Luminous Landscape Forum
Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: Rainer SLP on December 17, 2002, 06:13:57 pm
-
[font color=\'#000000\']Hi Petru,
How do I explain it?
Why is a 12mm wide angle lens for 35mm full frame a 16mm for the let us say 3/4" chip?
because the distance from the chip plane is the same to the focal central point as the 35mm full frame chip to the 12mm wide angle lens.
Draw again your sketch for example simulating the lens as well as the chip OK?
Now put the 3/4" chip on the 35mm full frame chip OK now draw lines from the 3/4" chip to the central focusing point of your 12mm lens OK and what do you get. A leser steep angle but a 16mm wide lens.
So assume instead of being a 12mm wide angle 35mm full frame lens that it is a 16mm. Again the same you get a lesser steep angle on the 3/4" chip but the focal length is then let us say 24mm OK?
If you want to have a real 16mm focal length wide angle lens on the 3/4" chip you have to move the central focusing point towards the 3/4" chip until you get the same angle as the 16mm on the 35mm full frame and Voila! your angle is the same as on the 35mm full frame chip.
Everything clear?
If you have a lesser steep angle then you do not have the same focal length! Very easy!
The bigger your negative the bigger your focal length compared to other negative sizes. With negative I mean Film or chip, does not matter.
Take a look at my sketch:
http://www.rainerehlert.com/fotos/wideangle.jpg (http://www.rainerehlert.com/fotos/wideangle.jpg)
Do not be confused because the green lines of the 16mm 3/4" do not meet the 16mm black lines of the 35mm chip, they do in infinity and there is where the focal length is calculated.
Or is somebody here than can tell me that I am wrong with my assumption?[/font]
-
[font color=\'#000000\']In that review there is vignetting, but what is the cause? The lens or the lens mount? I can't see why a FF CCD or CMOS sensor would vignette any worse than a piece of film of the same size.[/font]
-
digicam test results often show well over 120lp/mm while pro 35mm zoom lenses are typically back at 60lp/mm or less. Clearly 35mm designers are still pushing to improve, so it must be genuinely easier for the smaller image circle designs.
BJL,
I've never seen any test results of a digicam or P&S camera. 120 lp/mm is impressive by 35mm standards but not enough if the sensor is only 8mm x 6mm (or thereabouts).
There's confusion between 'system' resolution and lens resolution. 35mm zooms might typically produce a 'system' resolution limit of 60 lp/mm, but the lens itself must surely be capable of much higher than this. The D60 at 135 pixels (photosites) per mm has a maximum resolving power of 67 lp/mm. I think this is exaggerated on two counts. The Nyquist theory which suggests 2 pixels are required for each line pair is not perfect in practice. Somewhat more than 2 pixels are required (according to Norman Koren). Secondly, the D60 has an AA filter which, as we know, blurs the image slightly. Net effect is probably, a wild guess, 55 lp/mm. Doesn't matter how good your lens is, you're not going to get more than (55 lp/mm?) with the D60.
Now I'm going out on a limb here, and I stand to be corrected, but I suspect the opposite applies to P&S cameras. The sensors actually have higher resolving power than the lenses, hence no need for an AA filter. Any resolution test are therefore likely to apply to the lens itself. The sensor is probably capable of even higher resolution.
As I see it, from a very simple mathematical point of view, a 35mm standard lens which is truly diffraction limited at F2.8 could produce an image the same size, quality and DoF as a standard lens for an 8" x12" field camera, diffraction limited at F22, provided the 35mm sensor has sufficient resolving power. I can't see much point in pinning one's hopes on an even smaller format than 35mm.
-
Some of us are also old enough to remember half frame. A great idea that never caught on.
I believe that the cost differential between APS sized sensors and full frame sensors is a brief historical anomaly. That's one of the reasons why I believe that if Nikon is taking a 4/3rds or similar route they are making a mistake. Canon has just completed a major investment in a new CMOS imaging chip fabrication plant. This will allow them to bring the price of these sensors down to the point that there will be little financial advantage to smaller chips.
Add the installed base of tens of millions of full frame lenses and the desire by pros and serious hobbiests to have the highest possible image quality and I can't see much of an advantage for reduced frame sensors after another couple of years.
Michael
-
A 4/3" sensor gets almost twice the milage out of a 400mm lens as a "Full Frame" sensor. Glass for smaller sensor imagers is lighter, faster, cheaper.
Matthew,
But the bottom line for us enthusiasts is the final quality of the enlarged print. If technology advances to the extent that a 4/3rds format sensor can produce acceptably sharp Super A3 size prints with low noise and reasonably high dynamic range, then the same technology applied to the larger full frame 35mm sensor will allow for equally sharp but larger prints with perhaps even lower noise and higher dynamic range. The heavier lenses will have their purpose.
-
The bottom line is NOT the quality of the print. Or we all would be using 8x10 view cameras, or even 16x20.
Matthew,
Okay! Perhaps I'm guilty of a bit of hyperbole there. The bottom line is a compromise between quality and convenience. The 8x10 field camera has a significant quality advantage over 35mm but an absolutely huge inconvenience disadvantage - hence its lack of popularity.
What concerns me about the smaller format with lighter lenses is, how much quality might I be sacrificing for the greater convenience? At this stage it's simply not clear. The fact is, most people do not use super telephoto lenses, nor has there been any comparison that I'm aware of between a Canon 500mm lens and a 4/3rds designed 500mm equivalent.
For example, Michaels 500mm lens is so good it can be used with a 1.4x converter to produce an effective 700mm lens.
I would suggest that the 4/3rds size sensor would have such high pixel density, designed to extract the most out of its lenses, there'd be no room (resolution to spare) for a converter. From a quality point of view, the comparison might be between an equivalent 700mm (or even a 1000mm) 4/3rds lens and a 500mm 35mm lens + converter.
Supposing one were to find that a good 35mm lens with converter ( and bearing in mind the superior noise and dynamic range of the larger photodetectors) still gave a higher quality result than any 4/3rds system.
