens is launched. E.g. right now the FE 35/1.8 doesn’t have one, you can use only the “manufacturer profile” which comes from the lens itself via the RAW embedded info.
p.8 #8 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
racetratr wrote:
I use the Detail slider in ACR, set at or near 100. Small radius, easy on the amount. This is not my only sharpening step for my files, but is part of the process when diffraction is an issue.
+1
With diffracted images in LR, I keep the sharpening amount/radius the same and raise the 'detail' slider to 75-100 (from my default 30) depending on the aperture used. There is clearly improvement without much extra noise. In order to minimize noise, averaging 2 or 4 images is also a good idea, especially when using deconvolution.
p.8 #9 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
Fred,
Question on above post: you keep the sharpening amount/radius the same (meaning the LR default, or whatever the values you've customized to?
Fred Miranda wrote:
+1
With diffracted images in LR, I keep the sharpening amount/radius the same and raise the 'detail' slider to 75-100 (from my default 30) depending on the aperture used. There is clearly improvement without much extra noise. In order to minimize noise, averaging 2 or 4 images is also a good idea, especially when using deconvolution.
p.8 #10 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
StephenLowell wrote:
Fred,
Question on above post: you keep the sharpening amount/radius the same (meaning the LR default, or whatever the values you've customized to?
My sharpening settings changes depending on the application. For landscapes I like amount 50, radius 0.8 and detail 30 when aperture is larger than f/7.1. When shooting at smaller apertures, I just increase the 'detail' value.
I'm experimenting with different settings for the A7R4 and I am finding out that 'amount' can be increased a bit, perhaps as high as '60' while keeping the other settings the same.
In order for it to work, the lens needs to be supported by C1 via a “C1 lens profile”. The “manufacturer profile” (supplied by Sony in the RAW files) does not have sufficient information for this function to work. C1 generally supplies the “C1 profiles” a few months after a new lens is launched. E.g. right now the FE 35/1.8 doesn’t have one, you can use only the “manufacturer profile” which comes from the lens itself via the RAW embedded info.
So I thought I would answer this question with C1 and lenses that do have profiles. So this is Medium format digital from a few years back. This is a Phase One back on a Mamya 645 DSLR Style with a Mamiya 28mm F 3.5 lens which was a good lens but the corners did need some help. So as mentioned in C1 there is a slider setting called sharpness under lens correction which will increase the sharpness to the corners. Now the Mamiya 28mm is prime candidate as it is a Phase One lens or lets say heavily supported by Phase with there profile and such. But used correctly with this lens which shown here is a new version of it but back in the day I think i paid 3200 for it
Anyway these where all tweaked to give better corners all processed in C1 for a big custom home builder. This job lead into several more for me. So good gig and paid well.
Just wanted to show you guys a sample of what that can do with a lens that was not known for great corners.
p.8 #12 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
Fred Miranda wrote:
+1
With diffracted images in LR, I keep the sharpening amount/radius the same and raise the 'detail' slider to 75-100 (from my default 30) depending on the aperture used. There is clearly improvement without much extra noise. In order to minimize noise, averaging 2 or 4 images is also a good idea, especially when using deconvolution.
Those are very close to my usual settings. I might go down to .6 or even .5 on the radius, depending on the sensor and subject matter. Moving the Detail slider closer to 100 in ACR increases the proportion of deconvolution sharpening. I also use the masking slider freely, depending on the type of image.
As far as averaging images goes, my latest setting for hand-held cityscapes has been to use the camera's 2 second delay setting, which allows me to specify a high speed burst of five exposures. Using silent shutter, it's super easy to get five hand-held captures that are barely offset from each other in space and time.
Uncompressed RAW from the A74, plus five captures=lots of data. I compensate by ruthlessly discarding weaker images.
p.8 #13 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
philip_pj wrote:
What differentiates base resolution in comparisons is the 'law of diminishing returns'. You can't reasonably use a 2005 12mp sensor as a base and expect an increase from a 2019 42mp sensor as base to behave or be perceived in similar fashion. One left most users thirsty for (much) more, the other sates most appetites, even for high demand uses. What level of demand exists for 80mp? How large a print is required to impress over 61mp?
