Pages: 1 [2] 3   Go Down

Author Topic: Leaf backs at 28MP, 35MP  (Read 42955 times)

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Leaf backs at 28MP, 35MP
« Reply #20 on: December 31, 2005, 02:50:30 am »

Quote
But the fact is, I can get the same shot with my 35mm DSLR. I just have to use different settings. Instead of F2.8 and ISO 100, I use f8 and ISO 800, or instead of 1/200th, f2.8 and ISO 400, I use 1/200th, f8 and ISO 3200. Experience tells me the 35mm DSLR will still provide better image quality in the above examples. There's nothing the smaller format camera can do that the larger format can't, except be cheaper, smaller and lighter.
That "fact" rather neatly sidesteps the fact that ISO is not infinitely increasable. Just because a 35mm sensor can be constructed to function well at ISO 1600 doesn't mean that an 8x12 sensor can just as easily be constructed to go up to ISO 142400 (89 X 1600) and deliver the same overall image quality and dynamic range. In fact, if identical technologies were involved in both sensors, the 35mm sensor would deliver a better overall image than the 8x12 sensor because the 8x12 sensor would have hugely higher levels of dark current noise than the 24x36mm sensor. Something like 89X the dark current noise, actually.

Larger-format sensors only offer image quality improvements when they are used to increase the number of photons imaged during exposure. Without a larger aperture, a longer exposure time, or increased ISO, that is not possible. All of these are tradeoffs that can negatively affect image quality, either by decreasing DOF, increasing motion blur, or increasing noise. To achieve the increased image quality, you must choose some combination of these three options. You cannot magically gain image quality by simply spreading out the photons collected during an exposure over a larger imaging area; all else being equal, doing so is actually counterproductive. Making apples-to-oranges comparisons between sensors of different formats with different designs and technologies doesn't change that. It's worth noting that the extremely expensive medium format sensors do not yet match the ISO range of the one found in the 1Ds-MkII. Nor do larger-format film stocks (645, 8x10, etc) come in a wider range of ISO ratings than 35mm film. There is no free lunch.
« Last Edit: December 31, 2005, 02:51:10 am by Jonathan Wienke »
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Leaf backs at 28MP, 35MP
« Reply #21 on: December 31, 2005, 11:42:23 am »

Quote
That "fact" rather neatly sidesteps the fact that ISO is not infinitely increasable.

I think we're a long way from infinity making a jump from 35mm to 8x10 (12) which, as I said, is  about the same order of magnitude as going from the smallest digicam to 35mm (ie. 3x4.5mm to 24x36mm).

But you are right that in order to get the improved image quality and resolution from the same number of photons (which is what you're getting with an LF 400mm lens at f16 and ISO 6400 compared with a 50mm lens at f2 and ISO 100 on 35mm, ie. same shutter speed, same size aperture, same FoV, same DoF, same number of photons)) you would need to virtually eliminate dark noise. That would entail a lot of additional technology which would make the LF digital camera even heavier and more expensive.

I'm not suggesting such a camera is going to be a consumer product any time soon, but for all I know the US government might already have such a device.

If you lower the standards for the LF digicam, ie. given a static subject and long exposure it will always provide better image quality than a smaller format but never worse image quality using the same exposure as the smaller format, then the technological hurdles are reduced.

The basis for these apparently absurd statements I'm making is an observation that small digicams do not produce better image quality in poor light than the much larger format Canon DSLRs. I'm merely doing a bit of extrapolation.