As a general observation, it's my view that when you change whole systems which involve a lot of additional expenditure on infrastructure (in the case of cameras, on lenses and accessories) you have to make sure you're offering a substantial improvement in quality and/or convenience. There's no way that the 4/3rds format can compete quality wise with 35mm (I suspect, or am led to believe) and the greater convenience of the smaller format seems relatively marginal with the most commonly used lenses.
Nevertheless, I'm not against the introduction of a smaller format with different proportions. I'm all in favour of variety of choice.
-
[font color=\'#000000\']Michael, I think you have your economics wrong on the cost of imaging chips. The cost/area remains constant. Moore's law says you can put more in a given area, thus the price of a functional unit decreases. The correct assumption is that a 22 megapixel chip will cost no more than an 11 megapixel chip in a couple of years, but also that a given sensor size will remain about the same.
Dan[/font]
-
[font color=\'#000000\']Hi,
Sorry I meant 4/3" and not 3/4", but anyhow this is not so important.
Afterwards it came to my mind why do we speak of a 80mm focal length in MF as normal when we all are used to say that 50mm on 35mm is normal.
Because of the angle of view.[/font]
-
[font color=\'#000000\']Actually if you follow the costing models of the chips over the last say 4 years, we are following Moore's law pretty close.
The 460 Kodak was 35K, the 660 was 25K, then the huge drop was the 760, at 7K. Of course it didn't have a larger chip than the 660 but all new electronics.
Now Canon's 1ds is almost 2x the 6mp 760 and is less than 1K more than the 760 Kodak. I use the 760 as my model since its body model is the Nikon F5 which is the base eqivalent to the Canon. I realize that there are many differences, but as for overall body strength the two are very close.
If you really read Moore's law and get into the theory, it also gives mention to the other piece of this puzzle, the basic law of supply and demand. Canon has the only camera currently that can offer a high end body, and high end 11 mp chip. They basically can set the price where they want it. Any by picking a price right above the 760 I think that they did OK. I have made many posts to the effect that I also believe that they could have made the price lower and sold more, but only Canon knows their internal costing model.. I will have to say that there is more profit in 1 1ds, than about 8 D60s.
Until Nikon or Someone else brings their body out to equal Canon, there is not reason to change and this also effects Moores law, as it is looking at the industry in general. In general there are always more than one of any product, example PC's VCR's, you have to have more than one.
Someone also brought up the "sleeper" issue on the FF for Nikon. The mount opening. Nikon is smaller than Canon and if you look at Phil Askey's 1ds review, on some of the image you can see vignetting on the corners. As much as I hate to admit it, I don't see how Nikon will get around this, or Kodak with the 14n. It means that you won't be able to take advantage of the F2.8 speed of all your high end glass, instead having to start at around F4.0.
Nikon is in a pickel unless some sort of software correction can be made and I feel thats where they are headed. They have added the vignetting correction in Capture 3.5, but thats for current cameras.
Paul CAldwell[/font]
-
There's confusion between 'system' resolution and lens resolution. 35mm zooms might typically produce a 'system' resolution limit of 60 lp/mm, but the lens itself must surely be capable of much higher than this.
...
I can't see much point in pinning one's hopes on an even smaller format than 35mm.
I was talking about lens resolution only.
The figures I quote for 35mm pro lenses are for lens resolution, based on "MTF50", the lp/mm at which MTF is 50%, as recommended by Norman Koren and others. For example,
http://www.normankoren.com/Tutorials/MTF1A.html (http://www.normankoren.com/Tutorials/MTF1A.html)
gives an MTF50 of 61lp/mm for the Canon 28-70mm f/2.8L, one of the sharpest 35mm normal focal length zooms around.
I ignore the far higher lp/mm figures based on "extinction resolution" at which a very high contrast target is barely visible: those numbers correspond to resolving only 2-10% of true subject contrast, and so are of little relevance to human perception of a normal image.
Clearly the "one-piece digicam" test results are "system resolution", but the lens resolution is at least as good, and as you say only slightly better, since lenses are probably the main limit there.
The figures I quote for them might be a bit optimistic as they are not rigourous "MTF50", but nor are they the maximal "extinction resolution" figures.
If anyone wants endless data on 35mm and medium format lens resolution, measured as MTF at various resolutions up to 40lp/mm, check out http://www.photodo.com/nav/prodindex.html (http://www.photodo.com/nav/prodindex.html)
I pin my hopes on a somewhat smaller format than 35mm, probably no smaller than the Kodak/Olympus 4/3" format, because it could easily exceed the resolution needed by the overwhelming majority of photographers in a more convenient way: enough resolution for ANY SIZE of print so long as it is viewed "normally", meaning from a distance at least as great as the image size, so that most or all of the image fits into the field of view. Even with large murals and panoramas, the closest distance that makes sense to me for normal viewing is the short dimension of the print. About 3000 full colour pixels across a distance equal to ones viewing distance is all the eye can distinguish, pessimistically up to 4500 for Bayer pattern pixels. That is partly why 35mm lens MTF is typically measured only at up to 40lp/mm, which corresponds to just under 2000 lines by 3000 lines across the whole frame.
Thus I expect to eventually leave the bulky, more expensive lenses and other gear of larger formats (including 35mm format DSLR's) and their higher resolutions to the small minority of high end photographers who care about scutinizing small portions of the whole image, or who use primes and therefore more often have to crop significantly, or who need backward compatability with existing 35mm lenses.
P. S. I will also be happy to ditch the 2:3 aspect ratio of 35mm format.
-
People seem to forget that 35mm film was not created as a still picture format. It's spectacular success has probably been due to the fact that the cameras are about optimum size for hand holding by most people; certainly by the males who previously were the main users. Those who have a great vision about the opportunity to throw off old restrictions from the past are deluded. There were no such restrictions. By the late 70's Pentax, with their M Series, out-miniaturised the Olympus OM Series. Yet big Nikons and Canons remained popular. Don't men like their big EOS 1V's and F5's? Really? Pentax MX's and Olympus OM's were beautiful cameras, ideal for portability AND had the same size sensor as their larger competitors. Now the suggestion is that folk will clamour for a sensor half the size of what's not broke! Kodak have a great track record on new small formats. It's a bit of an obsession; 110, Disk, APS, ... Don't be surprised if 4:3 goes the same way. Suppose Canon announces a full frame D90 at a sensible priice. Who will want to sell their Canon gear to pick up an inferior 4:3 Kodak?
