Bill Pierce
Well-known
In the past years, the most obvious advance in digital cameras has been the increase in megapixels for any given sensor size. We’ve had some discussions about this in the past, but not talked about how to calculate how many pixels you might need to deliver a really sharp print. Say you were Edward Weston and you wanted to make very sharp and detailed 8x10 prints. You would only need 7.2 megapixels for a 300 ppi print, not too demanding.
One of the easiest ways to determine how “big” you can make a “sharp” print from a raw file is to put the file in a processing program that allows you to make prints, such as Lightroom, and in the print module check out the image size with the ppi at 300. I currently don’t have any cameras with less than 24 megapixels, but I dug out some old files from cameras with 16 megapixels. At 300 ppi I could make prints 16.321 x 10.881 inches, not bad. Next up from that is 24 megapixel cameras and their 300 ppi size came up as 20.792 x 13.862.
Of course, that’s a silly oversimplification. Any number of pictures are going to look good at lower ppi, and there are a huge number of other factors that effect the impression of sharpness in an image. Among the most obvious, lens quality, focusing accuracy, depth-of-field, subject motion, camera motion, contrast and tonal seperation. And with pixel count as high as it is in many of today’s cameras, these are really the important factors - not the pixel count. That’s something - 3 paragraphs about pixel count just to tell you pixel count isn’t so important. But, I suppose saying so is important because so much of what is seen in not only advertising, but editorial content, promotes pixel count. Yes, it can be important to some photographers. I have a friend, Eric Meola, who routinely exhibits 40 x60 landscape prints in galleries and whose work is printed in books with relatively large page sizes. For Eric and others like him, increase in pixel count and sensor size can be a real advantage. But, for the rest of us, it just lets us crop the image if we’ve taken the other necessary steps to preserve image quality (and crop from the center where lenses usually deliver their best performance). I wonder where we go to get the best image quality. I’ve actually started using a tripod and setting lenses at their optimum aperture - not all of the time, but some of the time. And, technically, the pictures are often better than those taken with my normal, somewhat carefree, shooting technique. Sometimes they're rather impressive. I wonder what you’re doing.
One of the easiest ways to determine how “big” you can make a “sharp” print from a raw file is to put the file in a processing program that allows you to make prints, such as Lightroom, and in the print module check out the image size with the ppi at 300. I currently don’t have any cameras with less than 24 megapixels, but I dug out some old files from cameras with 16 megapixels. At 300 ppi I could make prints 16.321 x 10.881 inches, not bad. Next up from that is 24 megapixel cameras and their 300 ppi size came up as 20.792 x 13.862.
Of course, that’s a silly oversimplification. Any number of pictures are going to look good at lower ppi, and there are a huge number of other factors that effect the impression of sharpness in an image. Among the most obvious, lens quality, focusing accuracy, depth-of-field, subject motion, camera motion, contrast and tonal seperation. And with pixel count as high as it is in many of today’s cameras, these are really the important factors - not the pixel count. That’s something - 3 paragraphs about pixel count just to tell you pixel count isn’t so important. But, I suppose saying so is important because so much of what is seen in not only advertising, but editorial content, promotes pixel count. Yes, it can be important to some photographers. I have a friend, Eric Meola, who routinely exhibits 40 x60 landscape prints in galleries and whose work is printed in books with relatively large page sizes. For Eric and others like him, increase in pixel count and sensor size can be a real advantage. But, for the rest of us, it just lets us crop the image if we’ve taken the other necessary steps to preserve image quality (and crop from the center where lenses usually deliver their best performance). I wonder where we go to get the best image quality. I’ve actually started using a tripod and setting lenses at their optimum aperture - not all of the time, but some of the time. And, technically, the pictures are often better than those taken with my normal, somewhat carefree, shooting technique. Sometimes they're rather impressive. I wonder what you’re doing.
Godfrey
somewhat colored
Yes, pixel count is useful up to a point, but when the average quality camera today has far more than enough pixel resolution to permit prints much larger than most people ever need or make, other factors in photographic craft take precedence. How many of us always make the very largest print possible? Very few, I imagine.
What do I do? Well, if I have any default habits, I tend to set the lens opening to between one and two stops down from wide open. And I keep the ISO setting as low as possible to maintain about 1/(2 x focal length) shutter speed ... or I get out my tripod. I spend time focusing carefully, and if I'm unsure on nailing just exactly the right exposure (and the opportunity is there) I'll make a second exposure with +0.3 to +0.7 EV more exposure.
