One thing that never gets mentioned is that perceived megapixels and file size can be dramatically increased in software such as Photoshop and Onone. . I took a 16 megapixel image from 25 megabytes to 175 megabytes and made an acceptable 40in by 60in print.
You are the best at covering this stuff. You cover all the bases, and you're no dismissive about any aspects of the subject matter. I don't see anyone else giving such comprehensive explanations that bring the technical side into such a user-friendly explanation. Nice work.
Love your lessons. I used to teach photography classes. So far your are the only one that consider all aspects and variables in a subject matter. Your technical knowledge WOW.. thanks.
Great video, I have watched so many video of yours and by many other good photographers in youtube but i have to admit but you are the one with the most analysis and i like this, keep going
Tony, you absolutely hit the nail on the head when you said, "300 dpi is the ideal image density for looking at something up close". The key words here are, "for looking at something up close". In my experience, as a professional photographer who has been in commercial photography for over 40 years, the whole dpi or ppi idea is ludicrous. If the 300 dpi theory is correct, then how on earth did they produce 10 x 8 ft billboard posters from my 35mm film transparency images, back in the day? The answer is (surprisingly) that the bigger the image you're printing, then the less dpi you actually need in order to produce what's perceived to be a 'pin sharp' print. Put simply, that's because you view a billboard from about 30ft away. A 20x16" print from a few yards away. And a 5x4" print with your nose pressed against the paper. Only the minuscule 5x4 needs that 300 dpi. If people really want to see their images at their best, then print big and don't be too concerned about the image resolution of your camera's sensor. The human eye won't notice. Indeed, as artists have known for centuries, we're not looking for exact mathematical precision ; the best figurative images are impressionistic. What appears to be a simple crude mark or brushstroke, when viewed from too close, is actually an elegantly expressive rendering when viewed from an appropriate distance. Only "rivet counter" self-appointed photography judges will stick their nose into a 20x16" print -- probably with a magnifying loup -- to base their judgement on what they like to think of as "scientific" judgements. And who cares what such people think. A truly great picture is best viewed in a large print size, where the viewer will stand well back, in order to appreciate the subject and its artistic rendering. Keep up the good work, Tony. You are a brave man to attempt to educate people that great photography is about art, rather than anything that can be broken down into pseudo-science. You'll take a lot of criticism from the rivet counting zealots, but your efforts to educate the wider community are very much appreciated. Rick Bear (Professional, award-winning, photographer.)
8 років тому+94
As a non-US viewer I would appreciate that you also add a column with metric equivalents. Perhaps there aren't that many of us, but from the comments in previous posts I get the impression that there are at least some. Well, just a suggestion for what is already splendid.
+Richard Rönnbäck Actually the US and UK are the last remaining significant markets using legacy inches. They are actually in the minority. Also US and UK somehow magically know about lens MILLIMETRES - LOL
The reason you need megapixels is so you can crop and process your images. When you straighten, add local contrast (clarity), sharpen, noise reduce, etc. you're doing arithmetic on neighbouring pixel values. That arithmetic will blur features that are close to one pixel on the sensor. But if you have a much higher megapixel image to work with so that the features you're hoping to see are five or ten pixels across, the processing algorithms will work much better and the end result will look sharper even if you export a 2 Megapixel JPEG at the end.
+Pavlos Papageorgiou Extremely well put and very much worth saying. I kept expecting Tony to make this point clearly but I suppose he figured it was a little too complex. He did allude to the idea when he showed the two 300dpi prints that looked different. He was right to emphasize the roll played by the lens but processing deserves more mention than it got.
This is too good of a reply to come from youtube comments. You gave a clear succinct answer that was on topic without being rude to anybody. Are you sure you understand how youtube comments work. ;) Thank you for being a sane voice on here.
Anonymity is a disinhibitor, like alcohol, so a lot of commenters are like bad drunks; feeling empowered but impotent and over-stimulated. They type like they’re drunk too. All those exclamation marks! And they’re never able to spell “they’re” correctly. I misspelled “role” in my original reply.
Very good answer. Yup, if you really haven't bothered to expend your efforts in learning to develop your photographic skills, then I guess you can never have enough gigapixels upon which to base your faint hopes of discovering a recoverable image in post processing. 😜, the Bear
Having been using my 16MP for 2 years. Just started looking for a new one, this saved me from the MP myth trap. Bad composed photos with more dots are still bad. Thanks Tony!
Thank you for being smarter then me. Things I would never think would even interest me or have any concern with my photography and your way of explanation just has me hooked and understanding. Thank you again for all that you do!
Thank you for this comprehensive, well-argued and informative video Tony. I especially liked that, by way of reassurance for those who might have been made insecure by the earlier parts of your presentation, you put things into perspective in your conclusion: namely that in the everyday world, viewing distance is a vital consideration. I'd only add that the same point applies to 4k TV's versus 1080p TV's (not 4k monitors, which one views at a far closer distance): the farther away from the screen, the less perceived benefit you will derive. Thanks again.
I forgot to add that I really enjoy videos that explain tech aspects. I'm just learning photography and you and Chelsea are my favorite resource. Thank You!
Very interesting. Never heard of the p-mp rating for glass. So what it looks like, is with 36mp and higher sensors we're starting to reach the point of diminishing returns as far as sharpness is concerned - at least with current lens technology?