-
Why then is virtually every standard enlargement bigger than 4"x6" cropped to a squarer shape than the 2:3 (1.5) ratio of 35mm's 24mmx36mm frame?
BJL,
I'm sure there must be lots of reasons and sometimes perhaps no reason at all. Irrationality rules the day.
To satisfy my curiosity, I had a look at all the popular formats. Here are the ratios.
6cm x7cm - 1 : 1.17
Old fashioned TV - 1 : 1.33
A4 - 1 : 1.41
Super A3 - 1 : 1.46
35mm - 1 : 1.5
Widescreen TV - 1 : 1.78
Widescreen Cinema - 1 : 2.35
The Golden Mean is represented by the irrational number 1 : 1.618. This ratio appears to be present in the growth patterns of many, many things in nature. It had a mystical quality for the Greeks. There are many references on the net which suggest the Rule of Thirds originated from 'The Golden Section' (also known as the Golden Mean). I'm no authority on the origin of these terms, so if you have information on this, share it with us.
I find it significant that the two most popular formats that approximate most closely to this mystical Golden Mean are 35mm and Widescreen TV. Is this pure coincidence? Why is there a movement to introduce a new!!?? 4/3rds format. This is the old fashioned TV format. Are we all creatures of fashion?
-
Ray,
you seem to draw very different conclusions than me from the same facts!
First a small numerical correction: 67 medium format is actually 56mm by 69.5mm, a ratio very close to 1.25, presumably chosen to fit 8x10 prints.
The two "longest" formats on your list are moving picture formats, a very different issue than still images; and the next longest, 35mm is simply NOT a heavily used format for serious prints even when it is what the camera gives; the overwhelming majority of prints bigger than snapshots are cropped squarer.
The evidence of the choices made by most serious artists in large format and medium camera design, printing paper sizes, and artists' supplies is that 2:3 is not a good compromise, it is off towards one end of the scale.
I repeat, the most common artistic choices are in the range 1.25 to 1.4, and if any one of the common numbers is nicely in the middle of the common range, to minimize average cropping needs, it is 1.33 or 4/3.
Also, 35mm format does not dominate in digital: the overwhelming majority of digital cameras have 3:4 format. Yes, a lot of 35mm and MF based systems have used 2:3, but probably for compatability with 35mm film scans in the commercial marketplace, and note also the recent trend of square and 3:4 formats in MF/LF backs.
Again, I agree that some subjects like panoramas favor wider shapes, and I look forward to digital panorama cameras in a variety of formats (including perhaps Widelux style, with a rotating lens and a linear sensor), but on the subject of cropping, it is rather clear that 35mm leads to more cropping (mostly at the ends) than any other commonly used film format, and formats like 645 and 67 medium format are favored in part for reducing the amount of cropping needed for the sort of photography most often done with such cameras.
-
[font color=\'#000000\']Dan,
You're quite right, and I likely sholuldn't ave invoked Moore's Law.
But, conceptually my point is the same. Based on some conversations with thos ein the know I'm pretty confident in saying that some manufactruer's are going to have solid yields from CMOS imaging chip wafers in the years ahead and this is going to bring down the cost of full-frame chips.
Certainly they will always cost more than smaller chips, all other things being equal, but the incremental cost of a complete camera might only be increased by a couple of hundred dollars at most. This isn't enough, in my opinion, to justify a paralell reduced-size line of lenses and cameras.
I could be wrong, but my gut tells me I'm not.
Michael[/font]
-
[font color=\'#000000\']Hi Petru,
somehow what you say will also apply on the smaller chips, because the image circle will be smaller and then you get the same problems of steep angle on the smaller chip.
If you have 16mm wide angle on 35mm full frame and 16mm on 3/4" chip you have the same angle, or am I wrong in my thinking?
Just make a sketch on paper and you will see.
So having said this, there will be the same vignetting on the DX lenses as we have right now on the 35mm full frame lenses.
If the frame covers diagonally the image circle, whatever lens you have, you will have the known problems.
Why should a lens specifiaclly designed for the digital e.g. 3/4" be better than the full frame lens?
I keep thinking that a 35mm full frame lens used on a not full frame chip is far better than those new? calculated DX or whatever lenses? Why, because you are not touching the limits of the image circle an therefore working on the better 66% ( or whatever focal correction factor) image circle of the lens. Just take a look at the MTF charts of the lenses and you see where they are good and where these begin to fall down.
Why do they make the aspherical grindings on the lenses? This specially grinding is always on the outer side of the circumference. Take a look at an image taken with the Canon 17-35mm zoom where you clearly see a nice curve in form of an Hunter's Bow. This is because of the aspherical correction.
The industry makes all as " Good as Necessary" and not as " Good as Possible ". OK there are exceptions like the 50mm f1.0 which for me is only a proove of the capability of a company but does not give nothing better than a well known 50mm f1.4.
Why do not we tell the industry to make as a chip in form of a automatically adjusting concave pan to the focal length of the lens (this already exists in the mirrors of the astronomical observatories) and then the light rays will all fall at the same distance at a 90° angle onto the corresponding pixel and the lenses can be made cheaper because we will have no chromatic and sharpness aberration which is the main problem when designing a lens :D Please allow me to keep dreaming of my SS-HAA-100MP (SSuper-Hyper-Automatic-Adjusting-100MegaPixel) Chip :D[/font]
-
[font color=\'#000000\']I'm sure there must be some advantage to a larger lens mount, but why would the F mount's diameter make it not possible to "effectively" cover a 24x36mm sensor? It's the same size as a 35mm frame and they cover that effectively. Also, Kodak seems to have accomplished the feat with their 14n?[/font]
-
[font color=\'#000000\']I too hope that Nikon’s decision to pursue a set of optics for their D series of digital cameras does not detract them from pursuing development of a camera with a full frame sensor.
