jaapv
RFF Sponsoring Member.
This has been the tendency of some posts around here... I've been crunching some numbers over the weekend, working from the other end, i.e. what is the biological resolution limit.
It is well-known that if a negative is good enough for A4 it is good enough for any enlargement, as the viewing distance will increase with the print size, making the net result about equal.
I postulated a 75x50 cm print, viewed at 75 cm distance. That would even tax film.
The finest detail the human eye can resove at that distance, for a relatively young person of say, 25 years old, is 0,25 mm.
We need, then, 4 points per mm. That makes 3000x2000 points, which is, obviously, 6 Mp. So 10Mp has plenty of reserve.
How does that compare to film? Well, normal 100 ISO slide-film resolves about 50 points/mm at 30% contrast transmission. That is just 1800x1200 pixels = 2.6 MP !!
Let's make the competition stiffer, use Technical Pan or Delta 100.
Those will resolve, used with an Apo-Summicron 90mm about 100 point/mm@ 30%.
Now we are talking! That means 3600x2400 pixels. Nearly 9 Mp. Very close to Leica's 10 Mp. But we needed the finest film and the best lens in the world to get that close.
But wouldn't more Mp's carry the M8 into the realm of medium format camera's? Not with a 1.33 crop sensor. The resolving power of the lens is, at 10Mp with a 1.33 crop, or 16 Mp with a 35 mm sensor, exactly the same as the sensors.So any increase in the number of pixels would necessarily involve reduction of the size. The only result would be that the same point projected by the lens would be picked up by more than one pixel, not creating any increase in resolution.
In terms of resolution sensors and lenses have reached the boundary of biological need. If we find one that we are happy with in terms of colour rendition, contrast and dynamic range, the only reason to"upgrade" will be mechanical, not electronic.
There is a myth that sensors do not have the dynamic range of film. That may still be true of the softest B&W film, provided they are developed accordingly, or with small second-rate sensors. In normal use however, a high-end sensor will render about 10 stops, which just happens to be the same as the number of zones in Ansel Adams' time-honoured zone-system. There is no way a normal slide film can better that.
There remains the issue of sensor noise. For one part I cannot understand that we accept grain in film as normal or even desirable, and will not grant sensors the same courtesy, on the other side sound engineers have been fighting noise for decades now, with some result, but it seems to have levelled out. I doubt there is much to be gained in that department.
Is this the death-knell of film then? I don't think so. There are - and always will be - artistic differences between the two media. I spent a lovely weekend listening to vinyl records - yes they sound nicer, albeit less perfect than a CD and they are still available, as are turn-tables and cartriges and tube amplifiers, and my daily car has been pushed into second place by my TR4 these summer days, which, even if cars have evolved dramatically the last 40 years, drives as well, but with more character. Morgans are still being sold.
Will the M8 turn into a vintage digital camera over the years? It seems to be fairly certain.
It is well-known that if a negative is good enough for A4 it is good enough for any enlargement, as the viewing distance will increase with the print size, making the net result about equal.
I postulated a 75x50 cm print, viewed at 75 cm distance. That would even tax film.
The finest detail the human eye can resove at that distance, for a relatively young person of say, 25 years old, is 0,25 mm.
We need, then, 4 points per mm. That makes 3000x2000 points, which is, obviously, 6 Mp. So 10Mp has plenty of reserve.
How does that compare to film? Well, normal 100 ISO slide-film resolves about 50 points/mm at 30% contrast transmission. That is just 1800x1200 pixels = 2.6 MP !!
Let's make the competition stiffer, use Technical Pan or Delta 100.
Those will resolve, used with an Apo-Summicron 90mm about 100 point/mm@ 30%.
Now we are talking! That means 3600x2400 pixels. Nearly 9 Mp. Very close to Leica's 10 Mp. But we needed the finest film and the best lens in the world to get that close.
But wouldn't more Mp's carry the M8 into the realm of medium format camera's? Not with a 1.33 crop sensor. The resolving power of the lens is, at 10Mp with a 1.33 crop, or 16 Mp with a 35 mm sensor, exactly the same as the sensors.So any increase in the number of pixels would necessarily involve reduction of the size. The only result would be that the same point projected by the lens would be picked up by more than one pixel, not creating any increase in resolution.