In the end, however, content wins over technical quality most of the time and if you make two slightly exposures of the same subject, and the subject includes people or other things that move, picking the one that's slightly worse technically but has better expressions and gestures will always make the better photograph and the superior print. In My Opinion, of course.
What image processing applications you like is irrelevant ... they're all pretty darn good, when it comes right down to it. Taking the time to think and look at what you're rendering, and learning whatever application it is thoroughly, and spending the time to build up your image processing skills so you can get what you want out of your photos consistently, etc etc ... that's the part that you just have to sit down and do to get your best results. A poor rendering of a great photograph is no better than a great rendering of a so-so photograph.
G
What do I do? Well, if I have any default habits, I tend to set the lens opening to between one and two stops down from wide open. And I keep the ISO setting as low as possible to maintain about 1/(2 x focal length) shutter speed ... or I get out my tripod. I spend time focusing carefully, and if I'm unsure on nailing just exactly the right exposure (and the opportunity is there) I'll make a second exposure with +0.3 to +0.7 EV more exposure.
Note: The justification for this last item is that most in-camera metering systems tend to underexpose by a half stop to a stop to protect against saturation in the highlights with contrasty lighting. Once you're aware of what your specific camera tends to do, and when you're faced with average scenes where highlight burnout is unlikely, a little more exposure can help with reducing noise and improving tonal gradation. This is particularly true when you've set the camera to capture raw data, since the histogram display in the camera is typically reading the generated JPEG thumbnail on the fly.
In the end, however, content wins over technical quality most of the time and if you make two slightly exposures of the same subject, and the subject includes people or other things that move, picking the one that's slightly worse technically but has better expressions and gestures will always make the better photograph and the superior print. In My Opinion, of course.
What image processing applications you like is irrelevant ... they're all pretty darn good, when it comes right down to it. Taking the time to think and look at what you're rendering, and learning whatever application it is thoroughly, and spending the time to build up your image processing skills so you can get what you want out of your photos consistently, etc etc ... that's the part that you just have to sit down and do to get your best results. A poor rendering of a great photograph is no better than a great rendering of a so-so photograph.
G
For me, it has been high ISO... that was the thing in digital that really made digital special.
Like you, my current cameras are 24mp or more. I've used, for the digital work that I still like of mine, anything from 10mp to 42mp. I have no issue with more MPs up to a point. I recently exhibited a few prints and the gallery wanted them kind of large... 20x30"... but some of the photos they liked were 10mp. They ended up passing on those and used some 16 and 24mp photos instead. If it was my decision, I would have just printed them smaller because I see no harm in doing so. People talk of proper viewing distance, but a lot of people get up close to look at details these days. I guess pixel peeping has infiltrated the museums and galleries too.
I also sold a few prints recently and again the customer wanted large prints. I got lucky because he chose 42mp files. It worked out really well because they were over 300dpi and they were impressive in the hand and to up close inspection.
I'm going to buy a Fujifilm GFX-50R next... and that will allow for a decent quality 30x40" print IMO. If that isn't enough for someone, I give up. I'll still most likely use lower MP cameras too.
Bill Pierce
Well-known
"For me, it has been high ISO... that was the thing in digital that really made digital special."
I think you are absolutely on target. The quality improvement was slow, but constant; so, I think some people (me) didn't pay attention. But it has certainly changed how many photographers work.
I think you are absolutely on target. The quality improvement was slow, but constant; so, I think some people (me) didn't pay attention. But it has certainly changed how many photographers work.
willie_901
Mentor
MTF50 And Perceived Sharpness
MTF50 And Perceived Sharpness
Being able to "determine how 'big' you can make a 'sharp' print" is complicated and depends on many variables.
What I do is make a guess at the typical viewing distance. Next I use the the pixel pitch values for my cameras' sensor diagonals (X100T pixel pitch = 4.76 µm, X-Pro 2 pixel pitch = 3.93 µm) to estimate the maximum print size.
For a 0.2 m ( (~8") viewing distance the minimum pixel pitch is 1.34 µm for a 16 X 24 " print. But at a 2 m (6.5 ft) viewing distance the minimum pixel distance becomes 13.44 µm.[1] The minimum optimum viewing distance for a 16 X 24 print from a X-Pro 2 image is 0.6 M ( ~2 ft).
The details are discussed below.
I also qualitatively evaluate the inherent image detail of the subject. The subject detail for a city scape requires much higher perceived MTF 50 than a foggy sunrise in a seascape.