Yeah! That was clear to me when I upgraded fromm my old tamron 70-200mm 2.8 non VR (9mpix on 70D) to its new Tamron 70-200mm 2.8 VR (16mpix, 70d). Its AMAZING the difference on the results, Im stunned and in love with this lens. I think 24-30 mpix is enough for everybody, 20 is perfect for sports. I've noticed that many lenses doesnt come close to 35 pmpix on 5Ds, only zeiss, thats something to think about.
This video is great, especially the calculations for getting between print dimensions and pixels, and perceptual megapixels for the lens camera combination. Brought together things I had noticed and suspected, now I don't have to blindly rely on the table I've been using for print size. THANK YOU! Just started stalking birds in my back yard with my new Zeiss batis 85mm on my Sony a7riii with magnified peak focusing, phenomenal wing detail.
+Bror Svensson Media It's actually the blending mode. They're just messing around with the curves to create a cross-process effect resulting in terrible greenish and oversaturated skin tones. A blending mode of soft light, color or hue with a 10% opacity would do the trick.
+Bror Svensson Media I think its the Wall paper, it seems to be difficult to light or balance, they have tried to warm up the picture recently, in fact Tony needs to decorate, or he needs to dye his hair violet or something as its camouflaged into the grey (ish) background. ;-)
One thing I notice is that it kind of reach the limit of lens resolution when switching to 50mp 5ds 5dsr as the p-mpx improvement is so minor, even using the Otus 85, the very lengendary lens in the world. 36 to 40 mp seems to be a sweet point as the extra mp improvement doesn't convert to more actual information you get.
Also, it depends on how you print it. I got one of my pictures, which I have taken with a Canon PowerShot S120 (12MP, really not that good), printed out at 23"*31" on a canvas, and it really looks sharp
I did an extreme extensive test on this Tony. I am OCD so I did every thing possible with this. I put the 5DSR up against the A7R, against the Canon 6D. Even with pretty extreme crops, it wasn't as night and day as I thought it was going to be. I have a pretty good size print and although the A7R has more resolution, it didn't seem to have much advantage over the 6D. The 5DSR has a lot of details, they are noticeable, especially when cropping. However, even then, it didn't make as big of a deal as I thought it would, unless you're doing a really small crop out of a picture. I could notice that it had more detail, however, the picture on the 6D, which was shot with a really sharp lens, had a ton of detail too and the extra detail I don't think a non photographer would have noticed. BUT with computer monitors, it gets to be even less of an issue. Most people don't have 4k yet, that's only 8 megapixels, we don't have any computer monitor that can see all the resolution of a Canon 6D. Of course we see the difference in zooming in and cropping, but unless it's an extreme crop, it just wasn't as big of a deal as I had made it out to be in my head. The reason I did this test is because I'm a landscaper, I was shooting with the A7R but missed my Canon 5D Mark iii, ergonomics, colors, etc. I thought about going to the 5DSR, so I tested it, but it just didn't offer as much as I thought it would over the Canon 6D. Which is what I have now. I didn't get the 5D Mark iii again because there was nothing it offered me as a landscaper that the 6D didn't have. Now, with that said, with ISO going down, and other things that come with adding more megapixels, I am hoping the next 5D is 28 megapixels. I really feel 28-36 megapixels is the sweet spot right now. The 24-70 2.8 ii and 135mm F/2 from Canon are extremely sharp, they provided a ton of detail with no softness. I used the Zeiss 55mm 1.8 on the A7R as well as the Zeiss 16-35 F/4.
I think sharpness is somewhat important for me being a filmmaker, I don’t have money for Ziess or Cooke lenses but clarity is still important to my art. You make a lot of sense Tony..... Love this channel
I use PPI when talking about digital files, and DPI when talking about actual print density. This way there is no confusion between file export setting and printer settings. I have seen files saved at 600 PPI becouse they tried to match printer's 600 DPI
Any idea why the DXO p-pmix scores are consistently higher for the Canon T5I than the same lenses on the canon 7D mark II? I was wanting to make the upgrade but I'm confused that there is such a consistent lower scores for the sharpness of the 7D mark II. Am I misinterpreting something?
Tony I have a question and you seem like the right guy to answer it. I have a pentax k-3ii with a 100mm macro lens and a 1.4x teleconverter. The pentax has a 1.5x crop factor and the macro lens shoots 1:1 magnification. Along with the teleconverter, what is my total magnification number at the closest focusing distance on the macro lens with the crop factor of the camera? How do I work out that formula? And what is my overall equivalent crop factor with the camera and the teleconverter?
I really appreciate the scope and range of the videos you guys put out - techy stuff like this, the new informative podcast and the creative focused 'live' shows really provide fantastic educational content - this has caused me to purchase your books which is I guess is the purpose :)
In a video you said that using a full frame lens on a micro four thirds camera produces less sharp images because the pixel density of micro four thirds is higher than full frame cameras. Am I missing something here?
DxO image testing is based on RAW, so images can still be sharpened in post-processing. Also depending on the type of photograph total P-Mpix across the frame might be completely irrelevant. A portrait lens for example might be extremely sharp in the central portion of the frame but unsharp in the extreme corners. You might likely place your subject toward the centre of frame (maybe slightly off-center to rule of thirds nodes), so you may not end up needing the corner pixels for capturing fine detail. Also, for portraits, these corner pixels might very well be representing out-of-focus blur instead of fine detail.
Hi Tony, do you mind making a video explaining the difference between depth of field and lens compression. i think its a really great topic to cover since a lot of people do not know the difference between those two
Hi Tony. Wasn't your formula predicting the 18PMP scoring Sigma 24-105mm on a 5D3 would rise by 14.5MP to give 32.5 PMP on a 5DsR? Seems your formula should be MPup/4 rather than MPup/2.
