[Bob 256]

My understanding was that a lens doesn't "resolve a certain amount of pixels". It's rather that you get diminishing returns when putting a lens in front of higher and higher megapixel count sensors, e.g. a lens might resolve (idk, just random numbers) 2000 lp/mm on a 25 mp sensor, but that might "drop" to 3500 on a 50 mp sensor (not proportional). Was that assessment wrong?

You're right. Pixels only apply to a sensor, while a lens is an "analog" sort of a thing. Details rendered by the lens don't just stop at some point. What happens is that the contrast between fine details begins to decrease (significantly at some lines per millimeter value) and falls off after reaching that point. After so much fall-off, there is little difference between lighter and darker details, so much so that it becomes harder and harder to distinguish the details. A sensor is simply interacting with this so the sensor electrical amplitude of these differences also falls off. Nonetheless, the MTF (modulation transfer function - a term engineers refer to which is a plot of contrast versus frequency) of a lens can be plotted and when this falls so low, no sensor, no matter how many pixels it has, can't see the details for noise. (It's interesting to note, photographic film's MTF can be plotted and when fine details beyond some value are recorded, they might be there but film grain creates "noise" which makes it impossible to see them).

Consider this: If you put a 3 Mpx sensor behind a lens capable of details that a 36 Mpx sensor can still record, what is limiting the actual number of pixels in the image? It would be the sensor of course. Now take a cheap lens from which any sensor beyond 10 Mpx can't extract additional details and use a 36 Mpx sensor. The lens is the bottleneck here and a lot of data space is just wasted to store the 36Mpx sensor output, since many adjacent pixels will be nearly identical due to the low MTF of the lens for those close spaced details. In this latter case, the lens becomes the limiting factor.

Lenses and sensors go hand in hand to determine the limit on fine image details but those limitations are best expressed as MTF for a lens, and pixel resolution for sensors.

[swanlefitte]

If you stand too close to a photo you can't see the forest through the pixels.

[biz-engineer]

You can make great large prints from 10 or 12 megapixels - ask most wedding photographers and there are plenty who actually think 12MP or 16MP are the sweet spot. 24 even allows you to "pixel peep" a large print. 100 megapixels is probably so you can analyze a print with a lab microscope? If that's a thing now....

We generally get such comment from people who never print. Yes, you can print 10 feet wide with 1 mega and if you stand far enough, 1 megapixels is good enough. There are many other things that people don't need in life. Do we really need a camera? Is that vital?

---------- Post added 04-03-21 at 05:17 ----------

More megapixels also means huge RAW files which are no fun to work with by the hundreds. And they eat up storage fast.

That is true. IMO, more megapixels means more work before the exposures, and much less images of much higher quality (both in terms of image quality and researched content). It's the approach opposite to "spray and pray".

---------- Post added 04-03-21 at 05:22 ----------

I agree with most all of you, that it gets to the point where the amount of megapixels gets to a point to where it makes no sense - except for extreme cropping.

I consider cropping is for bad photographers. A photographer who does his work before taking pictures don't have to crop that much. Not quite cost effective to use a high mega pixels camera and crop most of it.

---------- Post added 04-03-21 at 05:25 ----------

For storage its about 3750 shots per terabyte from the Fuji and about 25600 shots from a 24MP full frame.

3750 shots is way way beyond what a professional fine art photographer do. Some probably take a few hundred exposures per year, and a dozen of "keepers" per year. The quality of photographs goes down quickly with increasing quantities. At 25600 shots, I guess you can discard 99% of worthless files, maybe keep 256 images (1%) that have some interest, and 25 images that are really good. We can select images before we take them (like in film days, because each film exposure costs additional money..), or take a lot of pictures and think about them afterward, delete the ones we don't like (digital era).

---------- Post added 04-03-21 at 05:40 ----------

You need 50 mp for a 24"x36" (50cm x 90cm) print? Even for larger? I'm unconvinced.

