The logic of high MP and cropping photos?

LancsMak said:
But, and this is where I either validate my thinking or show my ignorance @sk66 (and there is much ignorance to show :) ) if (on ff body) I use a longer focal length on the lens to achieve the same equivalent focal length as on the crop sensor, then the larger sensor still provides the light advantages? E.g. Shot at 135 on a crop and 200 on a ff.
Click to expand...
But if you’re shooting sports, my guess is that you’re often at 200mm on the M50, and to get the same frame on the R6, you’re gonna need to crop a lot. And it’s within that crop that you lose all the advantages you just paid all that money for. You’ll be throwing away roughly half your pixels.
 
Phil V said:
But if you’re shooting sports, my guess is that you’re often at 200mm on the M50, and to get the same frame on the R6, you’re gonna need to crop a lot. And it’s within that crop that you lose all the advantages you just paid all that money for. You’ll be throwing away roughly half your pixels.
Click to expand...
No, mostly 135 is the furthest I go, rarely to 200.
 
LancsMak said:
No, mostly 135 is the furthest I go, rarely to 200.
Click to expand...
Really?
What are you shooting?

My ‘sports’ images are mostly shot on crop and mostly I was at the 200mm end of my 70-200, except on the rare occasions I shoot indoor sports (squash, badminton etc) when I’ll often shoot really wide.
 
Thanks to everyone that has helped me in this thread. Just pulled the trigger on a R6 mk2 from Panamoz. My first full frame and my first use of Panamoz thanks to this forum!

Mind you half the savings I got from buying through Panamoz have been spent on gifts for the other half to keep her sweet with me buying another body :ROFLMAO:
 
The Gods must be placated. You don't want to incur their wrath. :D
 
LancsMak said:
Interesting. So purely because you're looking more closely at the image you're now able to see some small elements that looked in focus are actually not adequately in focus to be acceptable when enlarged.

Destroys my thinking of shooting at shorter focal length and cropping... Just be the benefit of zoom lens and full frame light transmission then.
Click to expand...
Keep the entire DOF concept to simply remember that the base concept is that a 'point' can blurred so that resembles a circle, but if that circle is small enough it will fool your eye and brain into thinking 'in focus'...yet when that clrcle is enlarged -- by any means -- the eye and brain will detect it as 'not a point,...out-of-focus'. Why it is called the Circle of Confusion.

As for high MP = more ability to enlarge and crop...
The lens will only deliver a certain amount of detail resolution across the full expanse of the frame, so when you enlarge/crop the frame, you are throwing away detail resolution delivered by the lens, and you will not get finer detail within the piece that you chose to retain!
 
Last edited:
High MP is very useful to me as much of my photography is snap shots of people and vehicles without the time to compose so the ability to crop and frame later is a must.
 
LancsMak said:
Ok, so looking to test my thinking with the collective knowledge and experience on here. To say I've become infatuated with photography over the last year would be a gross understatement, and I'm giving serious consideration to having a two-camera setup for sports events. The question is whether my reasoning is sound....

- Currently a Canon M50 Mk2 with a f2.8 70-200 Mk2
- Debating getting a Canon R5 and adding a 24-70 to the mix. I'd swap the lenses over though and run the 70-200 on this mostly.

What's my logic?

Hopefully the split between the 24-70 and 70-200 is self explanatory in terms of being able to just switch as I need. The rationale for larger MP and cropping is that I often shoot down at f2.8 (or 3.2 if I can to minimise softness / still keep a fast aperture) due to indoor lighting. At best I'm after shutter speeds of 1/800 panning or 1/1600 headshots. So rather than a close-zoomed shot at 20MP with a very very shallow DoF (I manage this maybe 50% of the time, the rest are binned) my thinking is a wider shot at 45MP with a deeper DoF and then cropping down.

Smart thinking, or utter rubbish?

Thanks in advance.
Click to expand...
Your logic only stays correct if you get a pixel perfect image in your target crop area. That means:
  • Absolutely no shake. It usually means tripod + shutter release
  • Absolutely no subject movement unless intentional. That means fairly high shutter speeds
  • Low iso to keep noise reasonable
  • aperture maximum of f/8 to avoid resolution robbing diffraction
  • Lenses that are capable of resolving 8K. EF zooms are not the best for this, particularly outside of centre. 30MP 5DIV may be the ceiling for most of them, if that.
It is so easy to fail on at least one of these points, which alone can invalidate the whole concept and you are instantly down to 20MP effective resolution. 20-something MP cameras are very very forgiving for minor mistakes and flaws up to a certain point.