First of all I really like your calm and technical style :) If we are talking about sharpness, AA filter is the most important thing in my opinion. You will never get sharper images with a regular A7 with a Zeiss Otus than an A7r with a $250 Rokinon 85mm 1.8. It is a fact that people just ignore. A Canon 5D M3 can never be as sharp as a micro four thirds camera with no AA filter in it. I have tested it so many times. I don't understand people buying a $2000 lens for their A7II for a sharper image. It will never be as sharp as an original A7R if you avoid the shutter shock and use it at reasonable shutter speeds. This whole argument is pointless. Yes, better lenses usually create sharper images but seriously nothing can effect micro sharpness more than AA filters. Take care!
Tony, how well do the legacy lenses (70's Nikon, Canon, 50mm f1.8 and such) compare in pmp? Has big name glass gone up or down in true sharpness? (When sharpness used to be measured by how many line pairs per mm the film could hold, IIRC with 200 lp/mm used to be pretty good, but the big name glass never limited that?)
I remember as a kid mydad had banner canvases for products he handled. I was surprised to see groups of color dots with alot of white space between elements. Opened one up a little and went to the other side of the warehouse to see the image.
What I find the most challenging in resolution compartment is photo wallpapers. They tend to occupy whole wall in a room and you come close to them often. But theres always image stacking and stitching techniques...
Great video. As a self taught hobbyist photographer I never needed to print any of my photos, so I really don't have much knowledge when it comes to printing in general. As of lately though, I've been getting more interested in this topic, and I'm thinking about buying a printer (solely for fun, not an investment of any kind). Great timing hahaha
Also, fractal upsampling (e.g. Iterated Fractal Transform or IFT) enables you to dramatically blow past any such limits. Although the maximum upsizing depends on the source material type (e.g. architectural, nature, portraits, etc), if you use that together with sharpening appropriate to the material and paper material, coating and viewing distance, common low-Nyquist noise ratios in each dimension are 2, 4 and 8 times. In other words, 4. 16 and 64-times the pixels for even low viewing distances. Photoshop provides this as part of their upsampling options, but there are many cheap, free and other options. Years ago, in the 90’s, when Michael Barnsley first created the early IFT fractal encoders, it was used to create the image of a meteor crashed into a building for one of the end of the earth movies of that era. It was made the size of the side of a building from a 2 MB digital file. It looked so photorealistic from the street that they took it down the first week because it was creating traffic jams and crashes.
Hey Tony, did your print results confirm the dxomark measurement? That is to say, did the A7s/lens combo have less perceptual megapixels than you needed to hit 300dpi at that print size?
What about scaling images before print? If I have an 18MP raw file and process it in lightroom, and apply output sharpening for print, should I scale it down to fit my 8x10 at 300dpi, or send it to the printer at the full 18MP resolution?
I have question. As far as i've read, upgrading to very high resolution includes a lot more carefulness taking handhold pictures (Street Photography and so on).... is it true?
Tony you are an outstanding instructor & know infinitely more than I about photography but a question: would Modulation Transfer Function (MTF) not provide a more specific metric of lens image quality separate from the sensor than P-Mpix?
Hey Tony. Correct me if im wrong but, one does not need an optically perfect lens, just one which out resolves the sensor on the settings you are using
+Paul Redman Lenses can't really "outresolve" the sensor; they are always something less than perfect, and anything less than perfect will degrade the detail captured by the sensor from its maximum megapixel count.
Love the vid but this left me with more questions. What F stop did they use to test sharpness? Take the canon 50mm 1.2, how sharp is that lens on the focus point, obviously the rest will be blurred at F1.2. ISO? higher iso i feel like the image doesn't look as sharp but i think thats mainly noise. just my thoughts Thanks for the Break down of DPI and distance. this will help with prints i make from now on.
+Adrian Larios There's some info at sdp.io/dxo, if that's what your'e curious about. But don't get hung up on the specific measurements; lots of other factors impact sharpness.
I love your videos... Would not the surface on which the image is printed be included in the factors of megapixels needed for optimum results? After all, a Canvas Print will never show the detail that can be shown on a glossy surface or even more so on a metallic surface.
A bit rough and ready. To more accurately calculate DPI from MPixels (total number of pixels over the sensor area) its A = MPixels/300^2 (if you want to work to 300DPI, if not change the 300 to whatever DPI you want to work with). Then you have to decide what aspect ratio you're sensor is, assuming 1.4 (1.4 x height = width) the height of the photograph will be sqrt(A/1.4) and the width of the photo will be 1.4 x height. For those wanting metric units multiply the calculated height and width by 2.54 to give the dimensions in cm.
On DXOMark I compared my 18x135 stm kit lens (T6i) vs the 24x105 Sigma that you're always suggesting and it says 8 vs 10 Pmpix. With overall score of 13vs17. Poor results according to the website. Am I missing something here? I'll be glad to hear a response from anyone that actually knows something.
So, if I print an 8x10, 300dpi photo taken with my 9 P-Mpix camera/lens combination, it will have the same clarity as an 8x10, 300dpi photo taken with a 20 P-Mpix camera/lens combination?
If I print the same image taken with the same camera and lens at 300dpi and 600dpi would there be any difference ( 11x14"). Also I assume uploading to the web you have no real benefit going above 72dpi? Thanks!
