Problems getting sharp images by 60MP

[idusidusi]

Advertisement (gone after registration)

On 3/2/2022 at 4:04 PM, peterinkingston said:

trying to avoid the useless settings drives me nuts, I often fumble or give up. Give me any M any day of the week.

I identify with this very much I sometimes miss shots with the Sony A1 for the same reason, technology getting in the way.

[Advertisement]

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

[jaapv]

The solution with more complicated cameras is to select the relevant ones at home and save them under a user profile. Practical cameratas will have those profiles on a analog dial instead of the M where it is buried in a menu. I find my Panasonics are as straightforward to shoot -sometimes even more simple- than my M cameras. Come to think of it, such a mode dial would have been more useful than the present ISO one. 

[FlashGordonPhotography]

2 hours ago, jaapv said:

Actually, the images at 60 MP are exactly as blurry as the images one made at 10,18, or 24 MP. It is just that the camera shows the existing user error whereas lower MP cameras hide it. (At 100%) Will it show in any reasonably sized print? No. Can it be corrected? Yes, with Photoshop Sharpen or, even better, Topaz Sharpen AI. 
Would IBIS help? Only for camera shake, not for subject motion - which postprocessing will.

Not quite. To see blur it must cross more than one pixel. Smaller pixels show blur slightly differently to the same blur across a larger pixel. This is because the blur can travel across more pixels, maing blur more apparent. It's a subtle difference but a real one. When downsizing from 60MP to 18MP (assuming same sensor area) blur will be entirely output dependant (essentially pixel binning). When up-rezzing from 18 MP to 60MP blur will appear slightly less, as it travels across less pixels, but will be entirely offset by the lack of resolution.

Even the best of the current gen AI programs struggle once blur moves across more than a couple of pixels. They look like someone turned up the radius in sharpening because the program tries to equalise the colour over a broarder range of pixels. These programs work better on lower resolution files than larger ones.

Gordon

[Franka373]

4 hours ago, jaapv said:

Actually, the images at 60 MP are exactly as blurry as the images one made at 10,18, or 24 MP. It is just that the camera shows the existing user error whereas lower MP cameras hide it. (At 100%) Will it show in any reasonably sized print? No. Can it be corrected? Yes, with Photoshop Sharpen or, even better, Topaz Sharpen AI. 
Would IBIS help? Only for camera shake, not for subject motion - which postprocessing will.

I need to break this down to be sure I understand it.  
1. 60mp images are exactly as blurry (user error) as any lower mp image.

2. User error shows up more in the higher mp than the lower mp at 100% crop. 
3. Blur won’t show up in reasonably sized prints regardless of number of mp

4. User error blur can be corrected in software especially AI software

5. IBIS helps for user error but not subject motion. 
 

A few comments and questions.   1 and 2 seem to negate each other.  For 2 a 100% crop would be what print size at a given mp count?  Is there a way to calculate print size?  For 3 what is a reasonably sized print?   Reasonably is not defined.  Is 4 always true even for higher mp cameras? 2 seems to support what most are saying is that the higher the mp count the more important it is to use good technique and higher shutter speeds knowing that each person has a limit in ability to make sharp photos at slower shutter speeds which seems to negate 1 since more mp are more sensitive to human error therefor blur can’t be exactly the same regardless of mp count.  In other words, lower mp is more tolerant of human error therefore blur is less, not exactly the same for any mp count based on 2. And I think it is because in a lower mp count for the same size sensor the pixel is bigger.  I am certainly not an expert.  I am just trying to work through the logic which I admit may be wrong.  

[jaapv]

It is very simple. If the amount of blur falls within the pixel size it will not be recorded. If the pixels get smaller they will resolve a smaller amount of blur. Yes you can calculate maximum  print size from pixel numbers, depending on what DPI you are printing at.  Reasonable print size is related to viewing distance. And the size of your wall. 100%crop is a term used for viewing the image on a screen at pixel level. 4 is true for any pixel count, depending on the amount of blur. Forensic programs like Focus Magic can take it even further but the results are not always esthetically pleasant. 

[FlashGordonPhotography]

5 hours ago, Franka373 said:

I need to break this down to be sure I understand it.  
1. 60mp images are exactly as blurry (user error) as any lower mp image.