However, in developing this series of lenses, they can optimize them for the characteristics of the digital sensor. With a digital sensor significant MTF above the Nyquist frequency is undesirable (unless a low pass anti-aliasing filter is used), since it only contributes to aliasing (false image data) The design goal would be to maximize MTF in the usable range of the sensor and have it fall off to less than 10% above this range.
Schneider Optics has a series of lenses specifically designed for digital imaging and they discuss some of their objectives in a white paper posted on their website:
http://www.schneideroptics.com/info....phy.pdf (http://www.schneideroptics.com/info/white_papers/optics_for_digital_photography.pdf)
Norman Koren offers some interesting comments on his web site.
http://www.normankoren.com/Tutorials/MTF7.html (http://www.normankoren.com/Tutorials/MTF7.html)
Of course, a sharp cutoff of MTF above the Nyquist frequency cannot be obtained with a lens as is possible with the audio signal of a CD. It is not possible to have good MTF up to the Nyquest frequency and low MTF beyond. This approach implies some degree of over sampling. It will be interesting to see how the upcoming Kodak camera will perform without a low pass filter).[/font]
-
[font color=\'#000000\']Ray,
I'm no expert on CD, but if oversampling were done during the recording phase, there would be incompatabilities between players using 4x, 8x, or more oversampling. Here is a link which might be helpful:
http://www.repairfaq.org/REPAIR/F_cdfaqb.html#CDFAQB_005 (http://www.repairfaq.org/REPAIR/F_cdfaqb.html#CDFAQB_005)
I just estimated for the 4 micron; maybe it is actually 6.
When I said that the resolution of the small sensor camera was limited by the lens I was using Norman Koren's chart as a reference. He says that an 11 mm diagonal (3.4 um pixel) sensor such as used on the coolpix 5700 becomes diffraction limited at f4; I guess that's why the aperture range you mentioned is limited. Well corrected lenses are limited by diffraction when you stop down, rather than abberations--as I'm sure you know.
Thanks for your replies--you've brought up some good points.
Bill[/font]
-
[font color=\'#000000\']When I said that the resolution of the small sensor camera was limited by the lens I was using Norman Koren's chart as a reference. He says that an 11 mm diagonal (3.4 um pixel) sensor such as used on the coolpix 5700 becomes diffraction limited at f4; I guess that's why the aperture range you mentioned is limited. Well corrected lenses are limited by diffraction when you stop down, rather than abberations--as I'm sure you know.[/font]
[font color=\'#000000\']Bill,
That's a good point. I wonder if I've misunderstood that concept. I've been under the impression that diffraction is a function only of F stop and that any lens at F4 is theoretically capable of resolving (according to Rayleigh's Law) something like 1500/4 or maybe 1000/4 lp/mm. Taking the more conservative figure, that's 250 lp/mm. Not bad for a relatively cheap lens. Now, if only the lens designers/manufacturers could do this with 35mm lenses - ie. optimum performance at F4 instead of the usual F8!!
BTW, the oversampling in the recording stage of the new DVD audio format is just that - a new format. It's not compatible with a normal CD player.[/font]
-
The figures I quote for 35mm pro lenses are for lens resolution, based on "MTF50", the lp/mm at which MTF is 50%, as recommended by Norman Koren and others. For example,
http://www.normankoren.com/Tutorials/MTF1A.html (http://www.normankoren.com/Tutorials/MTF1A.html)
gives an MTF50 of 61lp/mm for the Canon 28-70mm f/2.8L, one of the sharpest 35mm normal focal length zooms around.
I ignore the far higher lp/mm figures based on "extinction resolution" at which a very high contrast target is barely visible: those numbers correspond to resolving only 2-10% of true subject contrast, and so are of little relevance to human perception of a normal image.
BJL,
Maybe I'm out of my depth here, but isn't there a lot of resolution between the MTF50% level and the MTF2% extinction. Maybe such 'buried' detail is irrelevant for the usual enlargements from 35mm, but I envisage some time in the future when people are making 24"x36" prints (and larger) from 24MP 35mm full frame sensors, they'll want that extra detail however faint. I believe the Rayleigh's diffraction limit applies to a 9% or 10% MTF. I would have thought even detail that's lost 90% of its original contrast can be enhanced and made useful.
But I agree there's a big future for the smaller than 35mm format for the non-photographic enthusiast or the occasional photographer who's not likely to print anything larger than A4 or Super A3.
-
Ah, but the field advantages of a smaller sensor and effective focal length multiplier continue to exist. I've come to like the 1.6 multiplier of my D60, and am really hoping that the EOS 3D or what ever it is called will not be full frame but somewhere between 1.3x and the current D60 1.6x. Reasons include using the sweet spot of the lens, and better reach for telephoto applications.
Dan
-
The ratio of 2 to 3 lends itself pretty well to historical compositional values. ... I'm referring to the 'Rule of Thirds', 'The Golden Mean' ...
Why then is virtually every standard enlargement bigger than 4"x6" cropped to a squarer shape than the 2:3 (1.5) ratio of 35mm's 24mmx36mm frame?
The common non-snapshot print paper shapes are 5:7 (ratio 1.4) , 8:10 and 16x20 (ratio 1.25), 11x14 (ratio 1.27), plus the newer "European" or "metric" A3 and A4 papers of ratio approximately 1.41 (actually the square root of 2). You see the same story in the range of large format and medium format cameras, where 6x9 is definitely at the "panoramic" end of the range. The same is also seen in artists' sketch pads and pre-stretched canvases, where the most common shapes seem to be 4:5 and 3:4 (9"x12", 12"x16" etc.), and while shapes of 2:3 and up exist, they are far less common. Everywhere I look other than snapshots, shapes in the range 1.25 to about 1.4 are predominant, and 3:4 (1.33) fits reasonably well.
Of course I like panaramas too, so I have no objections to such longer shapes being available, but I was talking about a convenient mainstream compromise shape.
The Rule of Thirds has nothing to do with the Golden Mean; that guideline works fairly nicely with a variety of image shapes. Prints with approximately the shape of the Golden Ratio (about 1.62) are not very common.
