Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Megapixels and limits of glass
#11
[quote name='Pinhole' timestamp='1287611520' post='3711']

Very informative post, thanks, Wim!



All discussions of resolution aside, I feel that with a Canon 5D MK II I am at a level that is perfectly adequate (prints above A2 are very rarely called for) - but what I really want is more dynamic range. Maybe I'm wrong here, but I feel that the industry is selling their new technology on the strength of megapixels because it's easier than improving dynamic range.

[/quote]

Thank you, Pinhole, it's a pleasure.



To a degree you're right. There are also several aspects to this. Sensor DR really is already better than film, at least in RAW files, namely approximately 10 stops, and if you push it you can get to about 12.



In order to get more out of a sensor, we actually need better AD and related circuitry. Current sensors, especially FF ones, are already capable of approximately 15-16 stops of DR, but making a good AD converter at a reasonable price that can handle this, especially at the noise levels required, is still a long way out of scope. I have been told there actually are cameras out there than can do approximately 16 real stops of DR, but those are rather expensive, and only made to order. Eventually, the technology will permeate down to general use, just like we see and have seen with, e.g., car improvements like disc brakes. I do think this still is some time off however.



A second thing to note is that in print, due to th elimited DR possible as paper can only be so reflective and inks only so black, you lose a lot of this DR, and in order to make prints that look right, you somehow need to compress the DR into a smaller space, in a way similar to what we used the Zone System in the past for. And even on screen this is true. Thsi does fool th ehuman eye and brain to think we actually see what was there in the original scene BTW.



I would therefore suggest that if you feel you are lacking in this area, to pick up some reading on the Zone System. There even are books on how to apply this with digital, and take it from there. I honestly think that that is the easiest and most satisfying solution, because it means we can also display our images to their fullest glory both in print and on the screen. Some very light, overexposed areas, and some dark, underexposed areas, never is a problem in a picture, because even the human eye struggles with large dynamic ranges. Just that the human eye dynamically adapts itself to the environmental light conditions without us noticing this adaptation a lot (unless we consciously do observe this, or step out from th every dark to the very bright or vice versa), and quite quickly too, but it is not how we really physically see the environment. Our brain does process and adjust things rapidly enough most of the time that we don't notice this.



BTW, in PP, even adjusting contrast, brightness and saturation often does the trick, either to expand or to compress the dynamic range the camera is capable of registering and making it fit the enviroment for display. For best results provided of course that exposure was done to fit with the end result. It honestly is amazing what one can get out of digital compared to film. From my experience, film actually doesn't come close, regardless of what some people say. The main difference between the two media, other than differences in DR and the actual curve, is the fact that film favours the top end of the exposure curve for detail, due to its transparant nature vs the blown out nature of digital at the high DR end, and digital the bottom end of the curve, due to the fact it still manages to retain detail at the lower end, where film just gets totally opaque.



Also note that with higher DR cameras, with fully utilized DR, the pictures will have a distinctive HDR look, as the viewing media can't handle this DR, and neither can the human eye, thus requiring DR compression to fit. Although the human eye is supposed to be able to handle a contrast range of about 1:1,000,000,000 (about 24 stops), this is not the actual DR range as we can see in a single instant, without any processing by our own hardware, both the iris (aperture) and sensitivity. Do note that this means we can't see this range in one go either, because we would have to simultaneously have our irises wide open and completely closed, and adjusted to high and low sensitivity. In reality we can handle about 10-14 stops in a single go (estimates), without processing or adjustments. In short, for this instantaneous type of looking, a digital camera is already there, considering digital can handle currently up to 13.5 stops of DR in some models (and 10 to 12 in the better ones anyway). And directly compared to th eye, this still means a camera can handle, for handheld shutter speeds of 1/60s and up, let's say, a similar "total DR" of about 23 to almost 25 stops (F/1.4 to F/22, 1/60s to 1/8000s, iso 100 to 800 without major loss of DR, assuming 10-13.5 stops of DR, exposed in the middle, i.e., 8 + 7 + 3 + 5, max. 7 = 23 to max. 25) , which is close if not the same, just not almost instantaneous or automatic. Use a tripod, and longer shutter speeds, and you get even more. Not DR as you want it, I know, but that is how DR in an eye works too.