The two zooms in question are the ones that I would in fact avoid using on 5Ds / R5 as much as possible, and actually got rid of the wider one altogether for a couple of primes.
 
to paraphrase, your logic may be quite sound if you have let's say 300mm f/4 or even 2.8 prime and want to get away with it instead of buying 500mm f/4. With a good technique that would probably work as an upgrade from something in 20MP range but obviously wouldn't give the same background blur and look if that was important.
 
LongLensPhotography said:
Your logic only stays correct if you get a pixel perfect image in your target crop area. That means:
  • Absolutely no shake. It usually means tripod + shutter release
  • Absolutely no subject movement unless intentional. That means fairly high shutter speeds
  • Low iso to keep noise reasonable
  • aperture maximum of f/8 to avoid resolution robbing diffraction
  • Lenses that are capable of resolving 8K. EF zooms are not the best for this, particularly outside of centre. 30MP 5DIV may be the ceiling for most of them, if that.
It is so easy to fail on at least one of these points, which alone can invalidate the whole concept and you are instantly down to 20MP effective resolution. 20-something MP cameras are very very forgiving for minor mistakes and flaws up to a certain point.

The two zooms in question are the ones that I would in fact avoid using on 5Ds / R5 as much as possible, and actually got rid of the wider one altogether for a couple of primes.
Click to expand...

I mind something along these lines being raised by Steven (SFT) as well, Is there a list or specific lens data that shows what lenses can properly resolve 8k cameras as I think mine is one (D810)?
 
gman said:
I mind something along these lines being raised by Steven (SFT) as well, Is there a list or specific lens data that shows what lenses can properly resolve 8k cameras as I think mine is one (D810)?
Click to expand...
It is a combination of QC which is per specific copy, MTF curve, resolution tests, etc. Primes have better chance than zooms, newer primes have better chance than older primes, longer primes have much much better chance than wide primes, etc.

D810 is midway there so will absorb more flaws without showing them, and if intention is cropping and cropping almost exclusively, then you may disregard the extreme corners but why not just buy a slightly longer one then instead?

Here is a little compilation of what me and Steve are fairly happy with, in DSLR mounts for landscape / infinity focus use:

Sigma 28mm ART (SFT)
Tamron 35mm f/1.4 - outstanding
Sigma 35mm f/1.4 ART - ideally around f/8
Sigma 40mm ART (SFT)
Sigma 50mm f/1.4 ART - ideally 5.6+
Sigma 85mm f/1.4 ART - from f/4+
Sigma 105mm f/1.4 ART - not really had much experience with this
Sigma 135mm f/1.8 ART - basically anything goes, DOF permitting.
Pretty much any recent prime at or over 200mm.

Zeiss Otus is a safe bet also. Lesser Zeiss you need to test.

Steve is also happy with his Nikon 70-200 2.8 E.

I can see a bit of flaws in Canons EF 70-200s ranging from lens to lens but they never ever get to ART level. They are best used for sports, etc. F4 IS mk1 is surprisingly good at wide end but doesn't maintain that throughout. Newer ones are reverse with peak around the middle. I think the old 200mm f/2.8 prime is a big leap from them if you can put up with long MFD and lack of zoom.

A very notable absence is the lack of anything wide on that list. I haven't tested Sigma 14mm, or Canon 24 TSE II. These would be the best contenders. 16-35mm III is annoyingly good at 20MP from get go without showing any real improvement after stopping down, but has precious little left to show at higher resolutions. SFT says ART 20 and 24mm primes are also not as good as longer ones. Whether that means good at f/8 or never I don't really know but my list basically improves sequentially with their release date and that tells something about these 2.
I looked at TDP samples long and hard and it appears the mid-range E-mount Contemporary Sigma DN 20 and 24mm f/2 may be the best options out there, with 1.4 samples being considerable poorer. Of course one needs a sony body to match and that is something I am seriously consider to access more quality lenses and perhaps some acceptable modern zoom
 
Last edited:
The reality is that cropping always reduces the quality compared to an uncropped image at what ever size.
The only real question is whether we would still be satisfied with this lowered quality.

