It might seem like there is a pretty straightforward correlation between sensor size, resolution, and maximum reasonable print size, but the relationship is a bit more complicated than just those specific variables. This great video examines the sort of print sizes you can reasonably create using photos from an iPhone, an APS-C camera, and a medium format body.
Coming to you from First Man Photography, this interesting video discusses the maximum size prints you can create from different sources ranging from an iPhone to a medium format cameras, along with an APS-C sensor. While the source of the images certainly has an impact, it is more than just this. Our own Lee Morris investigated this a few years ago when he shot an image for a billboard and he only needed two megapixels to do so. The issue mostly boils down to viewing distance. If you are viewing a photo up close, you will need fairly high resolution and detail, as your eye will be able to discern a lot more detail. On the other hand, for something seen from much farther away (such as a billboard), less detail is needed. Check out the video above for the full rundown.
Are the points that you’re making valid for home viewing? I understand viewing distance as a property in a museum, or on a billboard, and mostly for prints over 44”. For a print is an object. An object is meant to be examined up close. If the object exhibits flaws in its surface, like pixelation or mushiness as you state, then it’s an imperfect object, and it’s viewing distance is irrelevant. In the earlier days of digital printing, when I still avoided digital sensors and used film, around 2002, there were a lot of people displaying images in art galleries, that were pixelated. Pixelation must be one of the most awful aesthetics ever known to the human eye. My last point is that in the film world, medium format film, or 2 1/4 inch, could be printed to 40 inches. Perfectly. Why would that standard be any different today? One should settle for nothing less.
"For a print is an object. An object is meant to be examined up close."
Not always true. In the case of a photos, not all subjects require examining them closely. Most of the time the photo is meant to be enjoyed in totality. Details are not always interesting by themselves, are important in relation to all the other parts of the photo.
One example that might hold true for what you stated is a photo of a tree. The bark or leaves might be examined closely for their visual texture. But then, many people don't always appreciate that kind of thing and their attention is on the entire tree.
When details are less important than the whole, I'd consider the distance at which the photo is meant to be viewed and the size of the space in which it hangs.
So a cell phone image gives you a small print with mushy details at best viewed from a distance whereas a medium format camera gives you nice detailed large print which looks great at any distance.
You could have narrowed down your 16:48 minute video to one sentence !
I think that the issue can be broken into two distinct categories, 1) museum quality, and 2) hanging on your own wall at home. For those that are printing at museum quality, they are, I'm sure, well aware of the requirements on detail and resolution. So, to me, articles and videos like this one are trying to provide some guidance for people who want prints for home or office decoration, and the information on how much resolution is really needed can be quite contradictory and confusing. The video provides some nice qualitative information, but it doesn't quite go far enough to provide a rule of thumb for the casual or enthusiast photographer.
So, let me suggest a simple quantitative answer to that question.
At normal indoor lighting levels, the diffraction limit of the human eye is approximately 0.000122 radians (0.007 degrees) on average. This assumes a pupil diameter of 5.5 mm (typical for midrange indoor lighting) at the middle of the visible light spectrum, 550nm (green-yellow). Remember that the diffraction limit is the absolute limit that the human eye can perceive, if perfectly corrected, and it is determined solely by physics. For most people, our visual acuity does not approach the diffraction limit.
Generally, color perception and visual acuity are very good in the macular region of human vision, which is approximately +/- 9 degrees total field of view. By +/- 20 degrees total field angle, both visual acuity and color perception are beginning to drop off. Therefore, we tend to view photos at a distance for which the whole photo is between these two limits. Perceptual experiments have determined that beyond about +/- 15 degrees (i.e. being closer to the print) our eyes tend to dart around the picture and we don't necessarily perceive the whole. (note that this type of experimental result is highly subjective). This leads to the generally stated guideline that a comfortable viewing distance is greater than two times the longer of the two dimensions of a photo, TV screen, etc. This equates to 28 degrees total field of view.
Dividing the total field of view by the diffraction limit of the human eye gives us a maximum resolution on the longer dimension of 4000. So, we can say that 4000 pixels along the longer dimension will produce a print with detail to the limit of human diffraction, when viewed at a comfortable viewing distance. Assuming a 3:2 aspect ratio for most cameras, this tells us that 10.7 MP should produce an acceptably sharp print, at any size, when viewed from a comfortable viewing distance. Obviously, if someone views a photo from a foot away, its quantization will always be visible. However, this would be like watching your 60 inch television from two feet away, which no one really does.
As a point of reference, 4K ultraHD displays are 3840 pixels on the long dimension.
So, I would suggest for general printing, as long as a photo has 10 MP resolution, it can be printed at any size and still look good. I have prints that are 24x36 inches made with my old Canon 40D (10MP sensor) that look very good. Up close, one can certainly see some softness and pixelation, but no one casually views photos like that.
Higher resolution cameras give one much more flexibility in cropping to be able to maintain 10 MP.
So my rule of thumb is that as long as one retains 10 MP or more in a photo, print size is not limited for casual viewing. And in reality, I have photos that I have cropped to only 3 or 4 MP and printed at 16x20 that look just fine. Maybe a little soft, but not enough to bother me.
I know that there is a whole lot of subjectivity in this, but I humbly present it as a general rule of thumb for anyone who is looking for a starting point to figure out how big they can print.
From your experience am I to gather that an uncropped iPhone 12 (12mp coincidencly), should print at 16x20 without loss of quality ? Quality, of course being subjective.