2. User error shows up more in the higher mp than the lower mp at 100% crop. 
3. Blur won’t show up in reasonably sized prints regardless of number of mp

4. User error blur can be corrected in software especially AI software

5. IBIS helps for user error but not subject motion. 
 

A few comments and questions.   1 and 2 seem to negate each other.  For 2 a 100% crop would be what print size at a given mp count?  Is there a way to calculate print size?  For 3 what is a reasonably sized print?   Reasonably is not defined.  Is 4 always true even for higher mp cameras? 2 seems to support what most are saying is that the higher the mp count the more important it is to use good technique and higher shutter speeds knowing that each person has a limit in ability to make sharp photos at slower shutter speeds which seems to negate 1 since more mp are more sensitive to human error therefor blur can’t be exactly the same regardless of mp count.  In other words, lower mp is more tolerant of human error therefore blur is less, not exactly the same for any mp count based on 2. And I think it is because in a lower mp count for the same size sensor the pixel is bigger.  I am certainly not an expert.  I am just trying to work through the logic which I admit may be wrong.  

Blur is in the eye of the beholder.....

All of the above is subjective. You're more or less steady than I am and you shake differently to me. Your eye are better or worse than mine. You typically view prints from closer or further than I am. Your monitor at 100% is not necessarily the same size as mine. The subjects you photograph have typically more or less detail in them than the ones I shoot.

It's all about you.....

The only way to answer your questions to your satisfaction is to do your own testing. Basically print tests. Or the screens you typically view images on. Whatever output is normal to you is the benchmark you should be aiming for. Nothing more. You're not trying to please anyone else, unless they're paying you. What I rquire may not be what you require. You need to do your own testing.

If you need a number then that number is 240.....

240 ppi is generally accepted as a safe number for determining print size for a detailed scene. Kids can probably see more detail but most adults can't from a normal viewing distance. Divide your resolution in pixels by 240 for a rough idea of maximum print size. But consider it a start point, not an all encompassing answer. (That'd be 42.  )

But since you asked.....

This is my methodology for my needs. I print to A0. More often A1 but occasionally A0 or sometimes 24"x 65". I have a 24" printer. So I test to that standard. I'll take a shot on a tripod and then repeat the same shot handheld. Then I print segments resized to A0 on A4 paper (part of the image). At the point I'm struggling to see the difference I have my base limit. I'll also look at accessories or technique to see where I can improve. Now I know my threshold for general shooting to that standard. I also know shots of my kids aren't going to be at 60MP or printed at A0. So I interpolate my settings to reflect that. I also don't care if they're not technically perfect. I do as much family photography on Polaroid, Instax or my phone as my M. My standards are adjusted to suit.

Gordon

p.s. To your points for no1 add "at the same output size", to the end. For no4 stick a "sometimes" in the middle. Delete no3. That depends on how shaky you are and environmental conditions.

[harmen]

Advertisement (gone after registration)

The idea that blur that falls within the pixel size is not recorded is incorrect and has been discussed here many times.  What is correct is that once you get to that small level of blur, the limitations in the resolution will be of a similar level and as a result the recorded blur is not as obvious.

[IkarusJohn]

10 hours ago, FlashGordonPhotography said:

Not quite. To see blur it must cross more than one pixel. Smaller pixels show blur slightly differently to the same blur across a larger pixel. This is because the blur can travel across more pixels, maing blur more apparent. It's a subtle difference but a real one. When downsizing from 60MP to 18MP (assuming same sensor area) blur will be entirely output dependant (essentially pixel binning). When up-rezzing from 18 MP to 60MP blur will appear slightly less, as it travels across less pixels, but will be entirely offset by the lack of resolution.

Even the best of the current gen AI programs struggle once blur moves across more than a couple of pixels. They look like someone turned up the radius in sharpening because the program tries to equalise the colour over a broarder range of pixels. These programs work better on lower resolution files than larger ones.

Gordon

This is a helpful summary.  But it also needs to be clear, contrary to what might be concluded from Jaap’s comment, that for a given print or viewing distant, all other things being equal (other than ISO or aperture, obviously), one needs to use a faster shutter speed when using a 60MP full frame sensor than one would with a 24MP full frame sensor.  This is not just a pixel peeping issue.

