Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: ErikKaffehr on December 16, 2011, 10:19:04 pm
-
Hi Marc,
Just a few comments:
1) Each pixel contains essentially a photo sensitive area and about 6 transistors. The size of the transistors depends on the design rules used. So if you decrease pixel size without using narrower design rules, the area sensor vs. electronics will shrink. The sensitive area also works as a capacitor, reducing the size of the capacitor will reduce the "full well capacity" of the sensor. The "full well capacity" is the determining factor for the high end of DR. Optimal pixel size varies with design rules.
This article explains it well: http://isl.stanford.edu/~abbas/group/papers_and_pub/pixelsize.pdf
2) I agree on the issue of avoiding the AA-filter, but with shrinking pixel size the thickness of the AA-filter will also shrink. I'd also say that the effects of AA-filtering are overstated as it is obvious that it is diffraction and not AA-filtering that limits lens sharpness. Once we stop down beyond optimum aperture resolution drops.
3) Binning in hardware makes the pixels larger. It is good for increasing DR, specially at high ISO, but doesn't counteract aliasing.
Best regards
Erik
Or how about a FF 120mp sensor that is binned to 30mp! no AA filter (the lens resolution becomes the filter) less artifacts etc
Marc
-
Hi Erik
I was trying to read up on this last week and my understanding is:
1. Ideally a pixel should be as small as possible. The smaller it is, the better it can resolve.
2. However, the bigger the pixel is, the greater the dynamic range due to its light-collecting capacity.
Therefore, usually it's a trade-off. Am I right, or have I misunderstood completely?
-
Hi,
1) Yes, smaller pixels will resolve more details, but there are limitations. The main limitation is diffraction, once a lens a is diffraction limited we get diminished returns from shrinking pixel size.
2) The DR of the pixel is reduced with it's size, but we get more pixels and that compensates a lot. LuLa has an excellent article on the issue here: http://www.luminous-landscape.com/essays/dxomark_sensor_for_benchmarking_cameras.shtml
Combining pixels in software doesn't reduce readout noise the same way as binning them in hardware.
Here is a demonstration of diffraction: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=1 left column shows effects of diffraction.
Best regards
Erik
Hi Erik
I was trying to read up on this last week and my understanding is:
1. Ideally a pixel should be as small as possible. The smaller it is, the better it can resolve.
2. However, the bigger the pixel is, the greater the dynamic range due to its light-collecting capacity.
Therefore, usually it's a trade-off. Am I right, or have I misunderstood completely?
-
Erik
A few months ago Canon announced the development of a 120mpx APSc sensor But like our 10 mpx point and shoots the real resolution is diffraction limited and resolution limited to much larger than those little point and shoot pixels right? But like on the IQ's use hardware binning and let the diffraction limit and resolution limit of the lens be the AA filter. in other words if the lens cannot resolve a single photosite but can resolve 4 photosites don't we have a simple solution? sharper pixels, better DR, no AA filter, simpler de-mosaicing (each group of 4 pixels has an RGB value of its own).
Marc
PS I vote for 2.5 micron photosites binned in fours! And let the sensor area determine the gross resolution
-
Erik, thanks for those references.
-
Please let me see if I understand. Smaller pixels create more detail than bigger pixels. The smaller the pixel the sooner it will fall into diffraction obliterating the extra resolution. The smaller the aperture, the more diffraction at any pixel size. Ok these.
As I understand, the detail obtained from bigger pixels will never be more than the detail obtained from smaller pixels no matter how much diffraction they suffer. Right? Or.. it reverses? I don't think so.
Thanks in advance
Eduardo
Hi,
1) Yes, smaller pixels will resolve more details, but there are limitations. The main limitation is diffraction, once a lens a is diffraction limited we get diminished returns from shrinking pixel size.
2) The DR of the pixel is reduced with it's size, but we get more pixels and that compensates a lot. LuLa has an excellent article on the issue here: http://www.luminous-landscape.com/essays/dxomark_sensor_for_benchmarking_cameras.shtml
Combining pixels in software doesn't reduce readout noise the same way as binning them in hardware.
Here is a demonstration of diffraction: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=1 left column shows effects of diffraction.
Best regards
Erik
-
Thanks for the links, Erik. Appreciate it.
-
Hi,
Yes, your observations are correct.
With smaller pixels diffraction effects will be more obvious. You always seem to loose some but with smaller pixels you loose more when stopping down. The image below has been measured on a 6MP DSLR and on one with 10 MP. The 10 MP camera always gives more detail, but both start loosing out at the same time.
(http://echophoto.dnsalias.net/ekr/images/stories/difractionlimit.gif)
Some part of what lost with diffraction can be regained with correct sharpening, the enclosed image should illustrate this. That image is taken from this article:
http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures
Best regards
Erik
Please let me see if I understand. Smaller pixels create more detail than bigger pixels. The smaller the pixel the sooner it will fall into diffraction obliterating the extra resolution. The smaller the aperture, the more diffraction at any pixel size. Ok these.
As I understand, the detail obtained from bigger pixels will never be more than the detail obtained from smaller pixels no matter how much diffraction they suffer. Right? Or.. it reverses? I don't think so.
Thanks in advance
Eduardo
-
Hi,
Erik: It's good to assure this "assumed" truth.
Muy agradecido
Eduardo
Yes, your observations are correct.
Best regards
Erik
-
A dream sensor for me would be full-frame 40-48 pixel for high res still life / studio, binned to 4:1 10-12 megapixel for hand-held and high ISO.
Today it seems like small pixel sensors perform well up to ISO3200, but you still need large pixel sensors (for example D3s) to get decent quality in the ISO6400 - ISO25600 range. It does not seem to be possible to combine the best ultra high ISO performance with small pixel sizes yet.
Having to choose, I prefer high res over high ISO for my uses. However, the ultra-high range can be very useful for journalism/wildlife/sports/indoor events and is also useful for me at times, so if hardware binning could improve performance from ISO6400 and up that would yield a great all-around sensor. If I could get an ISO12800 with decent quality (similar to what D3s does today) and have that incorporated in the auto-ISO mode then I think I would very rarely feel limited by ISO range.