Some of us only ever see our images at their maximum size and quality When pixel peeping, however in many respects such an image probably looks at its best at around 50% of its maximum pixel size.

I suspect that most of us would get accustomed to the quality obtained from an uncropped or minimally cropped image.
And would soon become dissatisfied with more heavily cropped images.
Were this not the cas longer focal length lenses would not have remained as popular as they are.
People like to get the best from what ever sensor size that they are using, and that is not achieved by cropping.
 
Terrywoodenpic said:
The reality is that cropping always reduces the quality compared to an uncropped image at what ever size.
The only real question is whether we would still be satisfied with this lowered quality.

Some of us only ever see our images at their maximum size and quality When pixel peeping, however in many respects such an image probably looks at its best at around 50% of its maximum pixel size.

I suspect that most of us would get accustomed to the quality obtained from an uncropped or minimally cropped image.
And would soon become dissatisfied with more heavily cropped images.
Were this not the cas longer focal length lenses would not have remained as popular as they are.
People like to get the best from what ever sensor size that they are using, and that is not achieved by cropping.
Click to expand...
Cropping has no effect on image quality, only on what appears in the image.
If you take a given image, print at A3, then cut the print into two A4 halves, these will be identical to taking two copies of the same image, cropping one the left 50% half, the other to the right 50%, and printed at A4.

What can change the perceived quality is comparing a cropped and uncropped version of an image at the same physical size - such if you took the same A3 print and compared it to an A3 print of one of the half crops. The greater the crop, the greater any perceived difference will be.
 
Faldrax said:
Cropping has no effect on image quality, only on what appears in the image.
If you take a given image, print at A3, then cut the print into two A4 halves, these will be identical to taking two copies of the same image, cropping one the left 50% half, the other to the right 50%, and printed at A4.

What can change the perceived quality is comparing a cropped and uncropped version of an image at the same physical size - such if you took the same A3 print and compared it to an A3 print of one of the half crops. The greater the crop, the greater any perceived difference will be.
Click to expand...

You state the obvious. However if you are using a crop to substitute for an appropriate focal length. The it most definitely has an effect if printed to the same size. Which was the purpose of the thread.
 
Terrywoodenpic said:
You state the obvious. However if you are using a crop to substitute for an appropriate focal length. The it most definitely has an effect if printed to the same size. Which was the purpose of the thread.
Click to expand...
Actually the OP was looking to compare a 20Mp shot with a crop of a wider 45Mp shot - which is a more complex question.
In theory, if the lens used on the 45Mp camera was able to fully resolve the 45Mp, then cropping this to 20Mp should result in an image that can be printed the same size as the 20MP camera image with the same quality in both.
 
LongLensPhotography said:
It is a combination of QC which is per specific copy, MTF curve, resolution tests, etc. Primes have better chance than zooms, newer primes have better chance than older primes, longer primes have much much better chance than wide primes, etc.

D810 is midway there so will absorb more flaws without showing them, and if intention is cropping and cropping almost exclusively, then you may disregard the extreme corners but why not just buy a slightly longer one then instead?

Here is a little compilation of what me and Steve are fairly happy with, in DSLR mounts for landscape / infinity focus use:

Sigma 28mm ART (SFT)
Tamron 35mm f/1.4 - outstanding
Sigma 35mm f/1.4 ART - ideally around f/8
Sigma 40mm ART (SFT)
Sigma 50mm f/1.4 ART - ideally 5.6+
Sigma 85mm f/1.4 ART - from f/4+
Sigma 105mm f/1.4 ART - not really had much experience with this
Sigma 135mm f/1.8 ART - basically anything goes, DOF permitting.
Pretty much any recent prime at or over 200mm.

Zeiss Otus is a safe bet also. Lesser Zeiss you need to test.

Steve is also happy with his Nikon 70-200 2.8 E.

I can see a bit of flaws in Canons EF 70-200s ranging from lens to lens but they never ever get to ART level. They are best used for sports, etc. F4 IS mk1 is surprisingly good at wide end but doesn't maintain that throughout. Newer ones are reverse with peak around the middle. I think the old 200mm f/2.8 prime is a big leap from them if you can put up with long MFD and lack of zoom.