How many perceptual megapixels are there on auto zoom // non removable lens or entry level auto zoom? Can they still advertise it that way when it would be much lower and not changeable?
There you go again +Tony Northrup bringing your science mumbo jumbo into photography :-) . Very we ll done video you took the highly technical info and broke it down in terms that don't take a PhD in Mathematics to understand. Keep up the great work!
I've noticed that canon's 5dsr with 50 mp has a lower diffraction limited aperture than the 6d mark 2 with 26mp. The sensor area is the same on both ff cameras.
Correct me if I'm wrong, but isn't the dpi number measured diagonally? In that case you would have to divide the 300dpi by a geometrical factor to get the horizontal pixel density. 300/sqrt(2) = 200
This was most helpful Tony, thank ya. Just a simple question, is it generally safe to trust DXOMARK, when making buying decision for photographic equipment?
Something has been bugging me about DXOMark, specifically related to the 7D Mark II. If I compare the 70D with the 7D Mark II on DXOMark, ALL the lenses consistently measure lower P-Mpix on the 7D Mark II than on the 70D. They both have the same sensor size and megapixels, so why would DXOMark consistently measure less detail on the 7D Mark II? Example: Sigma 18-35mm F1.8 DC HSM A on Canon EOS 70D = 15 P-Mpix vs 13 P-Mpix on Canon EOS 7D Mark II
Ok I know this is a old video but is there like a website or something that might have a large database of different lenses and there approximate quality ?
+Tony Northrup You forgot two of the most important factors. 1. The Bayer sensor. It only achieves 50-55% resolution even without a low-pass filter. This creates a downsampling factor of 2! If you aren't shooting with a monochrome or multi-layer sensor you have to multiply the resolutions of the dpi table by two. 2. Transformation of contrast levels. This essentially explains what you are seeing with the higher-res cameras on DXO. They don't measure pure resolution but sharpness (resolution x micro-contrast). With the higher resolution sensor you are sensing the same lens contrasts at a higher frequency. Once you downscale to a smaller image scale you are transforming those captured contrasts. Low frequency contrasts of the lens become high frequency contrasts of the final output. Those are then perceived as sharpness. That's why the pixel crave does never really end. Definetly way after the two times downsampling factor. That's why one even can see a sharpness difference between an 80 and 60MP digital back on a small web presentation...
What about using a Metabones focal reducer to mount a high resolution full frame lens on an APS Sensorcamera such as a Sony 6500. How many megapixel sharpness in a full frame lens eould you need to maximize the sensors 24 MP resolution?
This is the best vid I’ve watched. I shoot in crop mode all the time with my a7iii because I think 10mp is enough. I’m kind stop doing it now, the end of the day, I need to print 8x10, I don’t think I have enough p-mpix for a sharp print
📚 Buy Our Books on Amazon! 📚
📕Stunning Digital Photography: help.tc/s
📘Lightroom 6 Book: help.tc/l
📙Photoshop Book: help.tc/p
📗Buying Guide: help.tc/b
One thing that never gets mentioned is that perceived megapixels and file size can be dramatically increased in software such as Photoshop and Onone.
. I took a 16 megapixel image from 25 megabytes to 175 megabytes and made an acceptable 40in by 60in print.
You really are the most reasonable and honest photography teacher out there. I like your straight forward style. Much appreciated.
Thanks!
totally agree!
I love the fact he provided everything with real facts
+1
You are the best at covering this stuff. You cover all the bases, and you're no dismissive about any aspects of the subject matter. I don't see anyone else giving such comprehensive explanations that bring the technical side into such a user-friendly explanation. Nice work.
+Terence Kearns Thanks!
Love your lessons. I used to teach photography classes. So far your are the only one that consider all aspects and variables in a subject matter. Your technical knowledge WOW.. thanks.
Love these videos where you really break it down to the technical info. Thanks for the video as always!
Great video, I have watched so many video of yours and by many other good photographers in youtube but i have to admit but you are the one with the most analysis and i like this, keep going
Another stunning video that helps photographers, and is presented in a methodical way and with great substance, thanks again
I love how technical and specific are your lessons. Thank you for so informative and useful classes in youtube.
Tony, you absolutely hit the nail on the head when you said, "300 dpi is the ideal image density for looking at something up close". The key words here are, "for looking at something up close".
In my experience, as a professional photographer who has been in commercial photography for over 40 years, the whole dpi or ppi idea is ludicrous. If the 300 dpi theory is correct, then how on earth did they produce 10 x 8 ft billboard posters from my 35mm film transparency images, back in the day?
The answer is (surprisingly) that the bigger the image you're printing, then the less dpi you actually need in order to produce what's perceived to be a 'pin sharp' print.
Put simply, that's because you view a billboard from about 30ft away. A 20x16" print from a few yards away. And a 5x4" print with your nose pressed against the paper. Only the minuscule 5x4 needs that 300 dpi.
If people really want to see their images at their best, then print big and don't be too concerned about the image resolution of your camera's sensor. The human eye won't notice. Indeed, as artists have known for centuries, we're not looking for exact mathematical precision ; the best figurative images are impressionistic. What appears to be a simple crude mark or brushstroke, when viewed from too close, is actually an elegantly expressive rendering when viewed from an appropriate distance.
Only "rivet counter" self-appointed photography judges will stick their nose into a 20x16" print -- probably with a magnifying loup -- to base their judgement on what they like to think of as "scientific" judgements. And who cares what such people think.