-
Hi Marc,
Just a few comments:
1) Each pixel contains essentially a photo sensitive area and about 6 transistors. The size of the transistors depends on the design rules used. So if you decrease pixel size without using narrower design rules, the area sensor vs. electronics will shrink. The sensitive area also works as a capacitor, reducing the size of the capacitor will reduce the "full well capacity" of the sensor. The "full well capacity" is the determining factor for the high end of DR. Optimal pixel size varies with design rules.
This article explains it well: http://isl.stanford.edu/~abbas/group/papers_and_pub/pixelsize.pdf
Eric,
The Stanford article you referenced is excellent. Their calculations were for a CMOS sensor using 0.35 micron technology, giving a fill factor of 30%. The optimum pixel size was 6.5 microns. If one were to use a smaller CMOS scaling one could improve the fill factor and the pixel size could be smaller. For a given fill factor, the use of microlenses can improve light sensitivity by focusing all the light of the pixel area on the active pixel site. However, the full well might not improve, since charge density would be the limiting factor.
CCDs have no electronics on the pixel, so their fill factor is larger.
With improved sensors with lower read noise, performance of smaller pixels at base ISO is often fully satisfactory.
Regards,
Bill
-
The amount of detail is not a simple function of pixel pitch. Which sensor would produce more detail, a 30MP 24x36mm sensor or a 30MP 48x72mm sensor? And when does detail become meaningless--if your viewer cannot perceive the detail or the output device unable to render the detail, what is the point of the detail?
-
Hi,
Both sensors will produce the same amount of detail, but the 48x72 mm sensor will have better microcontrast. The 48x72 mm sensor will also produce a lot of fake detail unless it has a very agressive OLP filter or you stop it down to f/22-f/32. (The Leica S2 seems to cease aliasing when stopped down past f/11, your hypothetical 48x72 sensor would have almost the double pixel size so it would need stopping down two stops more).
Moiré, which is a one form of aliasing, is a problem for many studio photographers with low res backs like 20MP MFDB, so the issue is not only hypothetical.
Regarding printer resolution your 30 MP sensor would resolve around 6700 pixels horisontally. New Epsons at finest setting seem to resolve 720 pixels/inch. So if you don't print wider than 9.3" then you will never exceed the capability of a modern printer. But my standard print size is 26 inch wide.
The best MFDB is probably the Phase One IQ180 and it has 5.2 micron pixels.
Best regards
Erik
The amount of detail is not a simple function of pixel pitch. Which sensor would produce more detail, a 30MP 24x36mm sensor or a 30MP 48x72mm sensor? And when does detail become meaningless--if your viewer cannot perceive the detail or the output device unable to render the detail, what is the point of the detail?
-
Just my 2cents. I don't think there is an "optimal" sensel size anyway. Sampling and DR are linked: it's very clear in astrophotography with very faint signals and markedly different focal lengths (real ones, not photographic equivalent). All other things being equal, there's always the unescapable trade-off between resolution and per pixel SNR (per area SNR is a different story). The range of sensel sizes we are in (say, 4-8 microns) is a good and polyvalent compromise. But yes, a camera with lots of 4 micron pixels with clean 2x2, 3x3 and 4x4 hardware binning would be really nice!
-
Hi,
Both sensors will produce the same amount of detail, but the 48x72 mm sensor will have better microcontrast. The 48x72 mm sensor will also produce a lot of fake detail unless it has a very agressive OLP filter or you stop it down to f/22-f/32. (The Leica S2 seems to cease aliasing when stopped down past f/11, your hypothetical 48x72 sensor would have almost the double pixel size so it would need stopping down two stops more).
Moiré, which is a one form of aliasing, is a problem for many studio photographers with low res backs like 20MP MFDB, so the issue is not only hypothetical.
Regarding printer resolution your 30 MP sensor would resolve around 6700 pixels horisontally. New Epsons at finest setting seem to resolve 720 pixels/inch. So if you don't print wider than 9.3" then you will never exceed the capability of a modern printer. But my standard print size is 26 inch wide.
The best MFDB is probably the Phase One IQ180 and it has 5.2 micron pixels.
Best regards
Erik
Thanks Erik, but it was more of a rhetorical statement. Also I don't think you will find an Epson will print a perfect series of line pairs at its maximum resolution and, even if it did, you would need to view it extremely closely to see it--probably with a loupe. I believe the finests setting for an Epson is 4880 dpi--720 is the lowest, but I would not confuse printer dpi with pixels.
-
Hi,
In the latest c2ps videos Jeff Schewe says that for some pictures the 720 settings matter.
To put things in perspective I have compared 12MP APS-C with 24MP FF in A2-print, using the same setup. In one case, a picture of a mural painting, the pictures could not be told apart, but mostly the difference was subtly visible. The files from the 24 MP camera were always much better.
Some authors say that no more than 12 MP is needed, there is something to it.
On the other hand, as far as the lens outresolves the sensor you are getting more and better information with smaller pixels. If the pixels are to large fake resolution will result, which may enhance visual impression, but nevertheless be fake. With smaller pixels you can sharpen better.
This article is by a well known expert on printing: http://theonlinephotographer.typepad.com/the_online_photographer/2009/02/why-80-megapixels-just-wont-be-enough.html
Best regards
Erik
Thanks Erik, but it was more of a rhetorical statement. Also I don't think you will find an Epson will print a perfect series of line pairs at its maximum resolution and, even if it did, you would need to view it extremely closely to see it--probably with a loupe. I believe the finests setting for an Epson is 4880 dpi--720 is the lowest, but I would not confuse printer dpi with pixels.
-
In the latest c2ps videos Jeff Schewe says that for some pictures the 720 settings matter.
I do my prints via a lab that has a Durst Lambda digital C-printer. I like digital C prints, unfortunately it seems like they're on their way out(?), Durst has stopped making the printers at least.
Anyway, that printer has a resolution limit of 400 ppi. It has some built-in scaling algorithm though, seems to be bicubic, which means that it smooths/smears the pixels somewhat to avoid aliasing. This smoothing is also apparent when there should be no upscaling required. This means that if I feed it exactly with 400 ppi I get a less sharp print than if I feed with with say 800 ppi. The smoothed 400 ppi is similar to effective 300 ppi. Not sure if it is possible to control the Durst to skip the smoothing at 400 ppi, I've tried to ask that to my lab but they're not really into tech nerd things so I have not got an answer (probably they don't know either and don't care to find out) :-).