Moreover, 42mp was chosen for its fine balance of characteristics and breakthrough sensor architecture - Sony was aware few needed them to match the low DR (two stops down) 50mp Canon unit.
It would not surprise if this a7rIV is close to the last roll of the resolution dice in the 135 format. Each throw builds the satiation percentage of the user base closer to 100%, apart from specialised use cases that will always get well (and disproportionately) aired on gear forums. But it's tremendous news for buyers of older Sony cameras....Show more →
Tagged for future reference.
Your strongest point here is that with successive higher resolution, the number of photographers that will make use of it will be lower.
At the time we got 12 MP, I felt that finally, we had a resolution in digital that I could work without limitations. There and then, I was not hungry for more.
With 50 MP, I feel that for the largest print size I make at home, 16 inches roll width, there is no point in more resolution, and I think 50 looks better than 24 and maybe 36. But we are only one step in print size from that point being 120 MP. And there are even more steps in print sizes easily commercially available and it is room for them on abut any wall, home or gallery.
It is not new with high resolutions. Scan tests from B&W film, like one made by onlandscape, show that large format film is possibly into the 400 MP territory. So with 150 MP we are not even close to catching up with what was available in the old days.
Now I hear all of you say that: but with small format like the 135 format there are so many problems with this and that, we will never get there. I have already been there in the center of the image with our DSLR lenses adapted on the Pentax Q at 360 MP equivalent. And I know that I have lenses where that will hold up out to he edges. That may however require using lenses that have been written off by mirrorless users for being way too large and heavy. So it is not for everyone to go there.
p.8 #15 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
Fred Miranda wrote:
Note: Even though the Loxia 35/2 is an older design (35/2 ZM) optimized for the Sony sensor, it has outstanding performance at center and f/4. (As you can see above, the extra megapixels are able to discern finer detail)
Thanks for these, Fred!
How are you focusing the L35 (edge, mid?) on the A7R4 for landscapes? Any difference in approach vs the A7R3?
I have a L35 in the box at home with no FE camera and was debating whether to get back in with a cheap A7R3 or new R4 for my mostly-stopped down travel and landscapes.
p.8 #16 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
krugorg wrote:
Thanks for these, Fred!
How are you focusing the L35 (edge, mid?) on the A7R4 for landscapes? Any difference in approach vs the A7R3?
I have a L35 in the box at home with no FE camera and was debating whether to get back in with a cheap A7R3 or new R4 for my mostly-stopped down travel and landscapes.
The Loxia 35 has wavy field curvature that only gets completely masked after f/5.6.
So, if I focus on center, the mid-field will be slightly out of focus but recovers as depth of field expands.
Yes, for the sample above, I focused it at center. If you want a better compromise for landscapes, focus between center and mid-field and step it down to f/7.1 or smaller.
p.8 #17 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
Fred Miranda wrote:
The Loxia 35 has wavy field curvature that only gets completely masked after f/5.6.
So, if I focus on center, the mid-field will be slightly out of focus but recovers as depth of field expands.
Yes, for the sample above, I focused it at center. If you want to a better compromise, you can focus between center and mid-field and step it down to about f/7.1 or smaller for landscapes.
Thanks! Can't remember what thread, but thought you also posted some edge photos where the L35 looked pretty good on the R4 (better than the R3)?
p.8 #18 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
krugorg wrote:
Thanks for these, Fred!
How are you focusing the L35 (edge, mid?) on the A7R4 for landscapes? Any difference in approach vs the A7R3?
I have a L35 in the box at home with no FE camera and was debating whether to get back in with a cheap A7R3 or new R4 for my mostly-stopped down travel and landscapes.