We can definitely print 5x8 feet with 8 megapixels. I printed 16x24" with 2Mpixels. I have about 100 prints 24x36, made print K200 files, K5 files, K3 files and K1 files. I admit the prints from K1 files are much better, but better isn't a must have.

For people who prefer to print from less megapixels, there is always the option to downsize the image file before printing large, e.g downsize 50Mpixels to 5 Megapixels before printing 5x8 feet large, it's perfectly allowed.

[HippyHippo]

I still wonder what percentage of all these increasingly large RAW files only see the light of day when exported from Lightroom as small .jpgs to put on Instagram. I’d love to know.

[Benz3ne]

The one benefit I see from the mega-megapixels trend is the ability to have heavy crops with significant detail.
24mp (Sony a7iii) is what I've settled on but I'm always curious about the 42mp offerings from the A7Riii, truthfully.

HippyHippo does have a good point - there are undoubtedly plenty who have gone for the 'bigger is better' approach only to have a 1024 pixel width photo uploaded to FaceNovels or the suchlike.
Even though that's largely an approach I take (Instagram, as mentioned), I like having the scope to have larger prints. Evidenced by my brother and sister-in-law's maternity photo which they had printed reasonably large and their baby photo which was also printed fairly large, twice (one for my parents).

[Benz3ne]

I consider cropping is for bad photographers. A photographer who does his work before taking pictures don't have to crop that much. Not quite cost effective to use a high mega pixels camera and crop most of it.

I agree with this sentiment for the most part. There are some situations (as I'm sure some of us have found ourselves in) where we cannot physically get closer, are stuck with the lens that is on the camera and can see something up ahead that is worth photographing. At this point, it's not for the lack of effort or ability of the photographer, but having that additional crop-reach brings benefit of getting what could be a one-in-a-lifetime shot.

[microlight]

Although I typically try to frame shots for minimal (or any) cropping, I agree with the above logic for the ‘once-in-a-lifetime’ shot. Same goes for using very high ISO to get a OIAL shot that would otherwise be lost; I had one of these in New Zealand in 2018, using around ISO12000 with the 55-300PLM on the K-3II to get a picture of a penguin moulting in some undergrowth. In this case, any picture was better than no picture, but in the event it cleaned up pretty well.

But this is a different discussion to the point about printing large-size art pictures where you’re optimising all aspects of a shot to get a particular look - especially if your target is to generate a large physical print, which is something I hardly ever do.

[noelpolar]

Coming from a K1.... I'd like the next K1 to have the same pixel denisty of the K3iii..... mostly so I can just use/buy one DSLR camera for all the stuff I like to do... love landscapes but also love flighty mach box size birds.... can be doing both on a walk.... but... I guess.... for me.... it really doesn't matter much.... because, like Norm, I'm pretty happy with what I already have.

[BarryE]

I'm surprised nobody (might have missed it, coz I got a little bored with the "how many MP you need to print large" debate, again), but cramming more pixels onto a sensor reduces the DR. Yes, fancy processing can, to some extent mitigate this DR reduction, but there's a loss as the MP count climbs.

For me DR is a critical parameter. A pixel count of around 24 MP for FF would have been the DR/resolution sweet spot for me. The K-1 is great, but if it had been 24MP with even greater DR it would have been even better.

[ThorSanchez]

For people who prefer to print from less megapixels, there is always the option to downsize the image file before printing large, e.g downsize 50Mpixels to 5 Megapixels before printing 5x8 feet large, it's perfectly allowed.

Which, of course, makes about as much sense as buying a $500k Ferrari and putting a 35 mph governor on it. It might be a fun experiment if you have $100M in the bank and just don't care about money. But for the 99.9% it makes infinitely more sense to buy a Volkswagen. The VW covers almost every conceivable use case, and costs 1/30th as much, without considering cost of ownership. For 99.9% of photographers buying a 100Mp+ camera so they can theoretically, occasionally print 2x3m prints at 300 dpi is as much overkill as buying a Ferrari over a VW.