In terms of resolution sensors and lenses have reached the boundary of biological need. If we find one that we are happy with in terms of colour rendition, contrast and dynamic range, the only reason to"upgrade" will be mechanical, not electronic.
There is a myth that sensors do not have the dynamic range of film. That may still be true of the softest B&W film, provided they are developed accordingly, or with small second-rate sensors. In normal use however, a high-end sensor will render about 10 stops, which just happens to be the same as the number of zones in Ansel Adams' time-honoured zone-system. There is no way a normal slide film can better that.
There remains the issue of sensor noise. For one part I cannot understand that we accept grain in film as normal or even desirable, and will not grant sensors the same courtesy, on the other side sound engineers have been fighting noise for decades now, with some result, but it seems to have levelled out. I doubt there is much to be gained in that department.
Is this the death-knell of film then? I don't think so. There are - and always will be - artistic differences between the two media. I spent a lovely weekend listening to vinyl records - yes they sound nicer, albeit less perfect than a CD and they are still available, as are turn-tables and cartriges and tube amplifiers, and my daily car has been pushed into second place by my TR4 these summer days, which, even if cars have evolved dramatically the last 40 years, drives as well, but with more character. Morgans are still being sold.
Will the M8 turn into a vintage digital camera over the years? It seems to be fairly certain.
V
varjag
Guest
I'm afraid there might be some flaws in your argument.jaapv said:
First, resolution is measured in line pairs per mm, meaning just that - two lines, normally black and white. To approximate two lines with discrete digital sensor you need 4 pixels (quantization theorem). Then, film has no Bayer color array over it, which effectively reduces resoultion by about 1/3rd. Also, I'm not so sure about 50 lp/mm of "average slide film", IIRC normally it's around 110-135 lp/mm for 100 ISO E6 transparency shot at contrast of 1:6 (typical for real-life shooting conditions).
I had scans from 35mm TMY handheld showing detail for about 17-19Mp resolution. They were taken on $15 Kiev with $10 Helios-103 lens, and I wasn't aiming for making technically perfect shot. Even with worst lens of my bunch, J-12, I normally have what amounts to 7-9 megapixels of detail on 100 ISO color neg.
That said am sure hope M8 will be great user camera, and all odds are for it. It is also clear that digital sensors will surpass film in resolution at some point, but this moment is yet to come. However, many of us folks sticking to film do that not for pixel bragging but for different still just as vaild reasons
EDIT: the film was TMY, not TMZ, sorry.
Last edited by a moderator:
S
Socke
Guest
Jaapv is correct with his estimate that 10MPixel will be plenty for most uses as is varjag when he points out the flaw in Jaapvs reasoning
I have another way to estimate my MPixel needs, I look at what I have at hand which suits me just fine.
I have an old Canon FS2720 slide scanner with some 2700dpi which gives me some 3600x2400 pixels, make that 8.6 MPixel, which is good enough for my 20x30cm prints and I'm just waiting for a 30x40cm.
My digital is a Canon D60 and I feel no need to replace it yet. Sure, I want the 5d viewfinder but I don't need the faster AF nor the 12 MPixel and I bought a 17-35 Zoom for my moderate wideangle needs, so the smaller sensor is not too much of a compromise.
If I can live with a 2nd generation 6MPixel DSLR and a very old slide scanner for years now, I think a 10MPixel Leica M8 will be fine for years to come.
And another thing, compared to newer Canon DSLRs the D60 is quite limited with noise at higher ISO, it's visible at ISO400 and anoying at 800, but newer PC software works wonders on the RAW files. Shooting RAW I get most of the high ISO qualtiy a newer Canon offers.
I have another way to estimate my MPixel needs, I look at what I have at hand which suits me just fine.
I have an old Canon FS2720 slide scanner with some 2700dpi which gives me some 3600x2400 pixels, make that 8.6 MPixel, which is good enough for my 20x30cm prints and I'm just waiting for a 30x40cm.