I'm suggesting that the reason Leaf backs and medium format sensors in general do not boast a low noise 6400 or 12800 ISO setting are due to niche market factors and plain economics rather than an intrinsic technological handicap due to their greater size. But I could be wrong   .
« Last Edit: December 31, 2005, 11:47:13 am by Ray »
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Leaf backs at 28MP, 35MP
« Reply #22 on: December 31, 2005, 12:07:44 pm »

Quote
The basis for these apparently absurd statements I'm making is an observation that small digicams do not produce better image quality in poor light than the much larger format Canon DSLRs. I'm merely doing a bit of extrapolation.
But not valid extrapolation, because if you took the number of photons involved in a well-executed ISO 800 24x36-format exposure, and dumped them on the digicam sensor, you'd still have blown highlights at ISO 100. The smaller sensor simply can't accept the increased photon load of the larger-format sensor. If you turned it around and dumped the photons involved in a perfectly-exposed ISO 100 digicam frame on to a 24x36 sensor, you'd have an underexposed mess which would at best be no better than the digicam shot, and would have a very good chance of being much worse. The Canon DSLRs deliver better image quality in low light because they process many more photons per exposure than the digicam, even at higher ISO setrtings. And that's why your comparisons are absurd.
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Leaf backs at 28MP, 35MP
« Reply #23 on: December 31, 2005, 01:07:34 pm »

Quote
I think we're a long way from infinity making a jump from 35mm to 8x10 (12) which, as I said, is  about the same order of magnitude as going from the smallest digicam to 35mm (ie. 3x4.5mm to 24x36mm).
To follow up on my previous post, let's conduct a thought experiment comparing an equal-photon exposure between the two formats you mentioned.

Let's start with an exposure optimized for the digicam. Assume the light level is such that a 4mm lens with a 1mm aperture (or f/4) yields perfect exposure for the digicam set at ISO 100, 1/100 second exposure. Let's translate that exposure to the 1Ds-MkII.

To get the same framing, the focal length needs to change from 4mm to 32mm. Keeping the 1mm physical aperture means we must shoot at f/32 to maintain equal DOF. To keep exposure at 1/100, the ISO would have to be raised to ISO 6400, which isn't even possible. If you raise the ISO to 1600 (the highest setting that is not based on firmware cheats manipulating the digital data after ADC conversion), you're faced with trying to recover a high-ISO image that is 2 stops underexposed, which is unlikely to be better overall than the perfectly-exposed digicam image.

Going the other way, a properly-exposed 1Ds-MkII ISO 1600 image is going to be overexposed by two stops given equal aperture diameter and exposure time, and ISO 100 on the digicam. The digicam simply can't handle the photon overload.

The best format for a given application (leaving lens limitations aside) is the one that is just large enough to handle the photon capacity of the desired exposure--sufficiently short shutter speed to minimize motion blur to acceptable levels, and sufficiently small aperture to achieve acceptable DOF--when the sensor is at its native ISO. No format is "best" in all situations.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Leaf backs at 28MP, 35MP
« Reply #24 on: December 31, 2005, 01:18:21 pm »

Quote
If you turned it around and dumped the photons involved in a perfectly-exposed ISO 100 digicam frame on to a 24x36 sensor, you'd have an underexposed mess which would at best be no better than the digicam shot, and would have a very good chance of being much worse. The Canon DSLRs deliver better image quality in low light because they process many more photons per exposure than the digicam, even at higher ISO setrtings. And that's why your comparisons are absurd.
[a href=\"index.php?act=findpost&pid=54845\"][{POST_SNAPBACK}][/a]

So basically you're saying my observation is wrong. That means I'll have to go back and check it. I've got a Sony T1 which I thought would be a great little camera to carry around everywhere in my shirt pocket. I no longer use it because ultimately image quality in any circumstance is compromised. It's only advantage is it's light and compact (although a little too dense for a shirt pocket).

I did a comparison with my Canon D60 at equivalent DoF and shutter speed but adjusted ISO to get the same shutter speed. Image quality was similar but the D60 was marginally better.

I've read reports on other forums of similar comparisons. I've never come across an example of a P&S digicam that can produce better image quality in poor lighting conditions (without flash of course) than a Canon DSLR at equivalent DoF.

However, because I haven't come acroos it means little. Put me straight and point me to such an example.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Leaf backs at 28MP, 35MP
« Reply #25 on: December 31, 2005, 01:41:27 pm »

Quote
To follow up on my previous post, let's conduct a thought experiment

No need to conduct a thought experiment. There are lots of practical examples.