A truly great picture is best viewed in a large print size, where the viewer will stand well back, in order to appreciate the subject and its artistic rendering.
Keep up the good work, Tony. You are a brave man to attempt to educate people that great photography is about art, rather than anything that can be broken down into pseudo-science. You'll take a lot of criticism from the rivet counting zealots, but your efforts to educate the wider community are very much appreciated.
Rick Bear
(Professional, award-winning, photographer.)
As a non-US viewer I would appreciate that you also add a column with metric equivalents. Perhaps there aren't that many of us, but from the comments in previous posts I get the impression that there are at least some. Well, just a suggestion for what is already splendid.
+Richard Rönnbäck
1 inch = 25.4 mm :)
+ObelixCMM And for rough estimations I use: 4 inches = 10 cm, 3 feet = 1 meter
+Dmitry Rudoy yeah & with your standard of accuracy & precision in establishing definitions, your mother is a virgin :-)
+Richard Rönnbäck Actually the US and UK are the last remaining significant markets using legacy inches. They are actually in the minority. Also US and UK somehow magically know about lens MILLIMETRES - LOL
Actually I come from the UK and was brought on Imperial feet and inches etc and then we had metrication forced on us! Rant over.. :-)
As usual, an eye opening video from Tony. Thank you. I have great respect for you & you have taught me a lot
+Omar Omari Glad to help!
As usual Tony, a very sensible, level-headed point of view. Keep up the good work!
+David Greenwell Thanks!
It also used to suck to have to crop an image only to end up with 1-2 mp worth leftover back when I was shooting with much lower megapixel cameras.
The reason you need megapixels is so you can crop and process your images. When you straighten, add local contrast (clarity), sharpen, noise reduce, etc. you're doing arithmetic on neighbouring pixel values. That arithmetic will blur features that are close to one pixel on the sensor. But if you have a much higher megapixel image to work with so that the features you're hoping to see are five or ten pixels across, the processing algorithms will work much better and the end result will look sharper even if you export a 2 Megapixel JPEG at the end.
+Pavlos Papageorgiou Extremely well put and very much worth saying. I kept expecting Tony to make this point clearly but I suppose he figured it was a little too complex. He did allude to the idea when he showed the two 300dpi prints that looked different. He was right to emphasize the roll played by the lens but processing deserves more mention than it got.
This is too good of a reply to come from youtube comments. You gave a clear succinct answer that was on topic without being rude to anybody. Are you sure you understand how youtube comments work. ;) Thank you for being a sane voice on here.
Anonymity is a disinhibitor, like alcohol, so a lot of commenters are like bad drunks; feeling empowered but impotent and over-stimulated.
They type like they’re drunk too. All those exclamation marks! And they’re never able to spell “they’re” correctly. I misspelled “role” in my original reply.
Very good answer. Yup, if you really haven't bothered to expend your efforts in learning to develop your photographic skills, then I guess you can never have enough gigapixels upon which to base your faint hopes of discovering a recoverable image in post processing.
😜, the Bear
Having been using my 16MP for 2 years. Just started looking for a new one, this saved me from the MP myth trap. Bad composed photos with more dots are still bad. Thanks Tony!
Tony always does a great job on the technical descriptions.
"Sharpness is not everything" somewhere Jared Polin is crying ;)
Jared is a quack anyway, let him cry all he wants xD
+Appelcaster JP is the MTV of YT... The Northrups are like old-school Discovery Channel.
+martinaee You mean when discovery was actually educational?
+driverv86 Hence: "old school" :)
martinaee I miss that version.
I always learn so much when watch your videos. Thanks.
Jon
Thank you for being smarter then me. Things I would never think would even interest me or have any concern with my photography and your way of explanation just has me hooked and understanding. Thank you again for all that you do!
Thank you for this comprehensive, well-argued and informative video Tony. I especially liked that, by way of reassurance for those who might have been made insecure by the earlier parts of your presentation, you put things into perspective in your conclusion: namely that in the everyday world, viewing distance is a vital consideration. I'd only add that the same point applies to 4k TV's versus 1080p TV's (not 4k monitors, which one views at a far closer distance): the farther away from the screen, the less perceived benefit you will derive. Thanks again.
I forgot to add that I really enjoy videos that explain tech aspects. I'm just learning photography and you and Chelsea are my favorite resource. Thank You!
Very interesting. Never heard of the p-mp rating for glass. So what it looks like, is with 36mp and higher sensors we're starting to reach the point of diminishing returns as far as sharpness is concerned - at least with current lens technology?
UA-camr: "I watched the first 10 seconds of your video, and here is why you are wrong about all the things I'm guessing you said!"
Yeah! That was clear to me when I upgraded fromm my old tamron 70-200mm 2.8 non VR (9mpix on 70D) to its new Tamron 70-200mm 2.8 VR (16mpix, 70d). Its AMAZING the difference on the results, Im stunned and in love with this lens. I think 24-30 mpix is enough for everybody, 20 is perfect for sports. I've noticed that many lenses doesnt come close to 35 pmpix on 5Ds, only zeiss, thats something to think about.
Which yields better low-light image quality? A high megapixel image down sampled or a sensor with large pixels.
This video is great, especially the calculations for getting between print dimensions and pixels, and perceptual megapixels for the lens camera combination. Brought together things I had noticed and suspected, now I don't have to blindly rely on the table I've been using for print size. THANK YOU! Just started stalking birds in my back yard with my new Zeiss batis 85mm on my Sony a7riii with magnified peak focusing, phenomenal wing detail.