I often print Tri-X scans on cotton, fine-art, photo rag paper to emulate a wet-chemistry analog print. The negative, and scan perceived MTFs are lower than the printer and high gloss paper perceived MTFs. Now the minimum viewing distance depends on the negative quality.
Details
The following assumes lens, camera/subject motion and subject detail (contrast) are not limiting factors.
This blog post, MTF50 And Perceived Sharpness, discusses some of the quantitative considerations.
The MTF of the human retina at 20/20 vision is one relevant parameter.
The linked blog post describes a conversion for the distance covered by one degree on the retina to the equivalent distance on a print or monitor.
For a 24 X 36 mm sensor, images printed "as 5x7s from 218mm away, 8x12s from 366mm away, fit to your 24″ monitor from 611mm away or 10x15m billboards from 18m away: they are all standard distances equal to the diagonal of the displayed image, hence project the same cycles per degree on the retina."
When the viewing distance is held constant, it turns out the ratio of the number of pixels on the sensor diagonal to the number of pixels on the monitor diagonal are important parameters for that limit perceived sharpness.
If the printer's diagonal DPI is insufficient, the perceived image sharpness (retna MTF) on a monitor will not necessarily translate to same perceived image sharpness on a print
The print media is is another factor. Epson's Legacy Etching’s textured matte finish will have a lower perceived MTF compared to their high-gloss Metallic Photo Paper.
1. For a 24X36 mm sensor the the corresponding minimal pixel pitches would be 2.06 µm and 20.61 µm respectively.
MTF50 And Perceived Sharpness
Being able to "determine how 'big' you can make a 'sharp' print" is complicated and depends on many variables.
What I do is make a guess at the typical viewing distance. Next I use the the pixel pitch values for my cameras' sensor diagonals (X100T pixel pitch = 4.76 µm, X-Pro 2 pixel pitch = 3.93 µm) to estimate the maximum print size.
For a 0.2 m ( (~8") viewing distance the minimum pixel pitch is 1.34 µm for a 16 X 24 " print. But at a 2 m (6.5 ft) viewing distance the minimum pixel distance becomes 13.44 µm.[1] The minimum optimum viewing distance for a 16 X 24 print from a X-Pro 2 image is 0.6 M ( ~2 ft).
The details are discussed below.
I also qualitatively evaluate the inherent image detail of the subject. The subject detail for a city scape requires much higher perceived MTF 50 than a foggy sunrise in a seascape.
I often print Tri-X scans on cotton, fine-art, photo rag paper to emulate a wet-chemistry analog print. The negative, and scan perceived MTFs are lower than the printer and high gloss paper perceived MTFs. Now the minimum viewing distance depends on the negative quality.
Details
The following assumes lens, camera/subject motion and subject detail (contrast) are not limiting factors.
This blog post, MTF50 And Perceived Sharpness, discusses some of the quantitative considerations.
The MTF of the human retina at 20/20 vision is one relevant parameter.
The linked blog post describes a conversion for the distance covered by one degree on the retina to the equivalent distance on a print or monitor.
For a 24 X 36 mm sensor, images printed "as 5x7s from 218mm away, 8x12s from 366mm away, fit to your 24″ monitor from 611mm away or 10x15m billboards from 18m away: they are all standard distances equal to the diagonal of the displayed image, hence project the same cycles per degree on the retina."
When the viewing distance is held constant, it turns out the ratio of the number of pixels on the sensor diagonal to the number of pixels on the monitor diagonal are important parameters for that limit perceived sharpness.
If the printer's diagonal DPI is insufficient, the perceived image sharpness (retna MTF) on a monitor will not necessarily translate to same perceived image sharpness on a print
The print media is is another factor. Epson's Legacy Etching’s textured matte finish will have a lower perceived MTF compared to their high-gloss Metallic Photo Paper.
1. For a 24X36 mm sensor the the corresponding minimal pixel pitches would be 2.06 µm and 20.61 µm respectively.
Beemermark
Mentor
An excellent 35mm camera and lens can make a fine photograph up to about 11x14, say a two page spread in Life magazine. However, and this is a big however, the same picture printed 16x20 and hung on a wall that makes you stand 2 or 3 feet away will look just as sharp. Viewing distance is everything. Some professional photographers on this forum have informed me that to sell pictures you need a gazillion megapixels since buyers well get their nose right up to the picture in the gallery to examine the sharpness. I think if someone needs to examine the photograph that closely the photograph isn't really making an impact.