As far as shapes with interesting geometrical properties, I like those "metric" papers A3, A4 etc., because when you cut a sheet in half, the two new pieces are the same shape as the original. Very convenient in the darkroom I would think. (The nearby 5"x7" happens to be my personal favorite shape for verticals.)
-
JBL,
In other words, there's a plethora of different formats that seem to have arisen for mostly practical, economic and accidental reasons, and none of them would appear to have any over all advantage from a purely esthetic point of view. Forget the Golden Mean, is that what you're saying?
Of course, the nature of the subject affects the format. Portraiture tends to fit a 4/3rds format and landscapes often require a more elongated format than even 35mm can provide. At the moment, photographers are well served in having a range of format options from square to 6cm x 17cm. If you can afford the camera, there's hardly a format not catered for.
But this is not the case with digital where the 35mm proportions still dominate, even for digital MF backs. Whatever the original reasons for 35mm becoming popular (and they appear to be economic) the fact remains that it's a good compromise between the extremes of square and panorama. Just a moderate amount of cropping gets you most formats.
The great disadvantage of the proposed 4/3rds format is that it's not only a much smaller format requiring a huge enlargement for a good size print, but becomes smaller still with the heavy cropping required for many landscapes. I can't see any over all advantage in moving away from the 35mm proportions (except perhaps to a completely square format to more fully exploit the lens's image circle) and I still find it significant that the Golden Mean ratio of 1 : 1.618 is slap in the middle of the two most popular formats - widescreen TV and 35mm, and I would say that widescreen's 1 : 1.78 is here to stay for many, many years.
-
JBL,
You certainly put up a convincing argument. I have not investigated in the world of painting, the preponderance of one format over another. I've got no further than observing there's a huge variety of formats, so for the time being I'll have to accept your word for that. But even if it proves to be true that the 4/3rds format dominates the world of painting, the question is still open in my mind as to whether there are practical, economic and/or traditional reasons for this, rather than esthetic considerations. Artists are subject to economic realities like everyone. For all I know, there are 'practical' preferences of the buying public that painters might, even unconsciously, cater to. This might have to do with the rectangular and fragmented shape of the average living room wall where windows and doorways leave spaces that are more suited to a particular format. One of my favourite photos, that I painstakingly created by stitching images together, is 96" x 13". I've often thought of marketing this photo. I sold one copy to the local council, which gave me a bit of encouragement. But I really think it's too long for the average living room. The resolution is there to make it 192'' x 26". (Any offers?)
I find from my own practice, that most of the cropping I do is to fit A4 or Super A3. Paper is expensive. Sometimes I can't avoid it and end up wasting a 3" wide strip along the length of Super A3. Sometimes I think, well if I make that a 4" strip, I can use it, but then I'm compromising (in my view) the artistic merit of the photo - and so it goes on. Practical realities.
-
[font color=\'#000000\']In your defense, die size is ever increasing, so QA standards are allowing fewer defects in a given area, which means more good chips off a given die, and thus lower overall chip costs.
Dan[/font]
-
[font color=\'#000000\']Hi Petru,
somehow what you say will also apply on the smaller chips, because the image circle will be smaller and then you get the same problems of steep angle on the smaller chip.
If you have 16mm wide angle on 35mm full frame and 16mm on 3/4" chip you have the same angle, or am I wrong in my thinking?
Just make a sketch on paper and you will see.[/font]
[font color=\'#000000\']I just made a sketch on paper And my conclusion is that the angle is less steep if the distance between the (redesigned) lens and the (smaller) sensor is the same. Of course, if you scaled everything (the sensor, the lens, and the distance between the lens and the sensor) you'd have the same geometry as before.
Sorry if my assumptions are incorrect, I'm not an optics engineer ...[/font]
-
[font color=\'#000000\']Is it possible that Nikon could be trapped by the diameter of its mount ? I mean, Contax changed its mount mainly to have a bigger diameter in order to effectively cover full frame sensors. And when Canon created the EOS mount, it made it huge compared to anything else that was on the market at that time.
I remember reading in 'Leica Lens Compendium' by Erwin Puts that one big reason for Leica to move from the 39mm mount to the M mount of the M3 was that it needed a larger diameter (albeit now small compared to today standards, but still big compared to the film/flange distance).
So it might be pay back time for Canon and grief time for Nikon and its pseudo F compatibility.[/font]
-
[font color=\'#000000\']Is this a myth that is having a life of its own? That ultrawide lenses have a problem because their cone angle is so great that the rays strike at too acute an angle? I believe it was Olympus(?) who raised the issue in their ads for lenses needing to be specially designed for digital? DSLRs with mirrors for viewing are a completely different ball game, to the ZSLRs, IMHO. The cone angle hitting the sensor is determined to a large extent by how far the rearmost lens element is from the sensor. In a non-reflex-mirror camera, like a mythical full-format Olympus non-mirror-reflex body, a 14mm focal length lens may have its rearmost element almost kissing the sensor, hence the Olympus ads. And of course you may have huge problems with rays arriving at very acute angles on the edges of the sensor. In a mirrored DSLR, however, the rearmost lens element has to clear the mirror when it flips up. The rearmost element is very far from the sensor (retrofocus lens designs) hence this issue of acute angles is no big deal. I would hazard a guess that no matter how short the focal length of the retrofocus lens is, the cone angle business is never "much" worse than for a 50mm lens. Typical 50mm designs seem to have their rearmost lens elements very close to the mirror when it flips up. Problems with vignetting and Chromatic Aberration remain with ultrawides however, but Jonathan Sachs (Michael's friend) has provided us with a one-click solution for each of these in Picture Window Pro...[/font]
-
[font color=\'#000000\']I shouldn't have used the term oversampling. At least in audo, oversampling takes place during playback, not recording.[/font]
[font color=\'#000000\']Bill,
I don't believe this is true. The latest DVD Audio formats record at enormously oversampled rates - 196KHz for one format and 2.8Ghz for the other, Sony's SACD system. But that's a 1 bit system, I believe.