Those are some funky ass colors, i think the saturation slider is a bit off haha
+Bror Svensson Media It's actually the blending mode. They're just messing around with the curves to create a cross-process effect resulting in terrible greenish and oversaturated skin tones. A blending mode of soft light, color or hue with a 10% opacity would do the trick.
+Bror Svensson Media I think its the Wall paper, it seems to be difficult to light or balance, they have tried to warm up the picture recently, in fact Tony needs to decorate, or he needs to dye his hair violet or something as its camouflaged into the grey (ish) background. ;-)
He is using a polarization filter as well, apparently to cut out the update effect of the TV.
One thing I notice is that it kind of reach the limit of lens resolution when switching to 50mp 5ds 5dsr as the p-mpx improvement is so minor, even using the Otus 85, the very lengendary lens in the world. 36 to 40 mp seems to be a sweet point as the extra mp improvement doesn't convert to more actual information you get.
13:27 why do you use the number 6870? Where does this come from?
another terrific offering, Tony....thank you
+Neil Hersh Thanks, Neil!
You are kind of the "Fred Picker" (teacher of film photography) of the digital world.
Good work.
Also, it depends on how you print it. I got one of my pictures, which I have taken with a Canon PowerShot S120 (12MP, really not that good), printed out at 23"*31" on a canvas, and it really looks sharp
I did an extreme extensive test on this Tony. I am OCD so I did every thing possible with this. I put the 5DSR up against the A7R, against the Canon 6D. Even with pretty extreme crops, it wasn't as night and day as I thought it was going to be. I have a pretty good size print and although the A7R has more resolution, it didn't seem to have much advantage over the 6D. The 5DSR has a lot of details, they are noticeable, especially when cropping. However, even then, it didn't make as big of a deal as I thought it would, unless you're doing a really small crop out of a picture. I could notice that it had more detail, however, the picture on the 6D, which was shot with a really sharp lens, had a ton of detail too and the extra detail I don't think a non photographer would have noticed. BUT with computer monitors, it gets to be even less of an issue. Most people don't have 4k yet, that's only 8 megapixels, we don't have any computer monitor that can see all the resolution of a Canon 6D. Of course we see the difference in zooming in and cropping, but unless it's an extreme crop, it just wasn't as big of a deal as I had made it out to be in my head. The reason I did this test is because I'm a landscaper, I was shooting with the A7R but missed my Canon 5D Mark iii, ergonomics, colors, etc. I thought about going to the 5DSR, so I tested it, but it just didn't offer as much as I thought it would over the Canon 6D. Which is what I have now. I didn't get the 5D Mark iii again because there was nothing it offered me as a landscaper that the 6D didn't have. Now, with that said, with ISO going down, and other things that come with adding more megapixels, I am hoping the next 5D is 28 megapixels. I really feel 28-36 megapixels is the sweet spot right now. The 24-70 2.8 ii and 135mm F/2 from Canon are extremely sharp, they provided a ton of detail with no softness. I used the Zeiss 55mm 1.8 on the A7R as well as the Zeiss 16-35 F/4.
I love how you answered it in the first 5 seconds :D
I think sharpness is somewhat important for me being a filmmaker, I don’t have money for Ziess or Cooke lenses but clarity is still important to my art. You make a lot of sense Tony..... Love this channel
A great educational video, Tony, thanks a lot for the time spent on this. Very clear and extremely informative.
So if I can get 8 perceived megapixels is that just good enough for 8x10s??
I loved the way you explained. you are a treasure.
will you please make a video about pixel size or Pixel pitch
"just want to let you know". -Tony ruining everyone's dreamy photography misconceptions with facts...Hahaha! I love it. Tony is too smart.
I use PPI when talking about digital files, and DPI when talking about actual print density. This way there is no confusion between file export setting and printer settings. I have seen files saved at 600 PPI becouse they tried to match printer's 600 DPI
Hi Sir Tony😄 I just want to know if the Nikon D5300 a good camera even though it have a 24.1 megapixels.
Is it a good camera??? Thank you😊
Yes, it's a great camera.
Very informative video. Thank you for all the useful information you have posted!
Any idea why the DXO p-pmix scores are consistently higher for the Canon T5I than the same lenses on the canon 7D mark II? I was wanting to make the upgrade but I'm confused that there is such a consistent lower scores for the sharpness of the 7D mark II. Am I misinterpreting something?
Today I bought your book here in Brazil! It will arrive in a few days. I'm excited! =D
Tony I have a question and you seem like the right guy to answer it. I have a pentax k-3ii with a 100mm macro lens and a 1.4x teleconverter. The pentax has a 1.5x crop factor and the macro lens shoots 1:1 magnification. Along with the teleconverter, what is my total magnification number at the closest focusing distance on the macro lens with the crop factor of the camera? How do I work out that formula? And what is my overall equivalent crop factor with the camera and the teleconverter?
I really appreciate the scope and range of the videos you guys put out - techy stuff like this, the new informative podcast and the creative focused 'live' shows really provide fantastic educational content - this has caused me to purchase your books which is I guess is the purpose :)
Thanks, Dave!
In a video you said that using a full frame lens on a micro four thirds camera produces less sharp images because the pixel density of micro four thirds is higher than full frame cameras. Am I missing something here?
The end was so motivating, thank you guys
REALLY DETAILED VIDEO. LOOKING FORWARD TO SEE SOME SMARTPHONE CAMERA REVIEW FROM THE EXPERT. THANKING YOU IN ADVANCE
13:27 Where is that 6870 coming from?