It's cool that machines with unbelievable capabilities and unreachable price tags exist. But for hobbyists almost all of them are completely irrelevant and wholly unnecessary.

---------- Post added 03-04-21 at 07:02 AM ----------

Although I typically try to frame shots for minimal (or any) cropping, I agree with the above logic for the ‘once-in-a-lifetime’ shot. Same goes for using very high ISO to get a OIAL shot that would otherwise be lost; I had one of these in New Zealand in 2018, using around ISO12000 with the 55-300PLM on the K-3II to get a picture of a penguin moulting in some undergrowth. In this case, any picture was better than no picture, but in the event it cleaned up pretty well.

But this is a different discussion to the point about printing large-size art pictures where you’re optimising all aspects of a shot to get a particular look - especially if your target is to generate a large physical print, which is something I hardly ever do.

The other side of that argument is that for many use cases you're better off with higher pixel density of a smaller format sensor with faster readout speeds than a 100Mp medium format camera. Medium format has never been known for autofocus capabilities or reach or frame rate, and if you're trying to capture an image of a far away animal without scaring it off you're probably better off with a camera designed for that situation rather than one that you might be able to do massive cropping on a wider angle shot. Even on full frame you'd need a 450mm lens to get the equivalent reach of the PLM on APS-C, so you're probably using a camera far less optimized for the shot that's much bigger, heavier and more expensive for little or no gain. Especially if the K-3 Mark III really does have equal or better high ISO performance compared to a K-1II.

All of these resolution and format arguments are about slightly moving the needle on a exponential or logarithmic curve. Each $1000 and pound of weight gets you a 2% better shot. In some edge cases.

[TonyW]

I understand that there has been a slight drift from megapixels to print resolution but they are still very closely related as they pertain to the native resolution of the camera in pixels. Printers also have a native resolution, actually they have 2!. The first being the requirement for the bitmap image they produce (spool) for any required print size measured in PPI. The PPI requirements are exactly the same regardless of your print size. Therefore a 20"x16" image will be printed with the same PPI as a 60"x47" print 360/720 PPI.

Also worth considering the real resolution of your Bayer sensor system is approximately half the declared MP So your 50MP system is only capable of resolving around 1/2 its pixel count. It takes at least 2 pixels to resolve something. To double the potential resolving power (making the assumption that nothing else has any effect) requires a quadrupling of pixels i.e. from 50Mp to 200Mp.

I believe modern lenses in particular will have been designed to take advantage of whatever MP count the manufacturer releases. 50 Mp is not really an issue. Even old lenses will have a sweet spot that takes advantage of the higher pixel count.

Also worth considering is viewing on a 'standard' monitor at 1:1 is like taking a 3x Loupe to a print. Viewing images at effectively 300% magnification is really only useful for editing/correcting image issues.

The real question is what do you need for the work you do now and aspire to do in the future. If you have no intention of making large prints regularly and more often than not only see your images on a standard low resolution monitor then do you really need more than 12-24 Mp?

...You need 50 mp for a 24"x36" (50cm x 90cm) print? Even for larger? I'm unconvinced....

I agree you do not necessarily need more MP for a print of any particular size IF you know that it will not be viewed at closer than viewing distance based on 1.5-2x the print diagonal.
The ideal native image size for your 24"x36" print from an Epson printer is 8640 x 12960 PPI = 112MP!!. Of course most of us do not posses such a camera but can still make great looking prints at this size from much lesser pixel count captures once we realise we need to view at an appropriate distance. All other things equal a theoretical 112MP camera would allow detail to be much better resolved and clear to a viewer standing closer than 'normal' to inspect the print.

Most people can't tell a difference in prints past 200dpi on most print substrates

Inkjet printers do not print as low as 200 dpi (DPI measure of dots/droplets of ink) so I assume you are referring to PPI, the measure of image resolution. While it may be true that most cannot tell the difference that would hold true particularly for those that are shown the images seperately. A side by side comparison of images may reveal subtle differences dependent on image content.