I don't know how far analogies with audio can stretch, but some of the terms and mathematical principles seem to be similar. Not sure how you arrive at 4 micron spacing for a 24MP full frame sensor. The Schneider Optics article you referred to earlier mentions 6 microns in relation to 80lp/mm and 24MP.
BTW, some of these point-and-shoot cameras do appear to have fairly good lenses. At least, that's what many of the reviewers claim. I recently bought the 4MP Nikon 4300 as a present for someone. A few test images I took appeared remarkably sharp. I would say capable of being blown up to Super A3. My main objection was, easily blown out highlights and much more noise than I'm used to with the D60. Well, actually I had a lot more objections when comparing it to the D60. The two cameras are not in the same league, but the lenses in these tiny cameras are not necessarily so poor that they're limited by diffraction. The Nikon 4300 has a 'real' aperture range of F2.8 - F4.9. Not much chance of reaching the diffraction limit at those apertures.[/font]
-
[font color=\'#000000\']Ray,
Your assertion concerning resolution as a function of the f number is true. I don't know how to reconcile this with Mr. Koren's statement. Perhaps I misunderstood it. Hopefully, a more knowledgable person will help me out.
I was confused by the relationship of the f number to the diffraction spot (Airy disk). It is related to the f number and is independent of the focal length. For green light (500 nm) it is 2.2 microns at f2, 4.5 at f4, 12 at f 11and 18 at f16. I had assumed that it would be bad if the spot were larger than the pixel element of the sensor, but this doesn't correspond the resolution figures in lp/mm. I give up!
Bill[/font]
-
35mm stuff is scary (try pointing a Canon 70-200 zoom at someone!).
Off topic: what was that old comedy movie where the spy confused his two gadgets: a camera disguised as a rifle, and a rifle disguised as a camera? :D
-
I think Nikon's move is wise.
A lot of talk about "full-frame" sensors. Why? Because lots of people have their old lenses and would like to use them at their normal characteristics.
But digital is a totally new medium. The difference is bigger then it was back when AF appeared. I don't think it is smart to bend all this new medium, the cameras and the lenses, to fit the characteristics of the old medium.
Two things are obvious: big sensors will continue to cost lots of money, as Moore's law does not apply to the size of the chip, but only the pixel count. The other is that if I have a camera with a sensor of a certain size, I would prefer to buy lenses that are designed for that sensor. Why would I carry lots of heavy glass that I never use? Besides, the design of a digital lens may (and will!) have to differ from the design of a film lens.
There is no need for "full-frame" sensors. Even the expression "full-frame" is absurd. Digital is a new medium, should set up a standard sensor size that provides the best price/quality.
Replacing film has lead to a general reduction in sensor size. P&S cameras are somwhere in the sub-aps range, SLRs (absurd as they are in the digital age) around the 1.5 sensors of today, and MF around the 35mm frame.
-
People seem to forget that 35mm film was not created as a still picture format. It's spectacular success has probably been due to the fact that the cameras are about optimum size for hand holding by most people; certainly by the males who previously were the main users. Those who have a great vision about the opportunity to throw off old restrictions from the past are deluded. There were no such restrictions. By the late 70's Pentax, with their M Series, out-miniaturised the Olympus OM Series. Yet big Nikons and Canons remained popular. Don't men like their big EOS 1V's and F5's? Really? Pentax MX's and Olympus OM's were beautiful cameras, ideal for portability AND had the same size sensor as their larger competitors. Now the suggestion is that folk will clamour for a sensor half the size of what's not broke! Kodak have a great track record on new small formats. It's a bit of an obsession; 110, Disk, APS, ... Don't be surprised if 4:3 goes the same way. Suppose Canon announces a full frame D90 at a sensible priice. Who will want to sell their Canon gear to pick up an inferior 4:3 Kodak?
I really don't know if the 4/3 format will succeed but the comparisons being made to other "smaller" formats are not quite valid. A lot of the smaller formats failed because the technology didn't produce satisfactory results or the execution (from a business point of view) wasn't good enough. APS, I think, was a case of unfortunate timing - digital cameras came along at about the same time. Had it been competing against 35mm alone, it might well have succeeded.
If full-frame sensors continue to be vastly more expensive than the smaller ones used in the 4/3 format and if the lenses of the 4/3 format can be made significantly cheaper than full-frame lenses, there is no reason why the 4/3 format shouldn't succeed. Picture quality isn't everything - if that were the case, medium format would have killed off 35mm. The 35mm format "succeeded" because it offered smaller size, lower weight and acceptably good picture quality for a significantly lower price than medium format. No reason why 4/3 shouldn't succeed if it can offer those same advantages.
- HCE
-
The real driving force for smaller than full-frame imagers will be the long-lens crowd, both old and new.
A 4/3" sensor gets almost twice the milage out of a 400mm lens as a "Full Frame" sensor. Glass for smaller sensor imagers is lighter, faster, cheaper.
Anyone who wants to hike and take images of birds, animals, etc. is a potential customer of a 4/3 system, a fast 300mm lens, and a TC. The poor schmuck who has to cart 40 pounds and $10,000 of heavy tripod, BWL 600mm, Wimberly mount to get the same coverage is going to be at a severe disadvantage. While some are willing to shoot from the roadside, others need reasonable gear that can be carried for miles and doesn't cost as much as a very nice used automobile.
-
The two "longest" formats on your list are moving picture formats, a very different issue than still images
JBL,
I forgot to ask. In what way are moving picture formats a very different issue?
-
[font color=\'#000000\']I read somewhere that the 35mm format wideangle lenses might have certain problems when working with full frame CCD/CMOS sensors. More specifically, some of the rays of light come from a very steep angle - vs the perpendicular to the surface of the sensor - and therefore the affected sensor pixels (e.g. the ones close to the edges of the sensor) will not "charge" fully.
I'm not sure if this is really the case, but I think that there is some truth to that. If it is a serious issue then maybe Nikon is right, from a strictly technical point of view. Oh, and let's not forget that both Nikon and Canon couldn't care less about their customers' so-called "investments" in lenses. What they want us to do is B-U-Y as much as possible, not continue to use the old lenses for the rest of our lives!!![/font]
-
[font color=\'#000000\']Ray,
I shouldn't have used the term oversampling. At least in audo, oversampling takes place during playback, not recording.