-
Erik,
I agree with you.
I am not against pixel resolution. 12MP image a great and more MP can give the image more--I actually like lower rez sensors for the more 35mm filmish look or structure or rendering (it does not look like film, buy rather a limit in frequency). Our perception of images is far more complex than just reducing the problem to resolving power and pixel resolution. But likewise just going for more and more pixels has diminishing returns. When making calculations, how many significant figures do you use? Would you build a house with a ruler accurate to a micron? In images, how many pixels?
Personally, I don't know where the limit is to pixel size or to pixel resolution, at least in a simple number. I guess what I react to it a simple view the smaller the better, the more the merrier. Given a choice, I would take fewer, fatter pixels and larger sensors. But at work we do scientific imaging with 1.25MP to 4MP microscope cameras with 7-9um pixels. I have even made 24" prints from those images. So maybe I have a different perspective from most. The manufacturers will keep pushing pixels like drugs; that is what the people (and addicts) want. I am also noticing a loss of acutance in optics because they need more resolving power, so there are systemic changes that do affect the results.
My too sense.
-
In images, how many pixels?
Personally, I don't know where the limit is to pixel size or to pixel resolution, at least in a simple number.
I did a simple amateurish calculation last week and the number I got was 5 microns - barring all else was constant and perfect. It was based on 20/8 vision (Only a tiny fraction of the population has this). Resolution of the 20/8 human eye would equate to 0.4 arc minutes at 20 feet. "Normal" vision (20/20) is 1 arc minute at 20 feet.
Since there are DSLRs out there with less than 5 microns per pixel, I suspect that it might due to resolution losses in the filtering and debayering methods adopted. Also, I would assume compression, sampling and color space all have their respective parts to play.
I can't find any data on filters on the sensors used. If I could, I might be able to get a more realistic number. But I'm just new to all this, and am still learning stuff.
-
Hi,
I'm aware of both 0.4 arc minutes and 1 arc minute, but I don't see how you arrive at the 5 micron figure. The figure would need to include a distance and also a magnification. You never look at pixels unmagnified. For instance if you print A2 and use APS sensor image would be enlarged about 25x so a pixel of 5 micron diameter would be 125 microns.
Assuming 1000 mm viewing distance that would correspond to 0.42 minutes of arc, but you need at leas three point two resolve two. The two points and a darker point in between. That would be 1.2 arc minutes, but I'm not sure about that. Let's say 0.8 minutes of arc. Are you looking at 25 cm which is feasible it would be 3.2 minutes of arc.
But I would say it is reasonable to say that A2 prints normally are viewed at about arm lengths distance. many authors claim that 12 MP is perfectly good for A2 size prints. It also reflects my experience, although I find that in some tests I have made the difference is quite noticeable between 12 MP and 24 MP, depending on subject and conditions.
Nevertheless, there are some advantages with smaller pixels, main advantage may be that risk of aliasing is much reduced. OLP filtering can be reduced and the image will respond better to sharpening.
Best regards
Erik
I did a simple amateurish calculation last week and the number I got was 5 microns - barring all else was constant and perfect. It was based on 20/8 vision (Only a tiny fraction of the population has this). Resolution of the 20/8 human eye would equate to 0.4 arc minutes at 20 feet. "Normal" vision (20/20) is 1 arc minute at 20 feet.
Since there are DSLRs out there with less than 5 microns per pixel, I suspect that it might due to resolution losses in the filtering and debayering methods adopted. Also, I would assume compression, sampling and color space all have their respective parts to play.
I can't find any data on filters on the sensors used. If I could, I might be able to get a more realistic number. But I'm just new to all this, and am still learning stuff.
-
Hi,
This article I have written a couple of years ago may be of interest and it has links to some good articles.
http://echophoto.dnsalias.net/ekr/index.php/photoarticles/24-how-many-megapixels-do-we-need
Best regards
Erik
Erik,
I agree with you.
I am not against pixel resolution. 12MP image a great and more MP can give the image more--I actually like lower rez sensors for the more 35mm filmish look or structure or rendering (it does not look like film, buy rather a limit in frequency). Our perception of images is far more complex than just reducing the problem to resolving power and pixel resolution. But likewise just going for more and more pixels has diminishing returns. When making calculations, how many significant figures do you use? Would you build a house with a ruler accurate to a micron? In images, how many pixels?
Personally, I don't know where the limit is to pixel size or to pixel resolution, at least in a simple number. I guess what I react to it a simple view the smaller the better, the more the merrier. Given a choice, I would take fewer, fatter pixels and larger sensors. But at work we do scientific imaging with 1.25MP to 4MP microscope cameras with 7-9um pixels. I have even made 24" prints from those images. So maybe I have a different perspective from most. The manufacturers will keep pushing pixels like drugs; that is what the people (and addicts) want. I am also noticing a loss of acutance in optics because they need more resolving power, so there are systemic changes that do affect the results.
My too sense.
-
Sorry Erik, but the concepts of standard viewing distance works really well and maximum print size has no baring on pixel resolution. When I publicly displayed a 24" print from a 4MP camera, everyone who saw it were amazed by it--and they did not know what camera it was from, they were just reacting to the quality of the image. Your model is too simple.
-
This article is by a well known expert on printing: http://theonlinephotographer.typepad.com/the_online_photographer/2009/02/why-80-megapixels-just-wont-be-enough.html
Roger Clark (http://www.clarkvision.com/articles/megapixel_myth/) has reached a conclusion similar to that of Ctein. Roger's experiments indicate that the highest print quality requires 600 pixels per inch on the print and this assumption is similar to the conclusion reached by Jeff Schewe in the latest Camera to Print and Screen tutorial. When printing with Epson printers with a native resolution of 720 ppi, Jeff suggests that if the native resolution of the image is greater than 360 ppi, one should print at 720 ppi to preserve maximal image quality. However, one must recognize that with ink jets using error diffusion technology there is a rather nebulous relation between the nominal dpi and the actual resolution.
To attain this resolution even with the Phase One IQ180, one would have to resort to stitching, and this is exactly what Jeff does in some cases. For prints viewed at average distances one can make do with less resolution, but as Bruce Fraser used to say, for some photographers the viewing distance is determined by the distance from the nose to the eye :).