I am not aware of any inkjet printers that use 200PPI as a native resolution, it will most likely be 300 or 360PPI. In any case native image sent to the printer at 55, 100, 200 PPI or any other PPI will be upsampled by the print pipeline to its required resolution

[Lowell Goudge]

What's a practical limit?

I can talk about the USA, but in Europe the standard photo poster size is 50x75cm (price is 7.99Euros in Fuji Crystal Archive DPII paper), which at 300pi requires 52.2Mpixels.

lets back up a little. in the film world, and still to this day, when people talk of depth of field, the circle of confusion, etc.. or blurred hand held shots and the impact on a point source of light all the discussion revolves around the use of 1/100 of an inch as the practical limit. with this definition, you only really need 100 dpi or a little more, not 300, therefore the example you give, from a practical limit is 9x the pixels you need. so a good answer is really in the order of about 15 MP. we reached that with every pentax camera (give or take) since about 2007, and the K20. conversely if you consider printing 3X bigger in each dimension, you actually cannot fit your 52MP within a persons field of view while at the same time discerning detail finer than 1/100 as other than a point, so the DPI falls off as the print size gets larger simply because you cant see it all.

so you need to consider that yes there is a practical limit, and you need to differentiate between what is possible, against what is necessary.

i have always maintained that we need better ISO before significantly more pixels.

---------- Post added 03-04-21 at 08:00 AM ----------

snip.....

[/COLOR]
I consider cropping is for bad photographers. A photographer who does his work before taking pictures don't have to crop that much. Not quite cost effective to use a high mega pixels camera and crop most of it.[COLOR=Silver]

.

you have obviously never shot wild life.

you need to remember that image size = subject size x focal length / distance


take a photo of a humming bird with any pentax DSLR and lens ever offered including the 2000mm reflex and do the math. the 2000mm reflex has a minimum focus distance of 20 meters, so a 100mm humming bird will be, on your sensor 10mm high.

while you might get shorter lenses with a better magnification, due to the play of focal length over MFD, you will almost always be cropping even on a crop sensor body.

[ncallender]

Yes, wildlife, nature, travel, events.
There are complete genres where you pretty much have to crop. This is where more mp is good. For example, product photography for ad agencies is one where I would choose a much higher mp sensor. You can control the light and shoot at base iso, and it is one field where by definition you HAVE to leave room for significant cropping. Because the packaging designers will need it. But I still think for most stuff, at least for me personally 24mp is great and 36mp on a full frame sensor is the perfect balance.

[normhead]

We've been through this so many times.
As for the crop argument.

Smaller sensors for the most part have smaller pixel sites producing more definition in the area of the crop.
It's debatable which is more effective, tight framing on a crop sensor or much looser framing on an FF sensor. You might get better composition on an FF camera. Most likely not but it could happen. You will definitely get more subject resolution on a crop camera, unless shooting FF over 54 MP, or 645 over 108 MP.

So if your at all into resolution, you either choose 54 MP or thousands of dollars more to get the same crop image as you would using APS-c.

Consider the APS-c image pre-cropped.

People are claiming you get better crops, but they aren' showing examples. maybe stop as yin it until you actually try it and prove it. The laws of physics say you're probably wrong 99% of the time. Talking about ice in a life time shots is pointless. You will have once in a life time shots once in a life time. You can uy a camera for your once in a lifetime shot... but so far I've probably owned 15 cameras in my life time. The odds that you'll get a once in a life time shot with s camera you buy next are 15-1 against you having that camera when your shot comes around, if you haven't already missed it. And the odds of the camera you have with you being the right camera for the jog depends on many other things besides megapixels. You could miss your once in a life time because you have too large a sensor and don't have the DoF you need at ƒ2.8. You could miss your once in life time because you left the right lens home and could have had the shot with something like the much despised 18-300. You could. miss it because the small pixels on your megapixel camera don't have enough dynamic range. Or you could miss it because your heavy gear is slow and cumbersome or too difficult to use ergonomically.