If I was marketing a small digicam that could produce better image quality in poor lighting conditions than the heavier and much more expensive DSLR, I'd play it for all it was worth.
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Leaf backs at 28MP, 35MP
« Reply #26 on: December 31, 2005, 01:45:13 pm »

Quote
I've read reports on other forums of similar comparisons. I've never come across an example of a P&S digicam that can produce better image quality in poor lighting conditions (without flash of course) than a Canon DSLR at equivalent DoF.

However, because I haven't come acroos it means little. Put me straight and point me to such an example.
Canon DSLRs are certainly capable of better low-light images sans flash, but such images are not obtained at digicam-equivalent near-infinite DOF. f/2 on a 1Ds typically involves a much larger physical aperture than f/2 on a digicam, so much so that even when shooting at ISO 1600, you're still imaging many more photons than you could with the digicam at ISO 100.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Leaf backs at 28MP, 35MP
« Reply #27 on: December 31, 2005, 03:15:52 pm »

Quote
Canon DSLRs are certainly capable of better low-light images sans flash, but such images are not obtained at digicam-equivalent near-infinite DOF. f/2 on a 1Ds typically involves a much larger physical aperture than f/2 on a digicam, so much so that even when shooting at ISO 1600, you're still imaging many more photons than you could with the digicam at ISO 100.
[a href=\"index.php?act=findpost&pid=54860\"][{POST_SNAPBACK}][/a]

Jonathan,
Point me to some real world results. Small 2/3rds digicams have approximately 1/16th the area of FF 35mm. To get equivalent DoF on FF 35mm, the F stop has to be multiplied by 4. Ie., F2 on small digicams equates to F8 on FF 35mm.

To get the same exposure (shutter speed) on the 35mm DSLR, one has to bump up ISO from 100  t0 1600.

The question thus arises, how does a small digicam shot at ISO 100 compare with a FF 35mm shot (5D specifically) at ISO 1600.

Qujite favourably is my impression.
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Leaf backs at 28MP, 35MP
« Reply #28 on: December 31, 2005, 04:19:58 pm »

The example I gave was based on your earlier digicam example. If you compare to a different format, obviously the comparison parameters change. But I'd say that an ISO 100 image from a K-M A2 would compare fairly well to an equal-aperture/exposure ISO 1600 image from a 1D-MkII.

I stand by my assertion that a larger format is only advantageous if you can use it to image more photons than a smaller format. If so, the larger format is more suitable. If not, you're breaking even at best and run a significant risk of degrading image quality.
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Leaf backs at 28MP, 35MP
« Reply #29 on: December 31, 2005, 08:19:04 pm »

Jonathan has stated an idea that I than has great merit:

the dominant benefits of larger format sensors relate to gathering more photons and thus "more information" from the subject.

That requires longer exposure times and/or larger effective aperture diameters (e.g same f-stop with a longer focal length), and thus forces resolution trade-offs related to the effects of camera-subject motion and/or increased OOF blur for all parts of the scene not at the exact focus distance. Look at where MF and LF film cameras are and are not used and you should get the picture.

Dark noise can be reduced, but at most this will reduce, not eliminate, the advantage of smaller sensors when an equal amount of light is gathered in different formats. And dark current comes from some fairly fundamental physical sources, so reducing it much below current levels probably requires fairly drastic measures like cooling the sensor. Cooling is already done in some bulky high end equipment, but might never come in a conveniently hand-holdable camera.
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Leaf backs at 28MP, 35MP
« Reply #30 on: January 01, 2006, 05:50:24 am »

Quote
The example I gave was based on your earlier digicam example. If you compare to a different format, obviously the comparison parameters change. But I'd say that an ISO 100 image from a K-M A2 would compare fairly well to an equal-aperture/exposure ISO 1600 image from a 1D-MkII.