It will. But the quality will not be the same as for a larger sized camera. The big difference between the iPhone and a larger camera is in the size of the optics. The iPhone optic has such a small diameter that its image quality will always be limited by the diffraction limit of the optics, regardless of the number of pixels in the sensor. Diffraction limit is inversely proportional to the diameter of the optics aperture, so a small optic will have a larger diffraction limit. Diffraction limit is the minimum angular separation that can be detected. So think about it as the minimum size object that can be resolved (hopefully, if you already knew about DL, that I am not insulting your intelligence). So blowing up an iPhone picture will not cause any loss of quality from the original image, but not that it will be as sharp as an image taken with a higher quality optic. The iPhone diffraction limit is about the same as the human eye, but in such a small optic, it is difficult to achieve optical performance that is equal to the diffraction limit. For a larger camera, the aperture will be 10x or more the iPhone aperture, so the diffraction limit will be an order of magnitude or more smaller than that of the human eye. So the optics can be designed more easily to get performance closer to the diffraction limit and far exceed the "resolution" of the human eye. This will result in a higher quality image.
Thanks for asking the question, because I should have addressed that in my post. I should have said that 10MP will allow a print to preserve the perceptive quality of the original image, but that it doesn't necessarily hold if the original image is "low" quality.
Well said! Finally it is all about the circle of confusion (or blur spot, if you like). Must be a 150 years we already know about it. Did he mention it in the video? I did not watch it.
Interesting and, on the surface, persuasive. Are these your own conclusions, or are they backed up by material published elsewhere?
They are my simple calculations based upon basic physics. It is easy to demonstrate the diffraction limit of the human eye by printing patterns (e.g. alternating vertical bars, etc) and moving further back until they are indistinguishable. One would definitely find that things get hazy before the spacing reaches the angular limit of diffraction. That is why it is hard to distinguish the "C" from the "O" and "H" from the "R" on the bottom of the standard eye chart. The information on the regions of human vision and declination of color perception and visual acuity are well-documented. I would guess that determination of the comfortable viewing distance is squishier (a technical term) because it is based on subjective measures of perception. One would suppose that there is a lot of variability amongst people. But changing the field of view for comfortable viewing distance within reason would not significantly alter the conclusion. Maybe 15 MP vs. 10 MP vs. 8 MP.
As I said, this is just my rule of thumb, certainly I wouldn't want anyone to take it as a hard and fast rule. But it seems a lot of people are under the impression that the larger the print, the more megapixels needed. But a 20 inch, 10 MP image viewed at a distance of 50 inches looks exactly the same as a 40 inch, 10 MP image viewed at a distance of 100 inches. All of the angular separations scale. I would guess that some of the confusion comes from the recommendations that when printing, images should be greater than 300 dpi (for example). Seems these recommendations were based upon images of the size in books and magazines.
I've never seen this type of analysis published anywhere, but I have never really looked for it.
If you have a blog, post this and I'll link to it from mine.
Obviously there's a relationship between print "resolution" and viewing distance. But "resolution" has many definitions.
That tired old "300 dpi" thing just never dies. It's an anachronism from print magazines of the past. On POD forums I see people endlessly confused and thinking they have to "upload at 300 DPI", or set their camera to it, or upsize their images in Photoshop.
I do know that I've sold quite a few big prints - 3, 4 even 5 feet across - from 24 mp images through a POD, and none were returned.
I don't have a blog, but thanks for the sentiment. You made me go look for published studies. As one would suspect, there has been much published on image quality based upon visual acuity. I found an excellent and comprehensive analysis at this website:
http://www.normankoren.com/Tutorials/MTF.html
It is very technical and not for the faint of heart! For those that like gory details, the math is nicely done. It contains a plot that summarizes things very well. It plots the contrast sensitivity of the human eye (i.e. the ability to distinguish objects) and is taken from :
J. L. Mannos, D. J. Sakrison, ``The Effects of a Visual Fidelity Criterion on the Encoding of Images'', IEEE Transactions on Information Theory, pp. 525-535, Vol. 20, No 4, (1974)
From their analysis, separations greater than 60 cycles per degree (.000291 radians/cycle) cannot be perceived by the human eye. (Note this is for 20/20 vision, which is not "perfect vision" defined by diffraction limit). A cycle is one pair of an alternating black and white stripe. In a digital camera, we would need two pixels per cycle, so the required resolution is 120 pixels/degree (0.000145 radians/pixel). My simple analysis gave me 0.000122 radians/pixel, or 143 pixels/degree.
"Up close, one can certainly see some softness and pixelation, but no one casually views photos like that."
Perhaps it would be best if you didn't speak for everyone.
"Maybe a little soft, but not enough to bother me."
That's better.
Sorry to offend you. Thanks for the impertinent grammar lesson. It adds to the heart of the discussion.
I just started the whole printing journey. And have discovered some of my personal boundaries.
Check out this 24mp image printed at 60". I kept the foam and my hand in for relative size and quality reference. After I printed this image (and another at 60") I immediately purchased the A7r4, for the 61mp. the 24mp image will print just fine at 36". However I don't want my print size to be limited by equipment, so I bulked up on MP's. Having done that - 4 out of my 5 lenses don't resolve up to 60mp. So I needed to purchase updated lenses that will resolve up to 60+mp
It's really been a pain in the ass.
Perhaps the most critical point was overlooked, namely critical focus. A 61 mp image hand held will likely produce a worse print than a 24 mp camera with similar settings on a tripod. Want sharp, large prints? Start with a decent body, choose the right subject/light, pair it with a sharp lens, use the right settings, and place the camera on a tripod.
Respectfully, you don't know what you're talking about.
(assuming you're referencing my comment)