A very notable absence is the lack of anything wide on that list. I haven't tested Sigma 14mm, or Canon 24 TSE II. These would be the best contenders. 16-35mm III is annoyingly good at 20MP from get go without showing any real improvement after stopping down, but has precious little left to show at higher resolutions. SFT says ART 20 and 24mm primes are also not as good as longer ones. Whether that means good at f/8 or never I don't really know but my list basically improves sequentially with their release date and that tells something about these 2.
I looked at TDP samples long and hard and it appears the mid-range E-mount Contemporary Sigma DN 20 and 24mm f/2 may be the best options out there, with 1.4 samples being considerable poorer. Of course one needs a sony body to match and that is something I am seriously consider to access more quality lenses and perhaps some acceptable modern zoom
Click to expand...


Thanks for the detailed replied, that's appreciated. I'm still thinking about the Sigma 40mm 1.4 ART F mount, but I'm guessing I'll need the dock as well?
 
gman said:
Thanks for the detailed replied, that's appreciated. I'm still thinking about the Sigma 40mm 1.4 ART F mount, but I'm guessing I'll need the dock as well?
Click to expand...
If you ask Steve, 40mm is probably the best lens under 100mm going for F mount. Just make sure you are happy with it 100% at f/1.4 before accepting. If you are not, then it is pointless buying and carrying it, ie. you can have much smaller and cheaper (used) 50mm instead.
I may [try to] get [another] one some day but I'm fairly well covered with Tamron 35mm and Sigma 50mm in the range. 16-24 is more in need of something failproof and I don't see any way out on Canon ecosystem, new or old.

You will only really need the dock for firmware update(s). If it comes with one already you probably don't need it. They are like £20 on ebay so hardly an issue.
I'm generally against having to adjust focus preferences particularly in docks. This usually indicates something is misaligned and not behaving right, and you are better off buying another that is set straight. Or in case of lesser lenses they may be hazy and soft wide open and AF systems has hard time period, but then you will just see inconsistencies instead; not anything than can be consistently fixed until you simply get the Z thing.
There is no need for any of it other that update on mirrorless.
 
Faldrax said:
Actually the OP was looking to compare a 20Mp shot with a crop of a wider 45Mp shot - which is a more complex question.
In theory, if the lens used on the 45Mp camera was able to fully resolve the 45Mp, then cropping this to 20Mp should result in an image that can be printed the same size as the 20MP camera image with the same quality in both.
Click to expand...
And as I explained in detail you need very nice lenses, mainly primes with great technique to get there.

EF 70-200mm f/2.8 II or III (same thing, different shade of cream) will NOT cut it. I spent some time with mine in the floods today (well just outside). It resolves a very nice 20MP (when scaled down in this case), but it does not resolve anywhere near 50MP outside of centre. And there is a very nasty focus shift going from 200mm f/2.8 to f/4 meaning you use it wide open or at f/8+ but not in between. 85mm f/1.4 walks circles all over it at f/3.2. I'm thinking this through if the use on R6 is enough to justify keeping it.
 
LongLensPhotography said:
Your logic only stays correct if you get a pixel perfect image in your target crop area.
Click to expand...
The logic is not correct...period.
You cannot crop an image and maintain the same DOF and perceived sharpness at the same size output. You can easily end up at the same point; but then there's no benefit.

LongLensPhotography said:
Lenses that are capable of resolving 8K.
Click to expand...
An 8k sensor (33MP +) will not resolve 8k of resolution unless the lens is resolving far more... every stage incurs losses; the lens (almost?) always incurs the greatest losses. A high resolution sensor w/o an AA filter might be .9x of the lens' resolution, but still a loss.
 
LongLensPhotography said:
If you ask Steve, 40mm is probably the best lens under 100mm going for F mount. Just make sure you are happy with it 100% at f/1.4 before accepting. If you are not, then it is pointless buying and carrying it, ie. you can have much smaller and cheaper (used) 50mm instead.
I may [try to] get [another] one some day but I'm fairly well covered with Tamron 35mm and Sigma 50mm in the range. 16-24 is more in need of something failproof and I don't see any way out on Canon ecosystem, new or old.