I stand by my assertion that a larger format is only advantageous if you can use it to image more photons than a smaller format. If so, the larger format is more suitable. If not, you're breaking even at best and run a significant risk of degrading image quality.
[a href=\"index.php?act=findpost&pid=54883\"][{POST_SNAPBACK}][/a]

Perhaps I'm not doing a good job getting my point across so let me try another tack. Whatever the formats being compared, if the physical aperture is the same, the FoV the same, the shutter speed the same then the number of photons will be the same. In my example of 35mm format at f2 and ISO 100 (50mm lens) compared with 8x10 format at f16 and ISO 6400 (400mm lens), the total number of photons reaching the sensor is the same, right?

So what's different? Well, I'm making an assumption here that it's possible to optimise an LF lens so it's truly diffraction limited at f16. I've not heard of a 35mm lens being diffraction limited at f2. Although the number of photons are the same, their arrangement is different. The image from the 50mm lens at f2 contains lots of aberrations. Many of that total number of photons are wasted or counterproductive, whereas the photons that have passed through the LF lens are as good as they get. You can't get a better image than one that is limited only by lens diffraction.

But you are right that higher dark noise will degrade the sharper image, but by how much is complete guesswork. I have no access to the latest cutting edge technology that's being developed under great secrecy in the labs. All I can do is compare what's already out there and available, and when I do, I find that Canon DSLRs can do just as good a job with the same number of photons as the smaller digicam and I wouldn't be surprised if the 5D does a significantly better job than the latest small P&S digicams.

There's a lot of room in in an 8x10 format camera for a lot of sophisticated technology. Your assertion that what I'm proposing is absurd and impossible reminds me of those mathematicians who thought they could prove that heavier than air machines could not fly.
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Leaf backs at 28MP, 35MP
« Reply #31 on: January 01, 2006, 10:22:23 am »

Quote
So what's different? Well, I'm making an assumption here that it's possible to optimise an LF lens so it's truly diffraction limited at f16. I've not heard of a 35mm lens being diffraction limited at f2. Although the number of photons are the same, their arrangement is different. The image from the 50mm lens at f2 contains lots of aberrations. Many of that total number of photons are wasted or counterproductive, whereas the photons that have passed through the LF lens are as good as they get. You can't get a better image than one that is limited only by lens diffraction.
Quote
Your assertion that what I'm proposing is absurd and impossible reminds me of those mathematicians who thought they could prove that heavier than air machines could not fly.
You know what they say about assumptions...and again, you are being absurd. You've forgotten the practical reality that designing a lens to cover a smaller image circle makes it far easier to achieve a high MTF at high LP/mm. The cheapest digicam lenses maintain 50% MTF at much higher LP/mm values than the best 35mm lenses, and many LF lenses do not perform well when attached to a 1Ds for the same reason. As a result, doubling the linear dimensions of the format approximately doubles the usable resolution in practice, instead of the theoretically-possible squaring it. It's no easier to make a LF lens diffraction-limited at f/16 than it is to make a 35mm lens diffrraction-limited at f/2. I've not heard of either.
« Last Edit: January 01, 2006, 10:36:06 am by Jonathan Wienke »
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Leaf backs at 28MP, 35MP
« Reply #32 on: January 01, 2006, 12:22:13 pm »

Quote
You've forgotten the practical reality that designing a lens to cover a smaller image circle makes it far easier to achieve a high MTF at high LP/mm.


No, I haven't forgotten it. I recall BJL has been playing down this advantage for lenses of focal lengths greater than about 65mm. The  resolution advantages I  see for the larger format (ignoring dynamic range and dark noise considerations for a moment) result from the use of the LF lenses at apertures where diffraction is the main aberration. You simply can't get a sharper image, in combination with the DoF that the aperture allows, when a lens is diffraction limited at the aperture being used, whatever the format.

F64 on 8x10' format does not produce a sharper, more detailed image at the focal plane (as opposed to the recorded image) than f8 on 35mm if both lenses are diffraction limited at those apertures. I'm fairly sure that most 35mm lenses are not diffraction limited at f8. If they were, their performance at f8 would tend to be the same, which it clearly isn't according to the Photodo tests.