Regards,
Bill
-
There are other image quality tradeoffs involved with small pixels besides noise and resolution. For example, with smaller pixels it is often much tougher to deal with (undesirable) crosstalk between neighboring pixels and uniformity.
-
There are other image quality tradeoffs involved with small pixels besides noise and resolution. For example, with smaller pixels it is often much tougher to deal with (undesirable) crosstalk between neighboring pixels and uniformity.
At what pixel pitch does this start to be a substantial problem? Do you see serious issues with raw data from compacts having 2µ pixels? Nikon 1 series (3.4µ pixels)? Canon 7D (4.2µ pixels)? Nikon D7000 (4.7µ pixels)?
-
Hi,
Blueray has just two MP resolution and it's said that normal cinema film has lower resolution than Blueray.
I have made moderately large (A3-A2) prints from images with low resolution in the 4-6 MP range that were OK. That is a good or great subject with a good sharpening can compensate for lack of detail.
I don't understand your reference to "my model beeing to simple", which model do you mean?
This image is about 4 MP I guess and is quite enjoyable in A3+:
(http://echophoto.dnsalias.net/ekr/images/SampleImages/HowManyMPixels/PICT6192-Edit.jpg)
Best regards
Erik
Sorry Erik, but the concepts of standard viewing distance works really well and maximum print size has no baring on pixel resolution. When I publicly displayed a 24" print from a 4MP camera, everyone who saw it were amazed by it--and they did not know what camera it was from, they were just reacting to the quality of the image. Your model is too simple.
-
BT
There are other image quality tradeoffs involved with small pixels besides noise and resolution. For example, with smaller pixels it is often much tougher to deal with (undesirable) crosstalk between neighboring pixels and uniformity.
BTW, this page has a couple of nice movies and images about micro-lenses and cross-talk
http://www.lumerical.com/fdtd/applications/cmos_image_sensor_pixel_microlens.html
-
Assuming 1000 mm viewing distance that would correspond to 0.42 minutes of arc, but you need at leas three point two resolve two. The two points and a darker point in between. That would be 1.2 arc minutes, but I'm not sure about that. Let's say 0.8 minutes of arc. Are you looking at 25 cm which is feasible it would be 3.2 minutes of arc.
I might be wrong Erik, but here is my line of thought:
The Snellen fraction does not talk about line pairs. At 20/20, a 'normal' eye subtends 5 arc minutes of an optotype. The most commonly used optotype (probably because it has 5 lines) is the letter 'E'. Each line therefore subtends 1 arc minute. For a 'perfect human', the angle would be 0.4 arc minutes, regardless of distance. If the object gets closer, the line width gets smaller, but the angle never changes. Even under magnification (microscopy or telescopy), the perfect human eye can only read at 0.4 arc minutes. But it is 0.4 arc minutes of one line only. Practically not very useful (which is why they use E), but makes calculations easier. It would be unfair to use line pairs on sensors.
How I arrived at 5 microns:
The sensor does not have the field of vision like an eye. The lens compresses the image it to a single plane - the back of the lens. It is as if the sensor is almost watching a movie in a dark hall. Sure, there's light scattering, and other 'leaks', but for our purposes, we can assume our camera lens to be perfect, and treat the back of the lens like an image plane. Therefore, it is not important to consider what is in focus and what isn't, etc. And I assumed the distance to be the focal flange distance (back of the lens to sensor). EF (44mm) and PL (55mm). I'm from the motion picture background. Some cameras like the RB67 (which I own) has a flange focal distance of 112mm. Obviously this figure varies as the lens zooms, focuses, etc. For simplicity sake, I have assumed it to be constant (I did say it was amateurish).
The smallest readable line that leaves the lens must subtend an angle of 0.4 arc minutes to a point behind the pixel - because the pixel is not a point object but a line itself. However, a pixel never has to resolve anything greater than 0.4 arc minute, so we can safely assume that a pixel must be smaller than the width of the line that is subtended on the back of the lens. Treating the sensor like the eye, and going by the fact that a pixel can always have one and only one value, it can be treated like a point object if only for theoretical calculation. In this case, if the angle is 0.4 arc minute and the distance is 44mm (for EF), say, then the width of the line is 5.112 microns.
For the EF mount:
@0.4 arc min, minimum line/pixel/dot width is 5 microns
@1 arc min, minimum line/pixel/dot width is 12.76 microns
For the RB67:
@0.4 arc min, minimum line/pixel/dot width is 13 microns
@1 arc min, minimum line/pixel/dot width is 32.5 microns
For the Pentax Q mount (the smallest I could find) (9.2mm flange)
@0.4 arc min, minimum line/pixel/dot width is 1 micron
@1 arc min, minimum line/pixel/dot width is 2.7 microns
Assuming a viewing distance of:
@ 12 inches (305mm): the line width @ 0.4 arc min must be 35.5 microns or 0.035mm. This corresponds to 726 ppi at 0.4 arc min and about 290 ppi (very close to 300 dpi isn't it?) for 1 arc min.
@ 2.5 feet (762mm) (normal computer monitor viewing distance): @ 0.4 arc min, the ppi is 287 ppi and @ 1 arc min the ppi is 115 ppi. Right now, the average screen is only 100 ppi. Professional monitors aim for higher ppi to justify this resolution.
@ 1000mm the width of one line is 0.12mm or 116.3 microns. This corresponds to 212 ppi @ 0.4 arc min, and 88 ppi @ 1 arc min.
@ 6 feet or 1830mm - the 'normal' viewing distance for home television. The ppi required is 120 @ 0.4 arc min. @ 1 arc min the ppi required is only 48 ppi. My LCD screen (42") has a horizontal ppi of 64. This is why 2 MP is okay for home viewing, even on a large monitor. Smaller TV sets need lesser ppi, and larger sets will be viewed farther away. The 72 dpi standard for TV is okay. The same standard is used for film projection (where half the resolution is lost in projection).
Side note: Line pairs per mm for EF mount:
Since the minimum required line width is 5 microns for an EF mount lens, a line pair would be 10 microns. Therefore, the lp/mm required would be 100 lp/mm corresponding to 0.4 arc min. @ 1 arc min, it is 40 lp/mm. The highest possible Canon EF lens (according to DXO) is the 85mm L which has a lp/mm of 67. Of course, it all depends on the testing and cameras used, etc.