Great video as usual! Just one thing, aren't recent videos affected by a blue tint? Tony's hair looks weird with this color temperature
DxO image testing is based on RAW, so images can still be sharpened in post-processing.
Also depending on the type of photograph total P-Mpix across the frame might be completely irrelevant. A portrait lens for example might be extremely sharp in the central portion of the frame but unsharp in the extreme corners. You might likely place your subject toward the centre of frame (maybe slightly off-center to rule of thirds nodes), so you may not end up needing the corner pixels for capturing fine detail. Also, for portraits, these corner pixels might very well be representing out-of-focus blur instead of fine detail.
Merci pour les super tutoriels que vous nous offrez chaque semaine! Thanks your are the greats person!
Great video! This was very educational. Thanks, Tony.
Hi Tony, do you mind making a video explaining the difference between depth of field and lens compression. i think its a really great topic to cover since a lot of people do not know the difference between those two
Hi Tony. Wasn't your formula predicting the 18PMP scoring Sigma 24-105mm on a 5D3 would rise by 14.5MP to give 32.5 PMP on a 5DsR? Seems your formula should be MPup/4 rather than MPup/2.
First of all I really like your calm and technical style :)
If we are talking about sharpness, AA filter is the most important thing in my opinion. You will never get sharper images with a regular A7 with a Zeiss Otus than an A7r with a $250 Rokinon 85mm 1.8. It is a fact that people just ignore.
A Canon 5D M3 can never be as sharp as a micro four thirds camera with no AA filter in it. I have tested it so many times.
I don't understand people buying a $2000 lens for their A7II for a sharper image. It will never be as sharp as an original A7R if you avoid the shutter shock and use it at reasonable shutter speeds.
This whole argument is pointless. Yes, better lenses usually create sharper images but seriously nothing can effect micro sharpness more than AA filters.
Take care!
Awesome video Tony, thanks for so much valuable information.
Tony, how well do the legacy lenses (70's Nikon, Canon, 50mm f1.8 and such) compare in pmp? Has big name glass gone up or down in true sharpness? (When sharpness used to be measured by how many line pairs per mm the film could hold, IIRC with 200 lp/mm used to be pretty good, but the big name glass never limited that?)
Thanks for the insightful and well thought out video Tony! I think it would be useful to do a similar one for bit depth.
I remember as a kid mydad had banner canvases for products he handled. I was surprised to see groups of color dots with alot of white space between elements. Opened one up a little and went to the other side of the warehouse to see the image.
What I find the most challenging in resolution compartment is photo wallpapers. They tend to occupy whole wall in a room and you come close to them often. But theres always image stacking and stitching techniques...
I would also add that more MP lets your crop the photo more, but great video!
Great video. As a self taught hobbyist photographer I never needed to print any of my photos, so I really don't have much knowledge when it comes to printing in general. As of lately though, I've been getting more interested in this topic, and I'm thinking about buying a printer (solely for fun, not an investment of any kind). Great timing hahaha
Also, fractal upsampling (e.g. Iterated Fractal Transform or IFT) enables you to dramatically blow past any such limits. Although the maximum upsizing depends on the source material type (e.g. architectural, nature, portraits, etc), if you use that together with sharpening appropriate to the material and paper material, coating and viewing distance, common low-Nyquist noise ratios in each dimension are 2, 4 and 8 times. In other words, 4. 16 and 64-times the pixels for even low viewing distances. Photoshop provides this as part of their upsampling options, but there are many cheap, free and other options. Years ago, in the 90’s, when Michael Barnsley first created the early IFT fractal encoders, it was used to create the image of a meteor crashed into a building for one of the end of the earth movies of that era. It was made the size of the side of a building from a 2 MB digital file. It looked so photorealistic from the street that they took it down the first week because it was creating traffic jams and crashes.
Can't wait for the 1DX II preview, Tony! When do you make it?
Hey Tony, did your print results confirm the dxomark measurement? That is to say, did the A7s/lens combo have less perceptual megapixels than you needed to hit 300dpi at that print size?
What about scaling images before print? If I have an 18MP raw file and process it in lightroom, and apply output sharpening for print, should I scale it down to fit my 8x10 at 300dpi, or send it to the printer at the full 18MP resolution?
I have question. As far as i've read, upgrading to very high resolution includes a lot more carefulness taking handhold pictures (Street Photography and so on).... is it true?
Tony you are an outstanding instructor & know infinitely more than I about photography but a question: would Modulation Transfer Function (MTF) not provide a more specific metric of lens image quality separate from the sensor than P-Mpix?
How do you feel about fractally enlarging photos before printing with something like On1's perfect resize?
Hey Tony. Correct me if im wrong but, one does not need an optically perfect lens, just one which out resolves the sensor on the settings you are using
+Paul Redman Lenses can't really "outresolve" the sensor; they are always something less than perfect, and anything less than perfect will degrade the detail captured by the sensor from its maximum megapixel count.
Love the vid but this left me with more questions.
What F stop did they use to test sharpness?
Take the canon 50mm 1.2, how sharp is that lens on the focus point, obviously the rest will be blurred at F1.2.
ISO? higher iso i feel like the image doesn't look as sharp but i think thats mainly noise.
just my thoughts
Thanks for the Break down of DPI and distance. this will help with prints i make from now on.
+Adrian Larios There's some info at sdp.io/dxo, if that's what your'e curious about. But don't get hung up on the specific measurements; lots of other factors impact sharpness.