Andrea Taurisano
il cimento
I couldn't agree more. I shoot mostly wide open, often with a ND filter on since Leicas don't allow fast enough exposures, and I value graininess, vignettes and everything that makes my images perfectly imperfect (or plain bad). My approach is more like you should be able to look at my pictures with narrowed eyes, and not miss much of the content.
But of course I am aware that photography is immensely more than my crappy images. I myself admire many of the photos I see, also here on RFF, which have a totally different approach than mine.
Sticking to the original question, though: I had quite a few A3-size prints made from both Leica M9 (18 Mp) and M8 (10 Mp) files, and no viewers or buyers ever complained about lack of sharpness.
I don’t think that’s a fair statement since I see people in museums and galleries getting up close all the time. That is unless you think everything in museums and galleries isn’t making an impact. Some photos have a simple, clear and big subject that takes up the whole frame, some images are more complex. I know that when I get a chance to look at the prints of my favorite photographers, I certainly am going to look at them from several viewing distances including very close. I’m also a fan of photo books which we also look at up close and personal...
bullterrier
Established
not much to say that hasn't been said but I can contribute too the thread with an image made with a 3 MEGA pixel camera
Sigma SD9+70mmf2.8
Sigma SD9+70mmf2.8
Andrea Taurisano
il cimento
I surely agree that a photo often needs more than a quick look to fully reveal itself and trigger an emotional response. But I am not sure I agree that what it takes, in addition to the first look, is getting so close that pixels and resolution become the limit.
To reveal more complex photos, I think spending more time with them helps much more than sticking my nose into them. Therefore I agree that photo books are wonderful: They give me all the time I want together with the photos, not because I can get into them. To illustrate what I mean, see the photo below, that I took from a train in Siberia. At a quick look, it's just a landscape. Give it a few more seconds and you'll see something very tiny in the middle. For me, that detail adds a lot to the image, even though the resolution of the image (and the weather) is too poor to see any detail about that human figure.
bullterrier
Established
nice image Andrea
Out to Lunch
Menteur
Instead of buying a tripod, I bought an IBIS camera. Cheers, OtL
I’m not sure we are having opposing viewpoints here. I agree time is a wonderful revealer of photos too. However, a photo book is sticking your nose into the photos... since you don’t hold them out at arms length. I know it’s fashionable to say MPs don’t matter. That’s ok... sometimes they just don’t. However, sometimes they do. Nobody every complains that someone used an 8x10” view camera to get high resolution. Why? Because that is perceived as cool. Digital isn’t cool. Sometimes you need fine detail and sometimes you do not. Your image didn’t need the fine detail. That doesn’t mean others do not. One photo does not make what I’m saying invalid. It completely depends on the image, your purpose, the content, and the method of distribution. The thing with large MPs is that you can still make small prints and stand back away from them. Also, let’s remember that enlarged film grain looks a lot better than enlarged digital artifacts. A LOT better.
20 years ago I was invited into my Boss's office to see the prints from a 100MPixel monochrome camera, he told me it was as good as a 4x5 camera.
I don't own a camera with more than 18MPixels. Stopped at the M9, M Monochrom, and Nikon Df. Nor do I make a living selling prints. I used to make a living writing firmware for digital sensors. I have an open door if I ever want to do that again.
I would like to see improvements made in saturation count for digital. Smaller pixels of the higher pixel count cameras capture more photons before saturating. The M8 saturates at ~60Ke-, single read-out. A 52MPixel full-frame sensor based on CMOS and double-sampled should be able to do 30Ke- if the material collected the same amount of light. The 12MPixel CMOSIS sensor used in the Pixii has a saturation count of 13500e-. The 47.5MPixel full-frame CMOSIS sensor has 14500 e-.
From many discussions on camera forums, I think high pixel count has done more for finding fault with lenses than much else.
I don't own a camera with more than 18MPixels. Stopped at the M9, M Monochrom, and Nikon Df. Nor do I make a living selling prints. I used to make a living writing firmware for digital sensors. I have an open door if I ever want to do that again.
I would like to see improvements made in saturation count for digital. Smaller pixels of the higher pixel count cameras capture more photons before saturating. The M8 saturates at ~60Ke-, single read-out. A 52MPixel full-frame sensor based on CMOS and double-sampled should be able to do 30Ke- if the material collected the same amount of light. The 12MPixel CMOSIS sensor used in the Pixii has a saturation count of 13500e-. The 47.5MPixel full-frame CMOSIS sensor has 14500 e-.