You will only really need the dock for firmware update(s). If it comes with one already you probably don't need it. They are like £20 on ebay so hardly an issue.
I'm generally against having to adjust focus preferences particularly in docks. This usually indicates something is misaligned and not behaving right, and you are better off buying another that is set straight. Or in case of lesser lenses they may be hazy and soft wide open and AF systems has hard time period, but then you will just see inconsistencies instead; not anything than can be consistently fixed until you simply get the Z thing.
There is no need for any of it other that update on mirrorless.
Click to expand...

Thanks, plenty to think about and it sounds like when I do actually buy this lens the most important thing is to buy from somewhere highly reputable so no hassle to return a bad copy?
 
gman said:
Thanks, plenty to think about and it sounds like when I do actually buy this lens the most important thing is to buy from somewhere highly reputable so no hassle to return a bad copy?
Click to expand...
Pretty much or ideally test on the spot
I have actually returned plenty to eBay :) but that one wasn't from there. I'm sure I was particularly unlucky with it as all the other arts I bought are optically fine.
 
sk66 said:
The logic is not correct...period.
You cannot crop an image and maintain the same DOF and perceived sharpness at the same size output. You can easily end up at the same point; but then there's no benefit.


An 8k sensor (33MP +) will not resolve 8k of resolution unless the lens is resolving far more... every stage incurs losses; the lens (almost?) always incurs the greatest losses. A high resolution sensor w/o an AA filter might be .9x of the lens' resolution, but still a loss.
Click to expand...
You are overthinking it way too deep. This is not a scientific experiment where you must maintain all equal settings. If you need to stop down an extra half or full stop you do. If you need sharper lens you buy that. No reason why 8k file cannot be pin sharp, only that it is not so easy.

By the way 8k = 45mp, so quite a bit more of everything
 
LongLensPhotography said:
You are overthinking it way too deep.
Click to expand...
The original idea was to use a WA on a higher resolution sensor in order to get a deeper depth of focus and increased sharpness; for indoor sports...
What is the definition of "pin sharp?" To me it means the image doesn't lose detail/sharpness before pixelation is visible; and yes, that is very difficult at high resolutions.

LongLensPhotography said:
By the way 8k = 45mp, so quite a bit more of everything
Click to expand...
The standard 8k format is 7680x4320...
 
sk66 said:
The standard 8k format is 7680x4320...
Click to expand...
A letterbox crop video format that you can't find in any mainstream stills or hybrid camera sensor. That's not only not useful but meant to confuse people reading the thread. We get 3:2, or 4:3. And in the more common former case that is 45MP. End of discussion.

sk66 said:
The original idea was to use a WA
Click to expand...
70-200 = WA. Ok fine. Not everyone is lucky enough to have a collection of 400mm+ primes.

sk66 said:
in order to get a deeper depth of focus and increased sharpness; for indoor sports...
Click to expand...
paraphrasing here a bit... going off tangent actually. All that OP wanted was useful images, not something scientific equivalent on a whole number of points.
 
sk66 said:
The original idea was to use a WA on a higher resolution sensor in order to get a deeper depth of focus and increased sharpness; for indoor sports...
Click to expand...
And that starting point itself becomes a limited benefit expectation! Let's analyze...assuming 24mm lens that resolves as well (100 lne-pairs per mm of film) as delivered with a 96mm lens (4 * FL) (96mm FL has 1/4 the FOV of the 24mm FL) for the purposes of discussion and analysis...and let us assume a distance to subject of 5m, the 135 format frame sees 5m tall area with 24mm FL

So...
  1. with 24mm film height, 1m tall object takes up 20% of the frame height, occupying 4.8mm of film and having 480 line-pairs of detail resolution across the 1m subject.
    The DOF zone depth with 24mm f/4 lens is 4.26m (note this is not using the too-liberal interpretation of human visual acuity assumed by 'manufacturer standard DOF assumption!)
  2. with 100mm FL, the 1m tall object takes up 85% of the frame height, occupying 20.4mm of film and having 2040 line-pairs of detail resolution across the 1m subject
    The DOF zone depth with 100mm f/4 lens is 0.21m
  3. with 24mm lens shot (#1) cropped to same 85% frame-fill as shot with 100mm (#2), it still only has 480 line-pairs of detail resolution for the 1m subject, enlarged to fill the same print size as #2...only 23.5% of the detail resolution as the 100mm FL shot!
    And the DOF zone depth in the cropped shot is 0.93m
Change the above from film to digital, the 'line-pairs per millimeter of sensor' is the only thing different from 'line-pairs per millimeter of film'
 