In short, if you do want more DR, for example for landscapes in the bright sun with, e.g., a path leading into a dark forest in the shadows, where you'd want to see details in both simultaneously, this isn't going to be possible soon, but then, the human eye can't do both simultaneously either. The human eye works like film really, e.g., it will correct for the highest DR values first, so that we can still see something in the brightest parts, while the darkest parts will completely clip, i.e., will show no detail at all. The challenge is to get digital to work in a similar way, where it fundamentally does the opposite, namely clip the highlights first.



Hence exposing to the right with digital, i.e, the highest DR where important stuff doesn't get clipped yet. That is IOW really an emulation of what our eyes and brain do <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='B)' />. Of course, when a human in that case concentrates on the dark forest, and one exposes the camera for the same, it'll look very similar.



Naturally, one could always do HDR, but that will generally look like (overprocessed) HDR too in most cases, even when done from a single RAW (and the latter is really a limit of the viewing medium). So, do we really need more DR? Or do we need/want to be able to record more intermediate tonal values? I really wonder <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='B)' />. However, in both cases we need better and faster AD converters and faster and better image processors in our cameras, and probably also better and faster computers, larger hard disks, new standard file formats, better monitors, better paper DR, etc. <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='Big Grin' />.



Kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....
#12
I agree with Wim in that the call for higher DR is usually not based on anything. High dynamic range results in bland images without contrast, which are NOT the images that would do much for us... just lower the contrast to the minimum in your RAW converter and you will get the idea.



Usually the call for higher DR is just repeating what has been seen on internet forums...



The only advantage of film over digital really is the way sensors collect and store the data... In the lower regions the doubling of the light does not have very many steps in between, making the dark data really rather coarse. But that can not be solved with higher DR.



What usually gets measured as DR (with the likes of DXOmark for instance) is more the noise... but a lower noise in shadows does not convey any different images to our eyes. There is a bad way of improving DR there (noise reduction, which does have its impact usually on IQ, like Nikon and Sony apply), and a good way (improving the AD conversion).



Improving the AD conversion does not have much to do with the photon collection itself, though.
#13
Actually, BC, my comments are not "repeated from an internet forum", but from years of experience using film (which I still shoot). I don't know about the science and the figures, but subjectively I see more shadow and highlight details in actual prints using low ISO film (either 35mm or 6x7) than my DSLR is able to reproduce.
#14
[quote name='Pinhole' timestamp='1287660810' post='3716']

Actually, BC, my comments are not "repeated from an internet forum", but from years of experience using film (which I still shoot). I don't know about the science and the figures, but subjectively I see more shadow and highlight details in actual prints using low ISO film (either 35mm or 6x7) than my DSLR is able to reproduce.

[/quote]

What you see has little to do with dynamic range, but rather with tonal contrast/curves, i am quite sure. Film has a lower DR than digital.



Your screen (and your prints) are not even able to show the dynamic range of your photos... they are dynamic range limited. Increasing the dynamic range of sensors will not help much.. it is in how to compress the dynamic range available from the sensor data into the range that you are able to print which makes all the difference.



Where film has an advantage is in how it rolls off the registration of bright stuff... it simply does not register it all. Film sort of registers light already in an S-curve (not sure if that is the correct way of saying it, if not I am sure Wim will correct me).



But in short... it is not the dynamic range of sensors that is lacking, it is the linear nature of the light capturing that makes the character of the tonal contrast different (and less appealing to you). Understanding this will bring you a long way in post processing images to approach the character of film.
#15
[quote name='Pinhole' timestamp='1287660810' post='3716']

Actually, BC, my comments are not "repeated from an internet forum", but from years of experience using film (which I still shoot). I don't know about the science and the figures, but subjectively I see more shadow and highlight details in actual prints using low ISO film (either 35mm or 6x7) than my DSLR is able to reproduce.