While blur is increased with detail crossing smaller pixels, the implication in Jaap’s post is that it doesn’t matter for printing/viewing size and distance, this is clearly not the case in practice.  If such blur is lost in the detail, then for practical purposes, there would be no need to take greater care and higher shutter speeds, as Gordon recommends.

However, if we look at four common sensors, the pixel sizes would suggest that this should be a bigger problem with Leica’s APS-C cameras:

• TL2/CL - 24.2 MP APS-C 3.92 µm
• M10 - 24MP full frame 6.05 µm
• M11 - 60 MP full frame 3.76 µm 
• X1D ii - 50MP medium format 5.29 µm

 

[aristotle]

3 minutes ago, IkarusJohn said:

 If such blur is lost in the detail, then for practical purposes, there would be no need to take greater care and higher shutter speeds, as Gordon recommends.

 

Actually, I'm not sure that's what he meant.  He was considering two images that had the same physical displacement of the sensor during the capture process.  One image is captured at 18MP and the other was captured at 60MP.  He was saying (I think) that if you upsampled the 18mp image to 60mp, you wouldn't notice the blur in the upsampled image as much as you would in the native 60mp image, but that's because the upsampled image doesn't have as much detail.

If you are generating a 5x7 print from either, you'll be downsampling both.  The required shutter speed to produce acceptably small camera induced motion artifacts would be the same in either case.  If you are generating a print that is large enough that it would require upsampling of the 18MP image to print at something on the order of 300dpi, and you want to really achieve the advantage of the 60mp sensor in that case, then you would in fact need to use a faster shutter speed in order to realize the gain.  The print size where this matters though is quite large.  

[harmen]

Again, there is no magic amount of movement below which it suddenly becomes invisible.  It is always recorded.  See post #129

Secondly, physical size of the pixel has nothing to do with.  We’re talking about light coming through a lens and being projected on the sensor.  Whether that sensor is 1cm across or 5cm does not have any impact.  If you rotate your camera during the capture, the amount of blur is a factor of the angle divided by the angle of the lens, e.g., a 28mm would show it less than a 90mm.  If you move your camera sideways during the capture, the amount of blur is a factor of that distance divided by the total distance captured left to right in the image.  Again a 28mm would show it less than a 90mm, however you will now also notice the blur will be more for nearby objects, while the moon would stay perfectly sharp.

[harmen]

11 hours ago, Franka373 said:

I need to break this down to be sure I understand it.  
1. 60mp images are exactly as blurry (user error) as any lower mp image.

2. User error shows up more in the higher mp than the lower mp at 100% crop. 

Correct.  And they do not negate each other.  '2' simply means that together with the overall enhanced level of detail for the image, you can now also study your blur in more detail.

The amount of blur is no different.  Blur, whether from moving the camera or from the object moving, is simply what it is.  How could it not be?  It's not like the camera is aware of the sensor within and physically moves less or more.  It doesn't even matter if you're talking about the tiny sensor in a mobile phone or a plate in a big view camera.  Movement is movement.  What is recorded is the projected image, so movement is always about movement of the projected image.  A 28mm lens on full frame captures 75 degrees from left to right.  If you rotate 37.5 degrees during the capture, you will see movement worth half the size of the image.  The projection doesn't care about the size of the sensor or the size of the pixels.

A camera with more pixels can capture this movement more precisely.  You can zoom in a tad further and see the impact more clearly.  However it's not like it would not have been recorded if only you had fewer pixels.  An analogous situation is when you capture detail that is smaller than one pixel.  It actually shows up.  You just can't see it very precisely.  Think about the converging lines on test charts.  At some point you will see that the camera cannot distinguish clearly between lines any more.  Does that mean that the image will suddenly turn pure white?  No it does not.  It still captures those lines, but as a grey clump.  They haven't gone; they are just less easy to study precisely.  When you take a photo of a forest, the smallest branches do not disappear; they just show up as brown tinting.  Going back to how blur is recorded, imagine these branches moving.  When they move the tinting moves to a different place.  Does it matter much?  No, since it's such a small part of the final image you're printing and hanging on the wall.