More megapixels provide few if any advantages after 16 MP. And much of the time, you can't tell the difference between 36 MP and 24 MP unless someone tells you which is which. Then you can go all nitpicky.

Pentax is going in the right direction with the evolution of the accelerator chip.

IN camera crop, in camera noise reduction. Look at the DXOMARK rankings. The Pentax 645z is still ranks #2 and the K-1 #11

The 51 MP Canon produces such inferior images, it's so far down the list you have to scroll down 3 pages just to find it. These are images that have been normalized for noise and resolution. SO bottom line, at present, more MP means inferior images.

Someone let me know when it changes.
Until then, keep hoping or the tech advances that will mean a 51 MP comer out performs a 36 MP camera. Someday in the future there may Coe a time when more MP is actually better. No-one has been more critical of what DxO means over the years. But, it's true standardized testing. It means what it means. If you do the same thing every time for every camera, the ranking will have some meaning.

I'll take that over the google eyed cawing of new camera purchasers any day. If you want the camera in your hands for the perfect opportunity, once in life time shot, you want one of these 14.

Hasselblad X1D-50c
102
Pentax 645Z
101
Panasonic Lumix DC-S1R
100
Nikon D850
100
Sony A7R III
100
Sony A7R IV
99
Nikon Z7
99
Sony A7R II
98
Nikon D810
97
Sony Cyber-shot DSC-RX1R II
97
Leica Q2
96
Pentax K-1
96
Nikon D800E
96
Sony A7 III
96

And those images will all be pretty much indistinguishable. It takes a diference of 5 DxO points to be distinguishable pixel peeping, At reduced size for normal viewing, there will be no discernible difference. People who buy expensive gear with high MP would have done better with one of the 14 cameras listed, and possibly saved a pile of money.

If I had a million dollars I'd put it up as a reward for whoever could prove me wrong. Pictures with artifacts, noise, heavy noise, cross talk at pixel level and low DR etc. are not what you want for your once in a life time image. Some have this all wrong. The camera most appropriate for the job is what you want. Assuming that will always be the highest res camera possible is a mistake. You're probably more likely to miss achieving your once in a life time shot because you have a hi res camera as any other reason. The Canon 5Ds will be noticeably worse than any of the 14 cameras I listed, by a considerable margin, almost certainly visible at normal viewing. So actually in practical terms the opposite of what has been stated is true. For your once in a lifetime shot, you don't want a hi res camera. Maybe at some point in the future, but not now, and possibly not ever.

If you consider the DxO figures accurate for one testing procedure and set of applications, the 5Ds isn't really any better than a K-5ii. People need to contemplate that for a while. That isn't based on any MTF numbers but those are based on real world image quality, ie the real world. No hype involved.

So bottom line....the highest MP cameras do not produce the best images.... unless you're buying a hassy, and the Hassy is only marginally better than a K-1, for what 50x the price? The costs of larger files, slower post processing etc. makes really high MP a really expensive mistake as far as I can tell, taking advantage of the gullible.

[texandrews]

I consider cropping is for bad photographers. A photographer who does his work before taking pictures don't have to crop that much. Not quite cost effective to use a high mega pixels camera and crop most of it.

Here I believe you are incorrect and I must disagree. Cropping is essential for professional photography. In my line of work, it is not possible to make images of all artworks in 3:2, 4:3, 16:9, & etc formats. Editors of catalogs and books pretty much always crop the images to suit the graphic design parameters of the final product. And I often crop out extraneous blank wall, ceiling, or floor, plus other distracting things from my finished shots. Only in exhibition views do I carefully compose in camera, as they are a record of the overall environment.

My work is niche, but I'm certain other pros in other niches have similar tales.

From an art standpoint, format is always a choice no matter what the medium. The final object dictates what that choice needs to be, not a capture device. It's fine to compose in camera, but it's not a dictum.