[/quote]

I dont want to shoot foul, but really, looking at pints from my slides (yes, properly exposed slides, printed by a professional lab for an exhibition) I see fundamentally less dynamic range in them compared to professional prints of my 50d and 5d Mark II files. The Prints from the slides show however a more pleasing rendition of the brightest highlights. Its not that the prints from digital clip earlier (this was counteracted in raw processing) but their highlights seem less natural and less smooth than the analagoue prints (difficult to decribe).
#16
[quote name='jenbenn' timestamp='1287663205' post='3718']

I dont want to shoot foul, but really, looking at pints from my slides (yes, properly exposed slides, printed by a professional lab for an exhibition) I see fundamentally less dynamic range in them compared to professional prints of my 50d and 5d Mark II files. The Prints from the slides show however a more pleasing rendition of the brightest highlights. Its not that the prints from digital clip earlier (this was counteracted in raw processing) but their highlights seem less natural and less smooth than the analagoue prints (difficult to decribe).

[/quote]



Maybe I used the term 'dynamic range' wrongly - in a more general way meaning the rendition of highlights and shadows - not necessarily in absolute terms. Like I said, I'm no scientist, but as I understand it digital sensors - just like in audio A/D converters - have no 'headroom' (once they clip, it's gone). And to my eyes, there are very fine subtleties in highlights and shadows in film that are not rendered using a DSLR.



Looking at some old B&W prints I'm seeing an incredibly smooth tonal range in the greys which I have never managed to achieve with a digital camera.
#17
[quote name='Brightcolours' timestamp='1287652927' post='3714']

I agree with Wim in that the call for higher DR is usually not based on anything. High dynamic range results in bland images without contrast, which are NOT the images that would do much for us... just lower the contrast to the minimum in your RAW converter and you will get the idea.



Usually the call for higher DR is just repeating what has been seen on internet forums...



The only advantage of film over digital really is the way sensors collect and store the data... In the lower regions the doubling of the light does not have very many steps in between, making the dark data really rather coarse. But that can not be solved with higher DR.



What usually gets measured as DR (with the likes of DXOmark for instance) is more the noise... but a lower noise in shadows does not convey any different images to our eyes. There is a bad way of improving DR there (noise reduction, which does have its impact usually on IQ, like Nikon and Sony apply), and a good way (improving the AD conversion).



Improving the AD conversion does not have much to do with the photon collection itself, though.

[/quote]

That's right, photon collection stays the same. We just need to improve AD conversion here <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='B)' />. More levels or tones if you like, with a lower SNR.



The dark areas sparse data is not really a problem, because we humans don't really see any detail in dark stuff anyway, if there is a much brighter area in the same view. That is why the noise suppression of Nikon works so well. Just throw away the lower 2% to 3% with all the noise during AD readouts and conversion of sensor data to RAW (set it to absolutely black), and seemingly there is not only much less noise, but it also seemingly retains the same DR <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='B)' />. Works very well with the way the human eye sees, however.



Kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....
#18
[quote name='Pinhole' timestamp='1287660810' post='3716']

Actually, BC, my comments are not "repeated from an internet forum", but from years of experience using film (which I still shoot). I don't know about the science and the figures, but subjectively I see more shadow and highlight details in actual prints using low ISO film (either 35mm or 6x7) than my DSLR is able to reproduce.

[/quote]

Pinhole, if you do see more, I sincerely suggest you try to shoot to the right in digital, and not rely entirely on the inbuilt lightmeter, and convert RAW to 16-bit tiff or so while adjusting the curve in the process. This to get an s-curve, and shift the entire image towards a DR-range where all tonal values you need are still present. Essentially that is the same as we used to do when developing film in the old days, just that we could only see the result after the development process rather than during th eprocess, as we now can with digital.



I think one needs to tinker as little as possible when converting RAW to an image to actually do the processing on, but correcting exposure (development time), preserving the maximum tonal scale (exposure, development time and developer selection and dilution), making sure it becomes an s-curve adapted to the human eye (extra step due to the linear logarithmic nature of digital) are of the essence IMO.