Printing:
I'm not an expert on printing, but the maximum 720 ppi limit seems justified, if all things are perfect @ 1 feet. However, the Epson 9900 is capable of 2880 ppi. Paper resolutions are easily greater than 60 lp/mm or around 720 ppi+ to 1440 dpi for Epson Fine Art papers. Of course, it is important to avoid interpolation when making these calculations. A drum scanner is capable of 3000 dpi to 24,000 dpi, and a flatbed scanner can top 5000 dpi.
As I've mentioned before, my calculations make many assumptions and are amateurish. I did these calculations just to understand how the standards that are being used have come about. I probably have made mistakes along the line (hopefully not grave ones). But I'm also secretly glad they correspond to the standards currently used in photography and cinema today.
For the full text with links to sources, please visit: http://sareesh.com/2011/12/resolution-visual-acuity-and-lpmm/
-
Hi Emil, it depends on the sensor design (e.g., filter stack height on top of the recording layer), but generally I've seen more and more crosstalk and uniformity issues in the progression towards 5 and 4 micron pixels over the past couple of years. Some of this can be addressed in software, but it is a tradeoff because of potential impact on resolution ... which of course is one of the motivations in going to smaller pixels.
-
Hi,
Thanks for sharing your calculations.
In my view you are ignoring a few factors. To begin with I don't think that sensor to flange distance is the key parameter. The optical center of the lens may be in front or behind the lens mount. I'd suggest that you could simply use the focal length in your calculations.
The other issue I have is, that according to my understanding, that you try to compare the angular resolution of the eye with the angular resolution of the sensor. But it has never been the case that we would opt for the same angular resolution in the imaging system. Let's just look at some glossy magazine. An article may be illustrated with some image, next page perhaps would contain a full page image and a third one have double spread image. All these images would be seen at the same distance having the same angular resolution, but very differing field of view.
The images would be observed using the same angular resolution, but the magnification would vary, like a factor of ten.
Another observation I may have made that the pixel size of the eye seems to be around 4 microns (cones are said to have a radius of 2 microns), that is pretty close to your estimate of 5 microns.
Best regards
Erik
I might be wrong Erik, but here is my line of thought:
The Snellen fraction does not talk about line pairs. At 20/20, a 'normal' eye subtends 5 arc minutes of an optotype. The most commonly used optotype (probably because it has 5 lines) is the letter 'E'. Each line therefore subtends 1 arc minute. For a 'perfect human', the angle would be 0.4 arc minutes, regardless of distance. If the object gets closer, the line width gets smaller, but the angle never changes. Even under magnification (microscopy or telescopy), the perfect human eye can only read at 0.4 arc minutes. But it is 0.4 arc minutes of one line only. Practically not very useful (which is why they use E), but makes calculations easier. It would be unfair to use line pairs on sensors.
How I arrived at 5 microns:
The sensor does not have the field of vision like an eye. The lens compresses the image it to a single plane - the back of the lens. It is as if the sensor is almost watching a movie in a dark hall. Sure, there's light scattering, and other 'leaks', but for our purposes, we can assume our camera lens to be perfect, and treat the back of the lens like an image plane. Therefore, it is not important to consider what is in focus and what isn't, etc. And I assumed the distance to be the focal flange distance (back of the lens to sensor). EF (44mm) and PL (55mm). I'm from the motion picture background. Some cameras like the RB67 (which I own) has a flange focal distance of 112mm. Obviously this figure varies as the lens zooms, focuses, etc. For simplicity sake, I have assumed it to be constant (I did say it was amateurish).
The smallest readable line that leaves the lens must subtend an angle of 0.4 arc minutes to a point behind the pixel - because the pixel is not a point object but a line itself. However, a pixel never has to resolve anything greater than 0.4 arc minute, so we can safely assume that a pixel must be smaller than the width of the line that is subtended on the back of the lens. Treating the sensor like the eye, and going by the fact that a pixel can always have one and only one value, it can be treated like a point object if only for theoretical calculation. In this case, if the angle is 0.4 arc minute and the distance is 44mm (for EF), say, then the width of the line is 5.112 microns.
For the EF mount:
@0.4 arc min, minimum line/pixel/dot width is 5 microns
@1 arc min, minimum line/pixel/dot width is 12.76 microns
For the RB67:
@0.4 arc min, minimum line/pixel/dot width is 13 microns
@1 arc min, minimum line/pixel/dot width is 32.5 microns
For the Pentax Q mount (the smallest I could find) (9.2mm flange)
@0.4 arc min, minimum line/pixel/dot width is 1 micron
@1 arc min, minimum line/pixel/dot width is 2.7 microns
Assuming a viewing distance of:
@ 12 inches (305mm): the line width @ 0.4 arc min must be 35.5 microns or 0.035mm. This corresponds to 726 ppi at 0.4 arc min and about 290 ppi (very close to 300 dpi isn't it?) for 1 arc min.
@ 2.5 feet (762mm) (normal computer monitor viewing distance): @ 0.4 arc min, the ppi is 287 ppi and @ 1 arc min the ppi is 115 ppi. Right now, the average screen is only 100 ppi. Professional monitors aim for higher ppi to justify this resolution.
@ 1000mm the width of one line is 0.12mm or 116.3 microns. This corresponds to 212 ppi @ 0.4 arc min, and 88 ppi @ 1 arc min.
@ 6 feet or 1830mm - the 'normal' viewing distance for home television. The ppi required is 120 @ 0.4 arc min. @ 1 arc min the ppi required is only 48 ppi. My LCD screen (42") has a horizontal ppi of 64. This is why 2 MP is okay for home viewing, even on a large monitor. Smaller TV sets need lesser ppi, and larger sets will be viewed farther away. The 72 dpi standard for TV is okay. The same standard is used for film projection (where half the resolution is lost in projection).
Side note: Line pairs per mm for EF mount:
Since the minimum required line width is 5 microns for an EF mount lens, a line pair would be 10 microns. Therefore, the lp/mm required would be 100 lp/mm corresponding to 0.4 arc min. @ 1 arc min, it is 40 lp/mm. The highest possible Canon EF lens (according to DXO) is the 85mm L which has a lp/mm of 67. Of course, it all depends on the testing and cameras used, etc.