I focus the Loxia 35 between center and mid-filed areas as I found it's the best compromise since this lens a wavy field curvature. I usually stop it down at least f/6.3.
p.8 #19 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
It's been largely eradicated in most of the modern lenses that many use on the high resolution sensors, but field curvature is the aberration that cannot be easily predicted or understood, because it occurs in longitudinal (along the axis) image space.
And the really strange about it? It changes not just by focus distance but by aperture as well. So the fancy lens you have that you (or the body's auto aperture) focus at widest aperture (for light and seeing the focal plane well) can turn on you when you (or the camera) stop it down to f5.6 or f8! A lot can change in the five stops between f1.4 and f8, and the four stops between f2 and f8.
People work around it but the slow fade of the focal plane at landscape apertures faithfully tracks the wavy pattern of the focal plane, still disrupting the image - the now covered part will be weaker than the axis and centre region. And it's obviously more obvious on high resolution sensor images. (i's a good reason to always have a flat field lens for certain uses). Roger Cicala explains in this 2014 article, back in auto aperture DSLR days:
'I’d also never really thought about what happens to field curvature when you stop down. If I thought about it at all, I probably assumed it would flatten out. Or maybe stay the same but the increasing depth of field would make it less noticeable. Turns out that isn’t always the case either. So, because I made assumptions and didn’t know what I didn’t know, I wasted a lot of time. Weeks of time.'
Like other 'secret' aberrations (focus shift and focus breathing), it's less likely to be encountered in slow lenses, because those ones only have one, two or three stops to work any nastiness on you, and the apertures that instigate the aberrations are often not in their aperture range.
Not surprisingly, fast lenses are set up for use at widest apertures, that's why people buy them. It's also why the aberrations that get the spotlight are those that are most obvious wide open: veiling glare, ghosting, lateral CA, axial CA, coma etc. while these other guys do their dirty work in the relative darkness of mid apertures.
They rarely get any press at all, despite (for example) turning your 35mm lens into a 40mm lens, or shifting your carefully set up focius point back a few inches, or bending your plane of best focus like a pretzel.
p.8 #20 · Comparing 42MP vs 61MP: Real-World Resolution and Contrast
philip_pj wrote:
It's been largely eradicated in most of the modern lenses that many use on the high resolution sensors, but field curvature is the aberration that cannot be easily predicted or understood, because it occurs in longitudinal (along the axis) image space.
And the really strange about it? It changes not just by focus distance but by aperture as well. So the fancy lens you have that you (or the body's auto aperture) focus at widest aperture (for light and seeing the focal plane well) can turn on you when you (or the camera) stop it down to f5.6 or f8! A lot can change in the five stops between f1.4 and f8, and the four stops between f2 and f8.
People work around it but the slow fade of the focal plane at landscape apertures faithfully tracks the wavy pattern of the focal plane, still disrupting the image - the now covered part will be weaker than the axis and centre region. And it's obviously more obvious on high resolution sensor images. (i's a good reason to always have a flat field lens for certain uses). Roger Cicala explains in this 2014 article, back in auto aperture DSLR days:
'I’d also never really thought about what happens to field curvature when you stop down. If I thought about it at all, I probably assumed it would flatten out. Or maybe stay the same but the increasing depth of field would make it less noticeable. Turns out that isn’t always the case either. So, because I made assumptions and didn’t know what I didn’t know, I wasted a lot of time. Weeks of time.'
Like other 'secret' aberrations (focus shift and focus breathing), it's less likely to be encountered in slow lenses, because those ones only have one, two or three stops to work any nastiness on you, and the apertures that instigate the aberrations are often not in their aperture range.
Not surprisingly, fast lenses are set up for use at widest apertures, that's why people buy them. It's also why the aberrations that get the spotlight are those that are most obvious wide open: veiling glare, ghosting, lateral CA, axial CA, coma etc. while these other guys do their dirty work in the relative darkness of mid apertures.
They rarely get any press at all, despite (for example) turning your 35mm lens into a 40mm lens, or shifting your carefully set up focius point back a few inches, or bending your plane of best focus like a pretzel. ...Show more →
I really miss all the testing that Roger did at the LensRentals blog.