I love your videos... Would not the surface on which the image is printed be included in the factors of megapixels needed for optimum results? After all, a Canvas Print will never show the detail that can be shown on a glossy surface or even more so on a metallic surface.
A bit rough and ready. To more accurately calculate DPI from MPixels (total number of pixels over the sensor area) its
A = MPixels/300^2 (if you want to work to 300DPI, if not change the 300 to whatever DPI you want to work with). Then you have to decide what aspect ratio you're sensor is, assuming 1.4 (1.4 x height = width) the height of the photograph will be sqrt(A/1.4) and the width of the photo will be 1.4 x height. For those wanting metric units multiply the calculated height and width by 2.54 to give the dimensions in cm.
On DXOMark I compared my 18x135 stm kit lens (T6i) vs the 24x105 Sigma that you're always suggesting and it says 8 vs 10 Pmpix. With overall score of 13vs17. Poor results according to the website.
Am I missing something here? I'll be glad to hear a response from anyone that actually knows something.
Watch my video, "should you use full frame lenses on crop bodies"
Thanks Tony!
Your book Stunning Digital Photography contains this kind of information? It is really useful information
Wow Tony you made me really understand megapixels concept. I get it! Thanks Tony
Fantastic. Thank you Maybe half my heated discussions about gear would be stopped dead if people watched this clip.
So, if I print an 8x10, 300dpi photo taken with my 9 P-Mpix camera/lens combination, it will have the same clarity as an 8x10, 300dpi photo taken with a 20 P-Mpix camera/lens combination?
If I print the same image taken with the same camera and lens at 300dpi and 600dpi would there be any difference ( 11x14").
Also I assume uploading to the web you have no real benefit going above 72dpi?
Thanks!
Technically fantastic video. Thanks!
Another great explanation. Thanks.
Can you please link the video where you test the Nikon 18-55mm kit lens you were refering to?
How many perceptual megapixels are there on auto zoom // non removable lens or entry level auto zoom? Can they still advertise it that way when it would be much lower and not changeable?
There you go again +Tony Northrup bringing your science mumbo jumbo into photography :-) . Very we ll done video you took the highly technical info and broke it down in terms that don't take a PhD in Mathematics to understand. Keep up the great work!
I've noticed that canon's 5dsr with 50 mp has a lower diffraction limited aperture than the 6d mark 2 with 26mp. The sensor area is the same on both ff cameras.
Correct me if I'm wrong, but isn't the dpi number measured diagonally? In that case you would have to divide the 300dpi by a geometrical factor to get the horizontal pixel density. 300/sqrt(2) = 200
+ACPushkin I've only seen it measured vertically or horizontally.
This was most helpful Tony, thank ya. Just a simple question, is it generally safe to trust DXOMARK, when making buying decision for photographic equipment?
+Rick Mentore
As long as you know what the numbers actually mean, yes. Here, watch this video: /watch?v=MRzONYmjH3M
Why is the image so blue? Did you intentionally changed the WB to blue side?
How can I determine the perceived megapixel if I use a manual focus lens from an old film cameras?
Something has been bugging me about DXOMark, specifically related to the 7D Mark II. If I compare the 70D with the 7D Mark II on DXOMark, ALL the lenses consistently measure lower P-Mpix on the 7D Mark II than on the 70D. They both have the same sensor size and megapixels, so why would DXOMark consistently measure less detail on the 7D Mark II?
Example: Sigma 18-35mm F1.8 DC HSM A on Canon EOS 70D = 15 P-Mpix vs 13 P-Mpix on Canon EOS 7D Mark II
Which camera/lens did you use for each of the two prints you made?
Ok I know this is a old video but is there like a website or something that might have a large database of different lenses and there approximate quality ?
+Tony Northrup
You forgot two of the most important factors.
1. The Bayer sensor. It only achieves 50-55% resolution even without a low-pass filter. This creates a downsampling factor of 2!
If you aren't shooting with a monochrome or multi-layer sensor you have to multiply the resolutions of the dpi table by two.
2. Transformation of contrast levels. This essentially explains what you are seeing with the higher-res cameras on DXO. They don't measure pure resolution but sharpness (resolution x micro-contrast). With the higher resolution sensor you are sensing the same lens contrasts at a higher frequency. Once you downscale to a smaller image scale you are transforming those captured contrasts. Low frequency contrasts of the lens become high frequency contrasts of the final output. Those are then perceived as sharpness. That's why the pixel crave does never really end. Definetly way after the two times downsampling factor.
That's why one even can see a sharpness difference between an 80 and 60MP digital back on a small web presentation...
What about using a Metabones focal reducer to mount a high resolution full frame lens on an APS Sensorcamera such as a Sony 6500. How many megapixel sharpness in a full frame lens eould you need to maximize the sensors 24 MP resolution?
Hi Tony. Tx for the video... will this apply to video too or are you talking (printed) photos?
+TheJWC56 Yes, though video is much lower resolution, and sharpness tends to be less scrutinized since it's always moving.
This is the best vid I’ve watched. I shoot in crop mode all the time with my a7iii because I think 10mp is enough. I’m kind stop doing it now, the end of the day, I need to print 8x10, I don’t think I have enough p-mpix for a sharp print
Definitely not; you definitely want to use all 24 MP. Plus, 8x10 prints are really small.
Any plans to break down why some lenses are good for stills versus video in terms of megapixels and image quality?
Great video.
Very informative