Printing:
I'm not an expert on printing, but the maximum 720 ppi limit seems justified, if all things are perfect @ 1 feet. However, the Epson 9900 is capable of 2880 ppi. Paper resolutions are easily greater than 60 lp/mm or around 720 ppi+ to 1440 dpi for Epson Fine Art papers. Of course, it is important to avoid interpolation when making these calculations. A drum scanner is capable of 3000 dpi to 24,000 dpi, and a flatbed scanner can top 5000 dpi.
As I've mentioned before, my calculations make many assumptions and are amateurish. I did these calculations just to understand how the standards that are being used have come about. I probably have made mistakes along the line (hopefully not grave ones). But I'm also secretly glad they correspond to the standards currently used in photography and cinema today.
For the full text with links to sources, please visit: http://sareesh.com/2011/12/resolution-visual-acuity-and-lpmm/
-
I might be wrong Erik, but here is my line of thought:
...............................
Printing:
I'm not an expert on printing, but the maximum 720 ppi limit seems justified, if all things are perfect @ 1 feet. However, the Epson 9900 is capable of 2880 ppi. Paper resolutions are easily greater than 60 lp/mm or around 720 ppi+ to 1440 dpi for Epson Fine Art papers. Of course, it is important to avoid interpolation when making these calculations. A drum scanner is capable of 3000 dpi to 24,000 dpi, and a flatbed scanner can top 5000 dpi.
As I've mentioned before, my calculations make many assumptions and are amateurish. I did these calculations just to understand how the standards that are being used have come about. I probably have made mistakes along the line (hopefully not grave ones). But I'm also secretly glad they correspond to the standards currently used in photography and cinema today.
For the full text with links to sources, please visit: http://sareesh.com/2011/12/resolution-visual-acuity-and-lpmm/
I think you may be inappropriately commingling the concept of PPI (pixels per inch) sent to the printer and DPI (dots per inch) at which the printer is specified for reproducing the image. The 2880 you mention is printer DPI. Even that metric is impossible to anchor as an end-point to the argument you are making, because the screening and dithering algorithms of these printers make such an exercise anything but straightforward; there is no simple relationship between PPI sent in and DPI coming out because of those algorithms, which are part of Epson's "secret sauce".
As a practical matter, I am now making prints with excellent properties of tonal gradation and resolution of fine detail from my 16.5 MP Sony NEX 5n sending the image over to my Epson 4900 at anything in the range of 240~296 PPI (no resampling) and printing it at about 16.3*11 inches on a Super A3 sheet, with printer set to High Speed OFF (not necessary but I do it anyhow) and Finest Detail (2880 DPI) ON (also not necessary - 1440 would probably do just as well). I just go all the way with these settings (per the Epson manual) to be certain the printer is doing the best it can for the PPI going in. So far I am very pleased with the quality of the results, whether viewed at normal viewing distance for this size image or examined in more detail at reading distance. The prints compete with output from my FF Canon 1DsMk3, which latter of course provides more leeway for cropping without compromising on the sweet range of PPI (no resampling) sent to the printer.
I think one can get carried away on theoretical arguments - which may or may not have their internal validity - but in the final analysis what matters most is what comes out of the printer and what your eyes see. There may be quite a range of theory-driven specifications over the which the real appearance of the image wouldn't differ either at all or enough to constitute an issue.
-
In my view you are ignoring a few factors. To begin with I don't think that sensor to flange distance is the key parameter. The optical center of the lens may be in front or behind the lens mount. I'd suggest that you could simply use the focal length in your calculations.
My aim was to avoid factoring the lens into the equation. I could be wrong, but the assumption I made is that the back of the lens acts like a screen to the sensor. Also, once the light crosses the focal point, it can still be affected by the 'rest of the lens elements' on its way out. But the extent of my knowledge regarding this subject ends here, and I probably took the easy way out.
The other issue I have is, that according to my understanding, that you try to compare the angular resolution of the eye with the angular resolution of the sensor. But it has never been the case that we would opt for the same angular resolution in the imaging system. Let's just look at some glossy magazine. An article may be illustrated with some image, next page perhaps would contain a full page image and a third one have double spread image. All these images would be seen at the same distance having the same angular resolution, but very differing field of view.
Just wanted to clear a doubt: Isn't magnification mainly the job of the lens? Even if a scene were magnified, the pixel (or eye) would still have the same built in angular resolution. The point was not to match the angles, but to find the minimum required angle, in this case 0.4 arc minutes. A pixel whose angle is lesser can out-resolve the eye, and help in interpolation to large print or display sizes. Have I understood correctly?
Another observation I may have made that the pixel size of the eye seems to be around 4 microns (cones are said to have a radius of 2 microns), that is pretty close to your estimate of 5 microns.
Best regards
Erik
One thing I read was about Hyperacuity. To quote wikipedia:
Under optimal conditions of good illumination, high contrast, and long line segments, the limit to vernier acuity is about 8 arc seconds or 0.13 arc minutes, compared to about 0.6 arc minutes (20/12) for normal visual acuity or the 0.4 arc minute diameter of a foveal cone.
. Link here: http://en.wikipedia.org/wiki/Visual_acuity
-
I think you may be inappropriately commingling the concept of PPI (pixels per inch) sent to the printer and DPI (dots per inch) at which the printer is specified for reproducing the image. The 2880 you mention is printer DPI. Even that metric is impossible to anchor as an end-point to the argument you are making, because the screening and dithering algorithms of these printers make such an exercise anything but straightforward; there is no simple relationship between PPI sent in and DPI coming out because of those algorithms, which are part of Epson's "secret sauce".
As a practical matter, I am now making prints with excellent properties of tonal gradation and resolution of fine detail from my 16.5 MP Sony NEX 5n sending the image over to my Epson 4900 at anything in the range of 240~296 PPI (no resampling) and printing it at about 16.3*11 inches on a Super A3 sheet, with printer set to High Speed OFF (not necessary but I do it anyhow) and Finest Detail (2880 DPI) ON (also not necessary - 1440 would probably do just as well). I just go all the way with these settings (per the Epson manual) to be certain the printer is doing the best it can for the PPI going in. So far I am very pleased with the quality of the results, whether viewed at normal viewing distance for this size image or examined in more detail at reading distance. The prints compete with output from my FF Canon 1DsMk3, which latter of course provides more leeway for cropping without compromising on the sweet range of PPI (no resampling) sent to the printer.