My digital is a Canon D60 and I feel no need to replace it yet. Sure, I want the 5d viewfinder but I don't need the faster AF nor the 12 MPixel and I bought a 17-35 Zoom for my moderate wideangle needs, so the smaller sensor is not too much of a compromise.
If I can live with a 2nd generation 6MPixel DSLR and a very old slide scanner for years now, I think a 10MPixel Leica M8 will be fine for years to come.
And another thing, compared to newer Canon DSLRs the D60 is quite limited with noise at higher ISO, it's visible at ISO400 and anoying at 800, but newer PC software works wonders on the RAW files. Shooting RAW I get most of the high ISO qualtiy a newer Canon offers.
V
varjag
Guest
Yep, I wouldn't dispute that 10mp is entirely satisfying for most real-life needs. Even 6mp is adequate for majority of publishing purposes.
boilerdoc2
Well-known
With the M8 not having an AA filter It should retain all of the lenses native sharpness straight out of the camera and before any sharpening in the computer. This should produce an immediate boost in quality of images, right? And if the M8 can produce clean high ISO files (which it should) then this camera should be darn near perfect. Look at what the R-D1 has done with 'only' 6 MP and bump that a factor of ~ 2!
It's gonna be a long wait until Photokina.
Steve
It's gonna be a long wait until Photokina.
Steve
jaapv
RFF Sponsoring Member.
varjag said:I'm afraid there might be some flaws in your argument.
First, resolution is measured in line pairs per mm, meaning just that - two lines, normally black and white. To approximate two lines with discrete digital sensor you need 4 pixels (quantization theorem). Then, film has no Bayer color array over it, which effectively reduces resoultion by about 1/3rd. Also, I'm not so sure about 50 lp/mm of "average slide film", IIRC normally it's around 110-135 lp/mm for 100 ISO E6 transparency shot at contrast of 1:6 (typical for real-life shooting conditions).
I had scans from 35mm TMY handheld showing detail for about 17-19Mp resolution. They were taken on $15 Kiev with $10 Helios-103 lens, and I wasn't aiming for making technically perfect shot. Even with worst lens of my bunch, J-12, I normally have what amounts to 7-9 megapixels of detail on 100 ISO color neg.
That said am sure hope M8 will be great user camera, and all odds are for it. It is also clear that digital sensors will surpass film in resolution at some point, but this moment is yet to come. However, many of us folks sticking to film do that not for pixel bragging but for different still just as vaild reasons
EDIT: the film was TMY, not TMZ, sorry.
Sorry,mate I used line-pairs /mm and that is what my numbers are based on. I understand the confusion,it may be my fault for not being clear, both line-pairs/mm and lines/mm are used indiscriminately in literature, making a lot of confusion. I used this correctly,I believe. But as we all seem to agree on the conclusion the argument is moot anyway.
As you are apparently well versed in theory, you'll be able to confirm that resolution for film is measured at contrast transmission. At a transmission of 90% the resolution is about 15 lp/mm, at 10% of far over 150. Usual is to take the resolution at 30% contrast transmission. You are certainly correct at the Bayer filter, however the AAfilter Canon uses is even more of a degradation. That is the main reason that the DMR outperforms the Ds II by a fair margin. Having said that, we can agree that resolution alone is a pretty meaningless parameter for picture quality. A better meausure are MTF or OTF, but I am not sure that a discussion of those would result in a manageble post.......
Having said that, resolution alone is about the worst
Last edited:
V
varjag
Guest
Jaap, if you account for contrast transmission, you can't just relate lp/mm to megapixels, unless you either shoot test chart through the same lens (giving you resolution of film/sensor and lens combo) or contact-print microchart directly onto film and digital sensor. Some magazines do the former in their digital camera tests, and resulting lp/mm figures, let's just say, not something unseen on film. If anyone here seriously believes that 35mm ISO 100 slide equals to 2-something megapixel resolution, I could mail them some of Powershot A60 snaps to try printing at sizes over 5x7"
P.S. Realising that I might sound like a film freak, I'm not. I would actually love to get hands on M8 at some point
P.S. Realising that I might sound like a film freak, I'm not. I would actually love to get hands on M8 at some point
Last edited by a moderator:
jaapv
RFF Sponsoring Member.