If LF lenses are nowhere near being diffraction limited at f16 then of course my argument is diminished. But the main point I've been making in these last few posts is that the smaller format has no advantage other than being small, light and cheap, and to prove it, when I next see my friend who owns a KM A2, I'll do some comparisons with my 5D at equal physical apertures and shutter speeds.
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Leaf backs at 28MP, 35MP
« Reply #33 on: January 01, 2006, 02:32:57 pm »

Quote
F64 on 8x10' format does not produce a sharper, more detailed image at the focal plane (as opposed to the recorded image) than f8 on 35mm if both lenses are diffraction limited at those apertures. I'm fairly sure that most 35mm lenses are not diffraction limited at f8. If they were, their performance at f8 would tend to be the same, which it clearly isn't according to the Photodo tests.
For 35mm lenses, diffraction usually starts kicking in beyond about f/11 or so. f/8-f/11 is generally the "sweet spot" where the lens performs best, with better lenses generally having their sweet spot at larger apertures.

You're actually proving my point here, namely that given an equal-aperture, equal-photon exposure and diffraction-limited lenses, the larger format offers no resolution advantage whatsoever over the smaller one, and the constraints of practically achievable ISO and dark current noise dictate that the large-format image will actually be worse. The only way larger formats "win" is if they are allowed to image more photons, by either increasing exposure time or aperture size.
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6600
Leaf backs at 28MP, 35MP
« Reply #34 on: January 01, 2006, 10:41:26 pm »

Quote
No, I haven't forgotten it. I recall BJL has been playing down this advantage for lenses of focal lengths greater than about 65mm.
[a href=\"index.php?act=findpost&pid=54931\"][{POST_SNAPBACK}][/a]
I think you are misquoting me. My comments were about getting roughly the same absolute resolution (lp/mm) from lenses of the same focal length for different formats when the lenses compared are both of narrower than normal angular field of view.

The comparison being discussed now involve lenses of the same angular field of view for different formats, and thus with the larger format lens having a longer focal length. Here both evidence and theory suggest that the shorter focal length lens for the smaller format will typically have higher absolute resolution.

One source of experimental data is the MTF graphs provided at the website of Schneider.
First compare lenses in the same family at different focal lengths: as the focal length goes up, MTF at equal lp/mm values tends to go down.
Then compare their new Digitars, designed for smaller image circles and generally of shorter focal lengths, to any other Schneider view camera lenses: the smaller format Digitars have dramatically better MTF performance. (Compare at 20lp/mm, the lowest value used for the Digitars but the highest value used for any other lenses.)

In fact, judging from the MTF curves at 20, 40 and in most cases 60 lp/mm, the Digitars seem to have absolute resolution second only to the Olympus Zuiko Digital lenses for Four Thirds format (for which MTF curves also available at 20 and 60 lp/mm). But even some good Olympus zooms outperform the Digitar primes, so once again, smaller formats win out on absolute resolution.
« Last Edit: January 01, 2006, 10:45:16 pm by BJL »
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Leaf backs at 28MP, 35MP
« Reply #35 on: January 02, 2006, 12:10:18 am »

Quote
For 35mm lenses, diffraction usually starts kicking in beyond about f/11 or so. f/8-f/11 is generally the "sweet spot" where the lens performs best, with better lenses generally having their sweet spot at larger apertures.

You're actually proving my point here, namely that given an equal-aperture, equal-photon exposure and diffraction-limited lenses, the larger format offers no resolution advantage whatsoever over the smaller one, [a href=\"index.php?act=findpost&pid=54942\"][{POST_SNAPBACK}][/a]

Yes, given equal physical apertures that are diffraction limited, equal FoV and equal photon count, there would be no resolution advantage to the larger format (assuming adequate sensors with more than sufficient pixels, reduced dark noise and no tilt to produce a DoF advantage).