From there on you should ideally get equal or better results than you did or do with 35 mm film. Not 6X7 yet, you'd need medium format to do that, although FF is getting close now (you'd need about 42 to 46 MP compared to a 6X7 frame).



Low iso B&W film still has a slight advantage over digital up to FF, but we are closing the gap very fast. At 100 iso we are there already, no problem. However, it does require different processing and different exposure than B&W or negative film in general requires.



Kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....
#19
[quote name='Brightcolours' timestamp='1287662703' post='3717']

What you see has little to do with dynamic range, but rather with tonal contrast/curves, i am quite sure. Film has a lower DR than digital.



Your screen (and your prints) are not even able to show the dynamic range of your photos... they are dynamic range limited. Increasing the dynamic range of sensors will not help much.. it is in how to compress the dynamic range available from the sensor data into the range that you are able to print which makes all the difference.



Where film has an advantage is in how it rolls off the registration of bright stuff... it simply does not register it all. Film sort of registers light already in an S-curve (not sure if that is the correct way of saying it, if not I am sure Wim will correct me).



But in short... it is not the dynamic range of sensors that is lacking, it is the linear nature of the light capturing that makes the character of the tonal contrast different (and less appealing to you). Understanding this will bring you a long way in post processing images to approach the character of film.

[/quote]

You are right, although generally there is a slight s-curve present already, essentially due to the nature of the processing from sensor data to RAW. However, it is a far step from the s-curve we naturally get with (B&W) film. Hence my suggestion to add that in the RAW-conversion process, something I personally always do, unless DR is well less than 5 stops (virtually never).



Kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....
#20
[quote name='Pinhole' timestamp='1287660810' post='3716']

Actually, BC, my comments are not "repeated from an internet forum", but from years of experience using film (which I still shoot). I don't know about the science and the figures, but subjectively I see more shadow and highlight details in actual prints using low ISO film (either 35mm or 6x7) than my DSLR is able to reproduce.

[/quote]

Hi Pinhole,



I would agree with 6X7, as that has more resolution available than FF and can record a larger transition, but with 35 mm it is about equal these days, or not far from even low iso 35 mm film, once in print. As I said in my other post, it does require a new way to process the negative (RAW), however, because it is different in nature to film.



[Image: castle09bw.jpg]



The above photograph had an s-curve applied from RAW to 16-bit tiff, nothing else was required there, and in PP it was converted to B&W with contrast and brightness changes to get a tonal range as full as possible available for view and print, and the path (and step) towards the tower were burned a little to lead one into the photograph, the underside of the supporting beam (which only supports itself now <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='B)' />) was dodged a little because the weathering on it made for a nice transition from the clouds, the tower was slightly dodged and sharpened to make it the focal point even better, and in downsizing I had to desharpen the image as a whole a few times in order to prevent sharpening artefacts to appear. I downsize in small steps, of about 1.4X or 1.5X to make sure the software makes a choice rather than just skips pixels.



Shot with 5D II and TS-E 17L, handheld, 1/200s, F/6.3, with a small amount of shift to make converging lines less convergent. I almost had my knees and shoes in this picture <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='B)' />. DR range retained in this picture is about 9.5 stops (Zone 0 to just over Zone 9), compressed to jpeg for viewing (on a calibrated monitor).



Could I have done this with B&W film? Yes, but beyond a 40 cm X 50 cm print it would suffer (and I would likely not be entirely happy with it). I print this very happily at 80 cm X 120 cm and even then can get closer to look at the detail that is actually there.



Kind regards, Wim
Gear: Canon EOS R with 3 primes and 2 zooms, 4 EF-R adapters, Canon EOS 5 (analog), 9 Canon EF primes, a lone Canon EF zoom, 2 extenders, 2 converters, tubes; Olympus OM-D 1 Mk II & Pen F with 12 primes, 6 zooms, and 3 Metabones EF-MFT adapters ....
  


Forum Jump:


Users browsing this thread:
1 Guest(s)