If your output requires 40 lp/mm you could quadruple the pixel density of the sensor and effectively sample at 80 lp/mm. To control alaising in this fashon Kodak would have to use 4 micron pixel spacing, quadrupling the pixel count of their sensor. The small sensor point and shoot cameras do not need antialaising filters because the lens+sensor MTF at full resolution is already so low that alaising is not noticable. If these cameras had better lenses used at apertures above the diffraction limit, I presume alaising would then appear.
Perhaps, as you say, some of the alaising can be removed by software without seriously degrading the resolution of the image.
Bill[/font]
-
[font color=\'#000000\'][font color=\'#000000\']I believe you are right about difraction spot size: the Airy diameter of the diffraction spot is "(aperture ratio) times 1.22microns" for green light (wavelength 0.5micron).
The problem is that for larger image circles, it gets harder to achieve a given sharpness in terms on lp/mm, as aberrations are harder to control. (Though Norm Koren sugest f/4 as an upper limit above which aberration limitation sets in, suggesting that this is independent of image circle size.)
For example, expensive pro grade prime medium format lenses are commonly less sharp than even the best "consumer" 35m zoom lenses in terms of lp/mm; the larger format wins in overall resolution by having "more mm of film" and so more of what really counts: "line pairs per picture height and width". (Check the extensive MTF lens test data at http://www.photodo.com/nav/prodindex.html) (http://www.photodo.com/nav/prodindex.html))
Likewise, moderately priced compact digital camera lenses with about 11mm or smaller image circles are sharper in lp/mm than many pro 35mm zoom lenses: digicam test results often show well over 120lp/mm while pro 35mm zoom lenses are typically back at 60lp/mm or less. Clearly 35mm designers are still pushing to improve, so it must be genuinely easier for the smaller image circle designs.[/font]
-
I can't see any problem with the 35mm format. I don't see any historical baggage that needs to be got rid of. The ratio of 2 to 3 lends itself pretty well to historical compositional values. Is it these values that some folks want to discard? I'm referring to the 'Rule of Thirds', 'The Golden Mean', 'The Golden Section', 'The Golden Rectangle', or whatever name you want to give it. The ancient Greeks discovered that the proportions 5 to 8 are visually pleasing. 35mm is 5 to 7 1/2. Pretty close. Essentially, a rectangle in the proportions of 5 to 8 has the interesting property that the ratio of the short side to the long side is almost the same as the ratio of the long side to the sum of both sides. (ie. 5 to 8 is approximately the same as 8 to 13.) If you're stuck for compositional inspiration, you can always fall back on this rule. Divide the 35mm frame into 9 equal rectangles and place objects within them, if possible connected by a diagonal flow from left to right, the way we read. Should look good.
-
I'm no authority on the origin of these terms, so if you have information on this, share it with us.
I can answer some of the questions; I invite corrections on the history though.
The Golden Ratio gives a rectangle such that if you add a square on the long side, the new rectangle is the same shape. This leads somehow to the biological/mathematical fact that spiralling growth tends to settle on this shape.
The history I have heard of the 3:4 and 2:3 ratios, plus wider ratios for moving pictures, is this:
a) Early movies settled on the 35mm format of Edison and Eastman, with a 3:4 shape (24mm across the film, 18mm along it).
TV copied this shape for obvious reasons; computer monitors copied TV's, early digital cameras used sensors originally designed for TV; hence the persistence of 3:4.
c) The 35mm movie frame was deemed too small for the higher quality standards of a stationary image, and Leica decided on the solution of "double frame": two movie frames became one 35mm still frame of 24mmx36mm. (Thus the so-called "half-frame" was the original, sometimes called "single frame" in older photography books.)
d) The movie industry reacted to TV by differentiating themselves with wider formats often just by cropping the frame to less than 18mm high). TV is only now responding with HDTV.
e) For many decades, almost every 35mm film print was cropped to a squarer shape; even standard prints used to be something like 3 1/2" by 5" (a ratio about 1.43) up until about the 1970's; probably originally a convenience based on halfing a 5"x7" sheet. (Why not quarter an 8"x10" sheet to get 4"x5" prints? I suspect that shape was considered a bit too square for snapshots, with too much sky and foreground in the average landscape.)
f) In the early, contact printing days of photography, various film sizes were somewhat commonly used, and the main ones I know of are 8 by 10, 5 by 7, 6 1/2 by 8 1/2, and 6 1/4 by 8 1/4. All are squarer than 2:3, and the last two get very close to 3:4.
g) The one new photographic frame shape adopted "from scratch" in recent times and widely accepted by serious photogaphers is 645 medium format; actually most brands are 41.5mm by 56mm, or almost exactly 3:4. Strangely, it is often said wrongly that its shape exactly fits 8x10 prints; in fact it is close but a bit "longer".
Given all the historical contraints in photographic equipment and printing supplies, I looked at painting and drawing supplies where shapes are far more flexible and there are many more choices of size and shape used; this lead to my observation that 4:5 and 3:4 are the mainstream favorites, with a huge array of other options appropriate for various situations of course.
Finally, the new 4/3" format is only a new size, preserving the predominant digital frame shape of 4:3 while about doubling the 2/3" format that is currently the upper limit in compact digital cameras; and with a cute double meaning in the name.
-
Ray,
perhaps we can end this discussion (or move it to a new thread on "The shape of things to come"), with at least one agreement: the shape of those "European (and Australian) paper sizes", A4, A3, A2, etc. (non super versions) is close enough to being the same as the 5 by 7 shape that I also like for many purposes, so perhaps we really want a "Euro format" camera option too, which would sometimes avoid all waste of pixels and paper, even when a sheet is used for two smaller prints of that same shape.
By the way, movie screen shapes "just are" different, I have no deep reasons to back up that comment.
-
[font color=\'#000000\']It is not possible to have good MTF up to the Nyquest frequency and low MTF beyond. This approach implies some degree of over sampling.[/font]
[font color=\'#000000\']Bjanes,
The above quote implies that oversampling is not possible. Seems to me oversampling is already taking place with small sensor 3 & 4MP point-and-shoot cameras. These cameras do not require AA filters. The lens is the filter.