However, in asserting that a diffraction limited LF lens at f16 would be just as difficult to make as a 35mm lens diffraction limited at f2, you are also making one big mother-of-all-stuffups' assumption. Where's your evidence for this? What facts or statements from LF experts are you using to deduce this?

I would agree that diffraction tends to dominate somwhere between f8 and f11 for 35mm, which is not necessarily the same as the sweet spot. The Canon 200/1.8 has a sweet spot at f4. It's marginally sharper at f4 than f8 but nowhere near diffraction limited at f4. If it was, its resolution in lp/mm would be double that at f8.

I think you've missed the point I'm making. With the exception of a few really expensive lenses such as the Canon 200/1.8, you are limited to a maximum resolution at around f8 to f11, which gives you the same image resolution and DoF as 8x10" at f64 to f90. With 35mm, I can sacrifice a bit of DoF and get an equally sharp image at f5.6, maybe. If a shallow DoF is desired, then of course it's no sacrifice.

Now according to statements from people such as Norman Koren, LF lenses used at f64 and f90 do not produce spectacularly detailed results because at such apertures the lenses are way beyond the diffraction threshold of the lenses. You have to open them up to f22 to get close to their full resolution potential. That's equivalent to 35mm f2.8. Show me the 35mm lens which can produce a dramatic increase in resolution by opening up to f2.8.

Add the tilt fator of LF cameras to the equation, then providing the subject lends itself to that process, you have both a dramtic increase in resolution whilst maintaining good DoF, but using just the same number of photons (provided you fit enough technology in that large box to dramatically reduce dark noise).
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Leaf backs at 28MP, 35MP
« Reply #36 on: January 02, 2006, 12:44:13 am »

Quote
The comparison being discussed now involve lenses of the same angular field of view for different formats, and thus with the larger format lens having a longer focal length. Here both evidence and theory suggest that the shorter focal length lens for the smaller format will typically have higher absolute resolution.

[a href=\"index.php?act=findpost&pid=54983\"][{POST_SNAPBACK}][/a]

BJL,
Well now you've confused me. If the term 'absolute resolution' refers to line pairs per mm, then absolute resolution is not what we should be talking about. Surely it's total picture resolution that counts. As camera format increases in size, the absolute resolution falls off in inverse proportion to the increase in format size, provided the lenses are used at apertures where diffraction is the only limit to resolution, and assuming equal FoV and DoF for the different format lenses

Once again, if I can choose a format so that I will always be using the lens at diffraction limited apertures, whether or not a shallow or deep DoF is desired,. then I have a potential resolution advantage. As I see it, an 8x10" format camera is the most likely to provide that opportunity, followed by MF.

In other words, with LF I have the option to trade off DoF for an increase in resolution. There's no such option with 35mm. I'm stuck with the resolution it can deliver at f9 (say) and I just hope that resolution is not too bad at f2 or f1.8. (My Sigma 20/1.8 is pretty bad at F1.8, but even the much praised Canon 50/1.4 is not too hot at f1.8 and f2).)
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Leaf backs at 28MP, 35MP
« Reply #37 on: January 02, 2006, 04:31:32 am »

Quote
However, in asserting that a diffraction limited LF lens at f16 would be just as difficult to make as a 35mm lens diffraction limited at f2, you are also making one big mother-of-all-stuffups' assumption. Where's your evidence for this? What facts or statements from LF experts are you using to deduce this?
Try re-reading BJL's post about Schneider lenses, for starters. It's exactly what you're asking for. And then do a bit of figuring with regard to the LP/mm and MTF achievable by cheap digicam lenses compared to the best 35mm and LF lenses and you'll find a pretty direct inverse correlation between image circle coverage and the LP/mm value where MTF falls below 50%. I'm not just pulling premises out of my butt here, all you have to do is go to any digicam review at DPReview and compare the sensor size and resolution, and convert that to LP/mm, then contrast that to LP/mm specs published for 35mm lenses or LF lenses. You'll find the digicam lenses are delivering pretty decent MTF percentages at LP/mm values that are far higher than 35mm or LF lenses can dream of. The only reason a cheap digicam lens can outperform an expensive LF lens so decisively is because it does not need to cover as large of an image circle as the LF lens.