All very true,of course it does not, but my argument is not the technical side, but that higher resolutions than say, 6 MP enter the realm of diminishing returns on account of the limitations of the human eye. Of course a close viewing distance for large enlargements and "100% crops" on the monitor will reveal far,far more, but for practical purposes that is meaningless. Otherwise we would fall into the same trap of senseless tests that the photomagazines tend to do.
fgianni
Trainee Amateur
Another thing to consider is that 6MPixel are worth 6Mpixel only on a Foveon Sensor (I think the biggest available is 3MPixel), on traditional CFA Bayer sensors you really need to halve that figure.
Toby
On the alert
jaapv said:
I agree with all this but in marketing terms 10mp is old news. If leica want to really make an impact in the digital market they will have to do better because of the camera magazine obsession with pixel counting. Look at the Oly E-1 -a fine camera but no one looks beyond its pixel count.
fgianni
Trainee Amateur
That said of course I am perfectly happy with my RD-1 with "only" 6MPixel.
S
Socke
Guest
Toby said:
Looking beyond it's megapixel count it gets ugly
Too noisy and where are the small and light lenses a 4:3rds sensor makes possible?
I'm sooooo disapointed by the Pana L1, 1Kg without a lens and then that big brute of a slow zoom in front of it.
S
Socke
Guest
fgianni said:
If it only had framelines for something in the 60° FoV range
The 28 is a tad too tight for my liking and I hate external viewfinders.
V
varjag
Guest
Amen to that. And people who line up for the coming M8 likely do so not for pixel count either: otherwise it's better to save for PhaseOne back or somethingjaapv said:
S
StuartR
Guest
I shot the same scene with my DMR and Velvia. Same camera, same lens. I scanned the Veliva using an Imacon 646 at 6300dpi. The DMR had higher clarity, the Velvia had better color, and the overall resolution was pretty close. The velvia fades away into blobs, and the digital fades away into squares, so at gargantuan picture sizes, I think film decays more gracefully. But up to its limits, the digital is crisper and has higher clarity. The film had the better color, dealt with highlights better (particularly blown highlights), and the finest detail looked more natural (softness vs. pixelation). I choose between the two based on what I feel like shooting, at a given time, but I still prefer film for black and white. I just cannot get the look I like from Acros in Rodinal using digital. Plus, I like to make fiber prints, so digital makes that impossible. But for certain types of work, (particularly with artificial lighting), digital is the easy choice.
Last edited by a moderator:
Gid
Well-known
boilerdoc2 said:
Where do you get this factor of 2 from? 10MP gives you a 30% linear increase in resolution over 6MP. In the real world this is fairly small and does not make a huge difference (10"x7" vs 13"x9" at 300dpi). Also, the M8 isn't here yet and who knows how it will perform.
I've been trying to find a link to some work done by a respectable source with regard how many mega pixels one "needs", but can't find it. His conclusion was that based upon normal viewing distance (visual acuity) you didn't need any more than 4MP. Clearly the bigger the picture the further away you would normally stand to view it - think about the paintings of Pissarot (hope I have the correct one) - just made up of dots that don't make sense close up.
I've also seen double page magazine spreads from the D2H - only 4MP.
Whilst I am at it. Viewing files on screen at 100% is a pretty pointless exercise. You need to look at the file at about 25% on screen to get an impression of what it will look like printed at 300dpi (300/72).
jaapv
RFF Sponsoring Member.
Toby said:
I am convinced that those who would buy a M8 on just on basis of a photomagazine "test" are bound to be deeply disappointed. I doubt that Leica even wants them as customers. It is a sad fact of life that having a red dot on the front of the camera alone does not make one take better photographs, as I myself have proven time and time again......
dll927
Well-known
After all this theoretical, scientific, and engineering discussion, I have one question: Has anyone seen a PRICE on this dreamed-of M-8?
Mark Norton
Well-known
We think it's going to be $4995 + tax, but none of us really know.
jaapv
RFF Sponsoring Member.
dll927 said:
Just think of the old Rolls-Royce salesman crack: "If you have to ask you can't afford it!"
(or did I mean this post for the snobbery thread??)