I think one can get carried away on theoretical arguments - which may or may not have their internal validity - but in the final analysis what matters most is what comes out of the printer and what your eyes see. There may be quite a range of theory-driven specifications over the which the real appearance of the image wouldn't differ either at all or enough to constitute an issue.
You are absolutely right, Mark. It was an overly simplistic assumption on my part.
-
Hi,
Look at the enclosed images, both are taken at the same time (within a few minutes) at the same spot (within a few meters) but with different lenses, one of the images is taken with 16 mm lens and the other with a 70 mm lens.
The 16 mm lens has a horizontal FOV of 132 degrees, the sensor here resolves around 6000 pixels that is 1.3 minutes of arc
The 70 mm lens has a horizontal FOV of 54.5 degrees, the sensor has still 6000 pixels so it resolves 0.54 minutes of arc.
If you would print both images at say 60 cm width and looked at them at 80cm the angle of view covering one pixel would be 0.43 minutes of arc for both images.
Best regards
Erik
Just wanted to clear a doubt: Isn't magnification mainly the job of the lens? Even if a scene were magnified, the pixel (or eye) would still have the same built in angular resolution. The point was not to match the angles, but to find the minimum required angle, in this case 0.4 arc minutes. A pixel whose angle is lesser can out-resolve the eye, and help in interpolation to large print or display sizes. Have I understood correctly?
One thing I read was about Hyperacuity. To quote wikipedia:. Link here: http://en.wikipedia.org/wiki/Visual_acuity
-
The 16 mm lens has a horizontal FOV of 132 degrees, the sensor here resolves around 6000 pixels that is 1.3 minutes of arc
The 70 mm lens has a horizontal FOV of 54.5 degrees, the sensor has still 6000 pixels so it resolves 0.54 minutes of arc.
If you would print both images at say 60 cm width and looked at them at 80cm the angle of view covering one pixel would be 0.43 minutes of arc for both images.
Best regards
Erik
Thanks for sharing, Erik.
Now I know where we differed. I did not consider the FOV angle of the lens because I wanted to disregard the lens from the equation. In my calculations, since I used the back of the lens as a screen, the angle does not matter at all - since it is the compressed result that is more important. E.g., for a 18-200mm lens, no matter what the angle is when the light enters, when it exits, it is compressed to the same plane, and the lens has the task of compressing resolution. The sensor remains an "independent" observer!
-
Roger Clark (http://www.clarkvision.com/articles/megapixel_myth/) has reached a conclusion similar to that of Ctein. Roger's experiments indicate that the highest print quality requires 600 pixels per inch on the print ...
My reading is that his experiments show an advantage of 600ppi over 300ppi, which is consistent with the threshold being anywhere above 300, not necessarily all the way up to 600. At one stage, laser printers moved to 400DPI, which seemed to improve the sharpness of text; after that there was a move to 600DPI and beyond, but driven at least in part to support the dithering needed to reproduce gray-scale images. Note my shift to DPI, because once the dots of ink or toner are being used to produce finer tonal gradations than the dots themseves have, the DPI needs to be higher than the PPI of the input image file. But with pure black on white text reproduction, PPI=DPI is reasonable.
So let me throw out for discussion and testing the hypothesis that for the most demanding case of pure black and white (where jaggies are most noticeable, due to the exteme edge contrast) 400PPI is enough while 300PPI is not always enough. And for almost all photography, other than photographs of text and high contrast test patterns, the visually distinguishable level is likely to be a bit lower.
By the way, at 400PPI (16pp/mm), A4 and US letter paper are about 16MP, so A3 is 32MP and so on.
P.S. on this viewing distance thing: with many large prints (including all billboards!) the effect of size is to lead to viewing from further away, like several people viewing and discussing a large print in a gallery. Even with huge 4'x5' from LF film, I most often see people staying four or five feet away, so not much closer than "normal viewing distance". But that is not always the case, so the photographer planning on large prints has to work out what the viewing style is likely to be, or what viewing style he/she wishes to support, not us online theoreticians.
-
P.S. on this viewing distance thing: with many large prints (including all billboards!) the effect of size is to lead to viewing from further away, like several people viewing and discussing a large print in a gallery. Even with huge 4'x5' from LF film, I most often see people staying four or five feet away, so not much closer than "normal viewing distance". But that is not always the case, so the photographer planning on large prints has to work out what the viewing style is likely to be, or what viewing style he/she wishes to support, not us online theoreticians.
But is it a failure if the viewer is able to spot image "flaws" by walking up to image distances that he/she was never supposed to do? What if she brings a magnifying glass? Is the value of a Picasso any diminished by knowledge about the limitations in the process of making it, or knowledge that the master changed his mind a few times, revealed by x-rays or multi-spectral imaging? I think not.
Someone will always find a flaw. I imagine that those people would never approve (or buy) the image in the first place, they are just looking for excuses.
I believe that a good image processing chain should look good at a range of distances, and that the consequence of walking "too close" should be a lack of expected details (lowpass filtering), not "stair stepping", "aliasing", "mosquitoes" or other nasty artifacts.
-h
-
But is it a failure if the viewer is able to spot image "flaws" by walking up to image distances that he/she was never supposed to do?
I agree; for my photography, such close viewing is "user error" (I call it "map reading"). That is why for my purposes, I am comfortable with about 12MP after cropping, no matter how big i might display the image, and aspire to more pixels than that solely to facilitate some macro and telephoto work when I cannot get close enough to fill the frame of the lens I have with me. I was however allowing that others have different intentions for their prints. (Hopefully there is better evidence than Roger Clark putting a big print on a staircase where it can only be viewed from close quarters, and then claiming this as evidence that people want to look at huge prints at very short range!)
P.S. I prefer to look at extreme resolution needs in terms of line pairs per picture height, avoiding the complications of how many pixels one needs to fully resolve a line pair, and issues like over-sampling to avoid moiré. For another day, I would argue that there is very, very little use for more than about 4000 lp/ph, due to the collision of diffraction and DOF needs: as the desired apparent image size increases (bigger prints/closer viewing) diffraction control requires larger aperture sizes while maintaining equal perceived DOF requires smaller aperture sizes until you are crushed between conflicting needs. 4000lp/ph could mean about 100MP, but nowhere near Ctein's 400MP.