Quote
In other words, with LF I have the option to trade off DoF for an increase in resolution. There's no such option with 35mm. I'm stuck with the resolution it can deliver at f9 (say) and I just hope that resolution is not too bad at f2 or f1.8. (My Sigma 20/1.8 is pretty bad at F1.8, but even the much praised Canon 50/1.4 is not too hot at f1.8 and f2).)
Again, that proves my point that in order for LF to provide a usable advantage it must image more photons. And stating that 35mm does not offer any ability to trade off resolution for DOF is simply stupid. No lens in any format is perfect; all formats suffer from lens aberrations that are far from the diffraction when the lens is wide open. There's a disconnect between the diffraction limit and what the lens can actually deliver; this gap becomes wider as the imaging circle of the lens increases. Between that and the exposure time increases mandated by the fairly narrow range of available sensor/film ISO capabilites, and increased cost, weight, and bulk, larger formats become progressively less practical; the circumstances under which they can deliver a useful image become progressively more restrictive as size increases, to the point at which they become nearly useless. If larger is always better, why stop at 8x10? Why not 16x20, or 24x36, or something even larger?
Logged

Ray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 10365
Leaf backs at 28MP, 35MP
« Reply #38 on: January 02, 2006, 09:04:26 am »

Quote
do a bit of figuring with regard to the LP/mm and MTF achievable by cheap digicam lenses compared to the best 35mm and LF lenses and you'll find a pretty direct inverse correlation between image circle coverage and the LP/mm value where MTF falls below 50%.

I've done the figuring. That's why I'm able to make and understand statements that all lenses at equal physical aperture and equal FoV, regardless of format, will provide the same total image resolution (excluding degradation caused by film or sensor). That shorter focal length lenses with smaller image circles deliver higher spatial resolution (more lp/mm) is implicit in the above statement.

Were it not so, then 35mm at a diffraction limited f8 could not provide the same resolution as 8x10 at a diffraction limited f64. Absolute resolution at f8 has to be 8x as great as LF at f64. That's simple maths.

The point I'm making, which again seems to have escaped you, is that lp/mm at f4 with 35mm is not even nearly 8x as great as the equivalent 8x10 f32. Where's the problem in appreciating that fact? It seems quite clear to me, or are you perhaps refuting my claim that opening up from f64 on 8x10 format to f32 does not substantially increase resolution?

I admit there's not much reliable information available on the net regarding MTF tests of LF lenses, but following is an extract from a Schneider Optics FAQ on large format lenses.

 7. Why does the sharpness of my lens increase as I stop down, and then decrease as I stop down further?

You are experiencing the effects of two primary optical aberrations, spherical aberration and diffraction. As you stop the lens down, spherical aberration is reduced, and the effects of diffraction are increased. This generally results in a "sweet spot" around f11, where the image has benefited from reduced spherical aberration, but not yet been degraded severely by diffraction. At small apertures such as f64, diffraction effects are quite large and the image will be noticeably softer.

The interesting thing here is that Schneider Optics LF lenses have a sweet spot at f11. You'd almost think they were referring to 35mm. I assume it's the smaller LF lenses for 4x5 format they're referring to. Even so, a sweet spot at f11 on 4x5 format is equivalent to a 35mm lens with a sweet spot at f2.8.

The most thorough series of LF lens tests I can find on the net is by Perez and Talman. Their results can be viewed here

Again, it's not clear which of the tested lenses would be suitable for 8x10 format, but I think it's likely that some of the 360mm and longer would be. I checked out some of the names and it seems some photographers are using some of these lenses with 8x10 format, such as the Fuji C-Series F12.5 for which there are 2 sets of results in the table, the first of which shows best resolution (and equal resolution) from f12.5 to f22. At f32, diffraction is already taking its toll. That would be equivalent to a 35mm lens having a sweet spot at f stops ranging from f1.6 to f2.8 with a progressive deterioration in resolution at larger fstop numbers.