[harmen]

2 hours ago, IkarusJohn said:

 If such blur is lost in the detail, then for practical purposes, there would be no need to take greater care and higher shutter speeds, as Gordon recommends.

This is true, but only if you don't mind hanging a picture on your wall from an original image with a lower number of pixels per inch.  You could scale up and print a 6MP image on A0 and enjoy the impressionist vibe without worrying about the lack of detail.  Then movement is indeed less of a concern.

 

Edit: let me add that I personally enjoy the vibe of grainy film and have far more film pictures than digital images hanging on my wall.  The above is just to add to the technical discussion here.

Edited March 29, 2022 by harmen

[IkarusJohn]

2 hours ago, harmen said:

This is true, but only if you don't mind hanging a picture on your wall from an original image with a lower number of pixels per inch.  You could scale up and print a 6MP image on A0 and enjoy the impressionist vibe without worrying about the lack of detail.  Then movement is indeed less of a concern.

 

Edit: let me add that I personally enjoy the vibe of grainy film and have far more film pictures than digital images hanging on my wall.  The above is just to add to the technical discussion here.

I guess this gets us into cropping.

If the viewer wishes to see the entire image, then there is an optimal viewing distance - the calculations have been posted before, but simplistically, you stand back for the larger images, equalising the pixel level detail.  Unless you just want wallpaper ...

PS - I should have said, I'm not sure about huge images.  1 metre across the long side is about as large as I want, and then it is viewed from the other side of the room.  

Where this gets me is I really don't see the point in 60MP.  I choose my lenses for their field of view and I don't really like cropping (there are exceptions, obviously), but framing happens in my head when I take the picture.  Lens selection only involves field of view partially - depth of field and lens character (foreshortening and wide view etc) are probably as important.

Edited March 29, 2022 by IkarusJohn

[harmen]

I'm with you.  Even though I now own an M11 and shoot at 60MP, I print A3+ at the most and have to ask myself why I don't save some memory and switch to a lower MP setting.  I bought the M11 because it has fewer limitations, especially those related to dynamic range and colour.  Luckily 60MP does not make my photos any more blurry, so it's not a drawback.

[IkarusJohn]

13 hours ago, harmen said:

I'm with you.  Even though I now own an M11 and shoot at 60MP, I print A3+ at the most and have to ask myself why I don't save some memory and switch to a lower MP setting.  I bought the M11 because it has fewer limitations, especially those related to dynamic range and colour.  Luckily 60MP does not make my photos any more blurry, so it's not a drawback.

Let me ask this - had the M11 had better dynamic range (lower base ISO, better highlight recovery etc), more malleable DNG files etc, but “only” been 40MP, would you still have bought it?

[Tailwagger]

1 hour ago, IkarusJohn said:

Let me ask this - had the M11 had better dynamic range (lower base ISO, better highlight recovery etc), more malleable DNG files etc, but “only” been 40MP, would you still have bought it?

Assuming all the other stuff, larger battery, new higher res EVF, etc. in my case, yes. I personally think that at the moment 40 Mpx is the sweet spot, given I rarely print past 17x22 and still retain crop-ability. But I honestly don't see the added 20M pixels in any way as burdensome and certainly not a reason to buy or reject. As folks ultimately came to recognize with the 10-R, there was a lot more to it than just 16M extra pixels over the 10, this is even more the case with the M11, IMO. 

[IkarusJohn]

And losing the baseplate …

[SrMi]

2 hours ago, IkarusJohn said:

Let me ask this - had the M11 had better dynamic range (lower base ISO, better highlight recovery etc), more malleable DNG files etc, but “only” been 40MP, would you still have bought it?

Yes. I do not care if the new sensor is 60MP or 40MP. However, Leica cannot pick the resolution, it has to use what is available.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[IkarusJohn]

7 minutes ago, SrMi said:

Yes. I do not care if the new sensor is 60MP or 40MP. However, Leica cannot pick the resolution, it has to use what is available.

Source?  

[SrMi]

6 minutes ago, IkarusJohn said:

Source?  

Latest Sony FF sensors are available either in 33MP or 60MP versions.

Edited March 29, 2022 by SrMi