Last edited:
wiltw said:
Change the above from film to digital, the 'line-pairs per millimeter of sensor' is the only thing different from 'line-pairs per millimeter of film'
Click to expand...
And the cropped image also only contains ~ 25% of the light originally recorded and which still exists in the 100mm image.
I.e. image noise will be much more significant (apparent); which can be a big hit on image resolution, especially for indoor sports (high ISO).
 
sk66 said:
And the cropped image also only contains ~ 25% of the light originally recorded and which still exists in the 100mm image.
I.e. image noise will be much more significant (apparent); which can be a big hit on image resolution, especially for indoor sports (high ISO).
Click to expand...
What is important is the light ' photons' falling on a unit area. Not the total light falling on a sensor. If you crop an image that photon distribution is the same. Smaller sensor sites inevitably capture fewer photons when the light intensity is the same. Noise is mostly the result of capturing fewer photons for a given area, for what ever reason.
 
This whole discussion has become about the perception of "quality".

It may be worth asking, in each case, how important such things as sharpness and detail are to the viewer or even if most viewers will care at all about such matters. If your audience are those who examine images with 10x magnifiers, then the foregoing is relevant. If your audience are the tens of thousands of other people who just want to be shown something, then I suggest that most of the foregoing is completely irrelevant.

As one historian so neatly put it: "In world war 2, Germany made the best tanks and fitted them with the finest guns. America, Britain and the Soviet Union made adequate tanks and fitted them with adequate guns. Why did the allies win? because they made 20 times more tanks!"
 
Faldrax said:
Cropping has no effect on image quality, only on what appears in the image.
If you take a given image, print at A3, then cut the print into two A4 halves, these will be identical to taking two copies of the same image, cropping one the left 50% half, the other to the right 50%, and printed at A4.

What can change the perceived quality is comparing a cropped and uncropped version of an image at the same physical size - such if you took the same A3 print and compared it to an A3 print of one of the half crops. The greater the crop, the greater any perceived difference will be.
Click to expand...

Exactly it is about magnification.
 
The ultimate question is whether the cropped image is good enough for the intended purpose. It doesn't really matter if an uncropped image would have higher quality if this goal is met. Nobody in the real world is going to view a cropped and uncropped image side by side
 
AndrewFlannigan said:
This whole discussion has become about the perception of "quality".
Click to expand...
Well, the original issue was the quality of the results achieved with current camera/lens, and the idea of adding a larger sensor camera with a wide angle lens to get "better results"... so yeah, the discussion is/was about "quality."

The question was not "can cropping be good enough?" Of course it can... and at a certain point it doesn't much matter what you do, because every option generates very similar results (cropping/crop sensor/TCs/longer lens).
 
Terrywoodenpic said:
What is important is the light ' photons' falling on a unit area.
Click to expand...
Yes, light per image area; not photosites/pixels.

If you record the same scene/subject (e.g. a light bulb in the dark) with the same composition and the same settings on a larger sensor; that IS more light... because either you were closer to it, or you used a longer lens with a larger aperture (same f#). And the smaller "crop sensor" will have recorded less light.

The corollary to that is; if you crop in post to change the composition to simulate having been closer, you are discarding light/image area... same as using a crop sensor at the time of recording. The exposure of what remains doesn't change. But when the crop is enlarged to the same size the remaining light (information, resolution, colors, noise, etc) is spread out farther... the enlarged image contains less light (good data) per area.
 
sk66 said:
Yes, light per image area; not photosites/pixels.

If you record the same scene/subject (e.g. a light bulb in the dark) with the same composition and the same settings on a larger sensor; that IS more light... because either you were closer to it, or you used a longer lens with a larger aperture (same f#). And the smaller "crop sensor" will have recorded less light.

The corollary to that is; if you crop in post to change the composition to simulate having been closer, you are discarding light/image area... same as using a crop sensor at the time of recording. The exposure of what remains doesn't change. But when the crop is enlarged to the same size the remaining light (information, resolution, colors, noise, etc) is spread out farther... the enlarged image contains less light (good data) per area.
Click to expand...
As I said in post 73 Magnification
Any deterioration is down to magnification.
 
You must log in or register to reply here.
Back
Top