Quote
If larger is always better, why stop at 8x10? Why not 16x20, or 24x36, or something even larger?

Do you not remember a thread a while ago about an American who built his own very,very large format camera. He had the lens poking out of one side of a van, exposed something like a 4'x6' image on the other side of the van and made contact prints of apparently amazing resolution. He drove around visiting schools to demonstrate the principles of photography. Pretty stupid, eh?  

I just came across some MTF details of a wide angle lens with an image circle of 500mm designed for a 9x18" format camera. Usable resolution is 30lp/mm. Such a lens used with a cropped 8x10 format would deliver 30 lp/mm even into the corners at no less than 30% MTF. With a crop factor of 5/3, the 210mm wide angle lens becomes pretty close to a standard lens for 8x10. To capture that resolution you'd need at least a 180mp Foveon type sensor or perhaps better a 300MB Bayer type since sensitivity is an issue. . Do you think there's a snowball's chance in he*ll of anyone designing a 35mm lens that can make use of 180 megapixels. We're only up to 16mp with the 1ds2 and already people are complaining their lenses are not good enough.

Your arguments seem to be flying in the face of the facts, Jonathan. Details of this lens can be found here .
« Last Edit: January 02, 2006, 12:01:49 pm by Ray »
Logged

Jonathan Wienke

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 5829
    • http://visual-vacations.com/
Leaf backs at 28MP, 35MP
« Reply #39 on: January 02, 2006, 02:51:36 pm »

Quote
Do you not remember a thread a while ago about an American who built his own very,very large format camera. He had the lens poking out of one side of a van, exposed something like a 4'x6' image on the other side of the van and made contact prints of apparently amazing resolution. He drove around visiting schools to demonstrate the principles of photography. Pretty stupid, eh? 
Not stupid, but not very practical, either. I doubt he was shooting surfing or weddings or fashion shows or car races or photojournalistic stuff with that rig. I think it's safe to say his exposures were measured in minutes or hours, not fractions of a second.

Quote
I just came across some MTF details of a wide angle lens with an image circle of 500mm designed for a 9x18" format camera. Usable resolution is 30lp/mm. Such a lens used with a cropped 8x10 format would deliver 30 lp/mm even into the corners at no less than 30% MTF. With a crop factor of 5/3, the 210mm wide angle lens becomes pretty close to a standard lens for 8x10. To capture that resolution you'd need at least a 180mp Foveon type sensor or perhaps better a 300MB Bayer type since sensitivity is an issue. . Do you think there's a snowball's chance in he*ll of anyone designing a 35mm lens that can make use of 180 megapixels. We're only up to 16mp with the 1ds2 and already people are complaining their lenses are not good enough.

Your arguments seem to be flying in the face of the facts, Jonathan. Details of this lens can be found here .
You're deliberately mischaracterizing my remarks, and straying far from the equal-aperture/equal-photon exposure comparison we have been discussing. In any such comparison, the smaller-format camera will (all else being equal) deliver a better result until it is no longer capable of imaging all of the photons without overexposure/clipping. Once the quantity of photons in the exposure exceeds the imaging capacity of the smaller format (and clipping occurs), then the larger format will produce the better result, and will continue to do so in increasingly decisive fashion as the photon count is increased until it's imaging capacity is reached. I have never denied that larger formats are capable of greater resolution than smaller ones, even though large-format lenses are not capable of resolving as many lp/mm as smaller-format lenses. But in order to make use of this increased resolution potential, large-format cameras must image more photons per exposure than their small-format counterparts. And that means that their practicality decreases as format size increases.
« Last Edit: January 02, 2006, 02:52:25 pm by Jonathan Wienke »
Logged
Pages: 1 [2] 3   Go Up