Of course, the consequences of this oversampling is low dynamic range and high noise, which presumably is one major reason why cameras like the 1Ds do not have 24MP (yet). However, it seems reasonable to assume that, as technology advances, it will eventually be possible to cram 24MP onto a 36x24mm chip without losing any of the quality currently possessed by the 1Ds. It's really all a trade-off. What is it that professional photographers appreciate most? What type of camera will sell best, one with a higher pixel count and higher resolution, but moderate dynamic range and noise, or one with a lower pixel count but better DR and noise. Hopefully we'll have a choice to accommodate different tastes and purposes, and the beginnings of a choice are already appearing with the 1Ds and 14n, both quite different cameras. Unfortunately, it's not a very useful choice for people who are locked into one or the other system. But as far as I can predict, the Kodak 14n will definitely beat the 1Ds in the resolution stakes, for two reasons; higher pixel count and no AA filter. How much of a hassle removing the inevitable aliasing artefacts becomes remains to be seen. But, as Paul Caldwell mentioned, it can be done in software without degrading the resolution nearly as much as an AA filter does.[/font]
-
The 35mm film format certainly does have some historical baggage that the digital transition might be a great opportunity to discard, and lens size reduction in particular will be a strong attraction even for moderately serious photographers: note for example, Michael R's decision against the 600mm lens partly due to its bulk.
On the other hand, drastically smaller sensor sizes run into some real "scaling problems", which is why I expect that nothing smaller than APS or maybe 4/3" format will catch on for the sort of photographers who want good manual image controls and a reasonable amount of quality, like nice sharp prints at up to 8"x10", or up US letter or A4 size in this inkjet era. Roughly, I am talking about the "consumer SLR and above" market.
My estimate now is that shrinkage by a linear factor of 1.5 is workable (e.g. Nikon DSLR's) but that a shrinkage factor of 1.9 (4/3" format) is at or beyond the limit. Maybe the size reduction will not be enough to be worth it and 4/3" will succeed only in high end "fixed lens" cameras like the Olympus E and Minolta 7 series, but I look forward to seeing experiments like those of Nikon, Kodak, Olympus and Fuji decide this in the market place.
The historical baggage includes the 35mm frame SHAPE: before digital, most serious photographers disparaged the 3:2 aspect ratio of 35mm, preferring squarer shapes like 5:4 or the 4:3 of 645 medium format (and of most digital cameras). Ironically, that same group is now the main one hanging onto 3:2, probably for the sake of backward compatability.
Film frame size also hit a lower limit due to resolution needs for standard enlargement sizes and the inertia of the well established 35mm lens market. Though emulsions are may now be good enough for many purposes at APS size, those emulsions and APS clearly arrived too late, relative to digital, to change mainstream film cameras designs.
A number of items do set a lower size for a "serious camera", though.
1. Camera size:
On one hand, I disagree with the claim that "real photographers WANT big, heavy, 35mm pro size cameras": many ACCEPT the size as the price of fast motor drives etc., while high end rangefinder camera users often mention their smaller size as an advantage.
On the other hand, down-sizing will reach a point where the extensive controls and usably large LCD's do not fit well, and so set a lower limit on serious camera sizes far bigger that a typical P&S digital. Looking at the high end fixed-lens digitals like the Olympus E and Minolta 7 series, they can still get distinctly smaller than 35mm though.
2. Selective depth of field (i.e. deliberately blurred backgrounds and foregrounds):
Smaller formats needs larger aperture ratios to achieve a given depth of field for the same angular field of view (they need the same aperture diameter at smaller focal length). Smaller formats do seem to make it easier to design lenses with larger maximum aperture ratios, but going to 4/3" format or below seems to require portrait lenses with f/1 or beyond, which sounds inherently difficult for lens design. Format shrinkage by factors of 1.5 (current Nikon DSLR's) should be able to handle DOF control though: for example a fast 50mm lens would probably do in place of an 80mm portrait lens. (Do any Nikon DSLR users care to comment?)
3. Pixel quantity and quality:
matching the quality of decent 8"x10" prints needs about 2000 pixels (or more) on the short edge of the frame, and reportedly at least 5 or 6 micron pixel size for adequate exposure latitude and noise levels, so there is a lower size limit, but it is distinctly smaller than 35mm: somewhere around 4/3" format in fact (13.5mm by 18mm). Even Canon D1s resolution of 2700 pixels on the short edge with 6 micron pixels would fit a "1.5 cropping factor" format.
4. Diffraction vs aberration trade offs; but the guidelines I have read suggest that the limits do not clash until one gets more than a factor of 2 smaller than 35mm format.
-
[font color=\'#000000\']Michael -- I understand the logic that full frame, higher res, lower noise sensors will be available at (eventually) lower prices and therefore you'd like to see Nikon continue to focus on this genre of equipment. Fair enough. As a Canon user, perhaps you should extend your point to cover the need for Nikon to do this to maintain some technical and price competition for Canon.
What you don't mention is an alternative market, that I find quite important: the digital Leica (without the baggage of the past). What I'm really talking about here is the need for a high quality digital system of reasonable size/weight. I see the need for this in landscape work b/c 35mm stuff is heavy! And, there is life beyond landscape. In the people/documentary world, 35mm stuff is scary (try pointing a Canon 70-200 zoom at someone!).[/font]
-
Ray,
The bottom line is NOT the quality of the print. Or we all would be using 8x10 view cameras, or even 16x20.
When even a top pro like Michael decides the 600mm lens doesn't make sense for him, even when shooting close to his vehicle, it is obvious that these lenses are just too big, too unweildy and (for 95%+ of photographers) too expensive.
The right tool for the job will prevail. Remember, 35mm was once the "inferior" upstart, but it managed to completely displace MF and LF in news, sports, and wildlife photography virtually totally.
-
Oops!! I didn't realise those A4, A3 and A2 sizes were 'non-American' standards. Just goes to show how ignorant I am. So you Americans are being influenced by the Europeans, eh?