-
Hi,
My lab suggests 200 LPI for sizes above A3. The largest prints I have 70x100 cm and 50x100 cm. One of the 70x100cm prints made from 67 Velvia scanned at 3200 PPI is extremely sharp, but has clearly visible "staircase" effect close up. Another 70x100 cm is made from 10 MP APS-C camera, it is OK at 80 cm viewing distance.
I would say that the philosophy that 12 MP is all you need has some merits, but the other philosophy saying that you need all pixels you have is not without merit.
Sometimes I feel that I want to walk "into" a picture and be immersed in it. Panoramas can be like that,
Best regards
Erik
My reading is that his experiments show an advantage of 600ppi over 300ppi, which is consistent with the threshold being anywhere above 300, not necessarily all the way up to 600. At one stage, laser printers moved to 400DPI, which seemed to improve the sharpness of text; after that there was a move to 600DPI and beyond, but driven at least in part to support the dithering needed to reproduce gray-scale images. Note my shift to DPI, because once the dots of ink or toner are being used to produce finer tonal gradations than the dots themseves have, the DPI needs to be higher than the PPI of the input image file. But with pure black on white text reproduction, PPI=DPI is reasonable.
So let me throw out for discussion and testing the hypothesis that for the most demanding case of pure black and white (where jaggies are most noticeable, due to the exteme edge contrast) 400PPI is enough while 300PPI is not always enough. And for almost all photography, other than photographs of text and high contrast test patterns, the visually distinguishable level is likely to be a bit lower.
By the way, at 400PPI (16pp/mm), A4 and US letter paper are about 16MP, so A3 is 32MP and so on.
P.S. on this viewing distance thing: with many large prints (including all billboards!) the effect of size is to lead to viewing from further away, like several people viewing and discussing a large print in a gallery. Even with huge 4'x5' from LF film, I most often see people staying four or five feet away, so not much closer than "normal viewing distance". But that is not always the case, so the photographer planning on large prints has to work out what the viewing style is likely to be, or what viewing style he/she wishes to support, not us online theoreticians.
-
Hi,
My lab suggests 200 LPI for sizes above A3. The largest prints I have 70x100 cm and 50x100 cm. One of the prints ... has clearly visible "staircase" effect close up.
... but the other philosophy saying that you need all pixels you have is not without merit.
Sometimes I feel that I want to walk "into" a picture and be immersed in it. Panoramas can be like that,
On the staircase effect: can that not be dealt with by upsampling and then dithering/smoothing?
As to "all the pixels you can get" ... as I hinted in my last reply, I can see thtomsk a considerable degree, but only up to the diffraction/DOF limit. Once you get the resolution that the pixels promise by using a large enough aperture, the pleasure of close examination will be limited to elements of the scene that are very close to the plane of exact focus, due to the magnification of OOF effects as you view at such great enlargement. Though I suppose that the dedicated big print landscapist can use focus stacking or such to overcome that.
-
Hi,
Regarding diffraction, I start to ask my self how much of detail lost to diffraction can be recovered by deconvolution?
If you check this series of images: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=1 the degradation of fine detail contrast when stopping down is quite obvious on the left column.
On the other hand, this part of the article http://echophoto.dnsalias.net/ekr/index.php/photoarticles/49-dof-in-digital-pictures?start=2 may indicate that restoration works well.
Best regards
Erik
On the staircase effect: can that not be dealt with by upsampling and then dithering/smoothing?
As to "all the pixels you can get" ... as I hinted in my last reply, I can see thtomsk a considerable degree, but only up to the diffraction/DOF limit. Once you get the resolution that the pixels promise by using a large enough aperture, the pleasure of close examination will be limited to elements of the scene that are very close to the plane of exact focus, due to the magnification of OOF effects as you view at such great enlargement. Though I suppose that the dedicated big print landscapist can use focus stacking or such to overcome that.
-
Erik,
That is a good point: AFAIK, diffraction transforms the image in a reversible way, so that with sufficiently high resolution, the right sharpening transformation could undo it completely. Though the algorithm might need to know a lot about the particular lens, like the shape and position of the aperture diaphragm.
So perhaps the ultimate landscape camera of the future will be a massively oversampled pinhole camera (no aberrations!) backed up by enough computing power to unravel the diffraction.
-
Erik,
That is a good point: AFAIK, diffraction transforms the image in a reversible way, so that with sufficiently high resolution, the right sharpening transformation could undo it completely. Though the algorithm might need to know a lot about the particular lens, like the shape and position of the aperture diaphragm.
So perhaps the ultimate landscape camera of the future will be a massively oversampled pinhole camera (no aberrations!) backed up by enough computing power to unravel the diffraction.
Any attenuation of signal prior to noisy sampling can only be corrected by amplification if this does not make the resulting noise objectional.
-h
-
Any attenuation of signal prior to noisy sampling can only be corrected by amplification if this does not make the resulting noise objectional.
Darn, you are right, and that means that in our attempt to reduce the effects of diffraction and OOF effects and shot noise and sensor read noise, we end up needing ever longer exposure times, and thus subject motion and such will set other limits on resolution.
My only conclusion is that we are only going to answer this question of pixel count for particular photographic situations, through experiment. My experiment is to keep upgrading until I stop seeing any significant benefit ... but frankly, I think I am already there.
-
But is it a failure if the viewer is able to spot image "flaws" by walking up to image distances that he/she was never supposed to do? What if she brings a magnifying glass? Is the value of a Picasso any diminished by knowledge about the limitations in the process of making it, or knowledge that the master changed his mind a few times, revealed by x-rays or multi-spectral imaging? I think not.
Someone will always find a flaw. I imagine that those people would never approve (or buy) the image in the first place, they are just looking for excuses...
Very ,very true. Especially the "buy" part. :) My (albeit somewhat limited) experience in galleries, with either paintings or photography, is that the people walking right up to a piece to inspect it are other artists, looking to judge the it on technical merits, or people who are just curious but not really looking at it as a whole. They're not buyers. People who do buy are those who connect to the piece emotionally at some level, and they typically maintain a "proper" distance from it.