From many discussions on camera forums, I think high pixel count has done more for finding fault with lenses than much else.
Hilmersen
Established
I had to read this: https://www.perkinelmer.com/CMSReso...CH_Determine Saturation (LI40 -BTv4)-8428.pdf in order to understand the previous post.
Dogman
Mentor
My prints are rather small--12x18 at the most, 6x9 normally. I have cameras up to 36mp but that's overkill for me. My 12mp Nikon D700 gives me excellent prints in these sizes, even with some pretty serious cropping. Megapixel race over for me. I won.
Like John above, the advances in ISO performance is what I really appreciate. It's nice to just set ISO to "A", set the parameters and let the camera do its thing.
Like John above, the advances in ISO performance is what I really appreciate. It's nice to just set ISO to "A", set the parameters and let the camera do its thing.
willie_901
Mentor
As long as we use photodiodes in CMOS chips, increasing "Saturation count" reduces sensitivity and vice-versa.[1]
Increasing the photodiode floating, diffusion-node capacitance increases the electrical charge capacity. However, higher diffusion-node capacitance also reduces voltage swing (conversion gain). Less conversion gain means lower sensitivity and increased read noise.
Product managers choose a conversion gain that is a compromise between high saturation count and high sensitivity. For better or worse, the market is driven by low-light image quality and the compromise typically favors sensitivity.
This white paper discusses the issue and a describes a solution - dual conversion-gain sensors. More technical detail is available here.
Cameras with dual, conversion-gain sensors became available about four years ago. At low camera ISO settings a low conversion gain is used to maximize full-well capacity. At a higher camera ISO settings (and above) the conversion gain is automatically increased to maximize sensitivity. Cameras with dual, conversion-gain sensors switch conversion-gain levels at different ISO settings - typically an ISO setting between 300 to 800.
1. CCD sensors use an array of coupled-gates which can store more charge. The last coupled gate's floating diffusion node does the charge-to-voltage conversion.
Jamie Pillers
Skeptic
As usual, I got to this thread late. So excuse me if this has already been said. I think the 300dpi 'rule' is probably not a great place to start when talking about how good a print will look. A better starting place for thinking about how good one's print will look to, say, a gallery observer, has probably more to do with viewing distance and desired sharpness. Personally, I don't expect to see 300dpi detail when I'm looking at a 30in. x 40in. print in a gallery. I don't want to view a print that size so close that I can detect 300dpi detail. I want to enjoy it a few feet away, or more. So the dpi number changes a lot.
The other issue is sharpness. Many, MANY photographs by famous photographers shown in places like SF MOMA are often not sharp. (I think most of these have been prints from film and thus tend to be smaller print sizes.) Those photographs astound one's senses with subject matter usually. Dpi is never something I even think about when I look at those photographs.
The other issue is sharpness. Many, MANY photographs by famous photographers shown in places like SF MOMA are often not sharp. (I think most of these have been prints from film and thus tend to be smaller print sizes.) Those photographs astound one's senses with subject matter usually. Dpi is never something I even think about when I look at those photographs.
Andrea Taurisano
il cimento
It really depends on (among other things) what the purpose of the print is. Below is my kitchen. The print behind the cooking plates is about 180 x 60 cm (something like 6 x 2 feet?). It was made from an iPhone panorama shot counting about 8850 x 3800 pixels. That gives a resolution of around 145 ppi, way too low for anything decent, in theory. But in practice, it just looks more than fair enough. No, I can't count larch needles, but that's not the point. But I would never exhibit or sell this print either.
PS: sorry if I pollute this nice forum with image samples of rather low "artistic value", but I think discussions like this are best held with examples at hand.
PS2: Yes, Lasagne were very good yesterday
PS: sorry if I pollute this nice forum with image samples of rather low "artistic value", but I think discussions like this are best held with examples at hand.
PS2: Yes, Lasagne were very good yesterday
Probably because DPI doesn’t come into play when we talk about analog photography printing? The reason I mention it is because it’s an easy starting point to reference in digital printing. It’s your best quality print at a given MP... and quality goes down from there. Of course you don’t need 300dpi. We all know this. I will state once again that I always look at prints in a gallery up close and while standing back... whether it’s film or digital...and there’s nothing wrong with this. It depends on what’s within the frame if this is worth doing.