Ad
Ad
Ad
Pages: [1] 2 3 ... 6 »   Bottom of Page
Print
Author Topic: The Missing Factor in Sony F828 Reviews  (Read 32947 times)
Laco
Guest
« on: January 24, 2004, 06:29:36 AM »
ReplyReply

Quote
If we were to assume that the F828 Zeiss lens is sharpest at f5.6, then the fall-off at f8 is likely to be negligible, which would make the F828 an ideal camera for all those shots that require great depth of field.
Quote

Who needs great (or greater) depth of field at focal lengths of 7 - 50 mm?
Logged
Howard Smith
Guest
« Reply #1 on: January 26, 2004, 09:35:44 AM »
ReplyReply

I think you need to read (review) Micheal's tutorial on understanding depth of field.  Using a 7 mm lens at f/8 then blowing up the result to same image size as your 200 mm lens will not change the depth of field.  They will be the same.
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5083


« Reply #2 on: January 27, 2004, 09:24:16 AM »
ReplyReply

Ray,

  you seem to working harder than necessary to prove an obvious advantage of larger formats: in exchange for the extra cost and weight of larger format equipment, including sensors with more pixels of the same pixel pitch compared to the smaller format alternatives, one is never worse off for image quality than with a smaller format; as a last resort one can crop down to a part of the image corresponding to what you would have got with a smaller sensor, using the lens's sweet spot in the process. Throw in a few reduced image circle wide angle lenses if you need to preserve wide angle coverage with such crops.

  Some more good news for big formats; in principal, diffraction limitation and DoF limitation scale the same way with image size and focal length: if you half the linear dimensions of the sensor (and of its pixels if you want to keep the same amount of detail in the image), half the focal length to get the same angular FoV, half the aperture ratio (i.e. keep the same aperture diameter), then you will half the diameter of the diffraction spot size and roughly half the CoC for any given out of focus part of the subject. Thus when you double the magnification to make prints of the same size, DoF and diffraction effects will be essentially the same.

  However, if one goes too far down in size, the larger aperture ratios needed become a problem; either completely impractical, or leading to increased optical abberations, so the image quality possible with very small formats will suffer. If one accepts the folk wisdom that f/8-11 is typically optimal for 35mm format, and Norm Koren's claim that f/4 is a lower limit in order to keep optical abberations well under control, the the lower limit for best quality seems to be about half of 35mm dimensions, or roughly 4/3" format.

  But maybe modern aspherical lens designs and such can push good performance down to f/2 which would bring 2/3" format back into the game; it is not fair to judge solely from the very wide ranging zoom lens of the Sony 828, so maybe the Leica lens of the Leica/Panasonic Digilux 2 is a better test.
Logged
Ray
Guest
« Reply #3 on: January 28, 2004, 04:11:57 PM »
ReplyReply

Quote
Why not keep ISO the same (same "film type) and let shutter speed take care of exposure?
Some good reasons; mainly because we can now use different ISO settings with the same ease as changing aperture or shutter speed, and shutter speed requirements are usually more critical for any given shot. If I'm hand holding the camera with a telephoto lens, I've got no choice but to use a higher shutter speed, or get a blurry image. If I'm shooting a Kung Fu demonstration out of the range of my flash, I've got no choice but to use a high shutter speed, or get unrecognizably blurry shots.

Unless it's a static subject and one is using a tripod, it generally makes more sense to choose the shutter speed and aperture and let the ISO fall where it will, rather than choose the ISO and aperture and let the shutter speed fall where it will.

That Photodo contest between 35mm and 9x12cm is fascinating. I've been aware of that for some time. It tends to imply if lens technology can progress to the point where it's actually possible to manufacture a lens that is diffraction limited at, say f2, and if it ever becomes possible to manufacture sensors that are capable of capturing that resolution of a diffraction limited F2 lens, then the larger format camera will no longer have an advantage, except perhaps for shallow DoF.
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5083


« Reply #4 on: January 29, 2004, 10:03:03 AM »
ReplyReply

Quote
"Luminous Landscape" sot of implies the subjet and envionment we may be interested in. Normally, cameras will be tripod mounted and shutter spedd will not be much of an issue.

If smaller is better, it strikes me as odd that folks much smarter than I (the Canon and Nikon pros for starters) are staying up nights trying to make bigger sensors. What are they thinking?
Agreed on the first point: in the context of landscape photography, surely one of tests to do is ultimate image quality; and it seems almost certain that bigger sensors will continue to have the advantage there. Then again, I am sure that Michael never intended to use the Sony 828 as his primary landscape camera!

   About Canon, Nikon et al "staying up nights trying to make bigger sensors", the facts suggest otherwise.

   Agreed, in a small very high end sector, there is an understandable effort to produce "full frame" sensors which make the most advantage of existing high end 35mm and 645 format lenses, and it seems that with the recent 22MP MF backs, this ideal has been more or less reached in both formats. With no serious modernization going on in any of the larger versions of MF, I expect that 645 (56x42mm) will be the permanent upper limit on DSLR sensor size.

   On the other hand, these cameras and backs account for a tiny fraction of the total digital camera market. For the other 99.9%, there has been absolutely no trend to bigger sensors for some years, nor to bigger pixels. Instead, the dominant trend is gradually reducing photosite size and increasing pixel counts as technological change reduces noise levels and increases pixel read rates, while using the same or smaller sensor sizes.

    Since Howard mentioned Canon and Nikon in particular: Canon is not increasing sensor size in this "99.9% sector"; instead it seems to have settled on three DSLR formats of FF, 1.3x and 1.6x for three market sectors of "highest quality/MF replacement", "high quality with high speed", and "amateur", with no size increase ever in the last two sectors. Nikon has produced six DSLR models with no increase ever in sensor size from the 1.5x of its DX format, and is quite emphatic about staying at that size.
   Other DSLR players Pentax, Fuji, Sony (as sensor maker), Minolta and Olympus have also given no sign of increasing their DSLR sensor sizes, and in the one piece digicam market, there has been no sensor size increase since 2/3" format arrived more than three years ago.
Logged
Howard Smith
Guest
« Reply #5 on: January 30, 2004, 07:55:53 AM »
ReplyReply

Let's asume you are correct and one day there will be a 10gb sensor smaller than a bread box.  Even if such a sensor were perfect, the resulting images would be no better than the lens that formed it.  Just the same as an image formed by a perfect lens is no better tahn the film that records the image.  It does not make economical sense for one technology to progress much beyond the other.

There is room to grow in both optics and imaging, but don't too far ahead of yourself.
Logged
Ray
Guest
« Reply #6 on: February 02, 2004, 02:04:44 AM »
ReplyReply

Quote
OK Ray, now I'm really confused. On Jan 31, you said that the advantage of LF was not necesarily its superior lens, but the superior capabilty of the larger piece of film. Then you seem to say that large format lenses are not superior to 35mm lenses. Whih one is it?
Howard,
You're either playing games or being deliberately obtuse. (Never mind! I forgive you  Cheesy )

Some 35mm lenses are superior to some LF lenses, and some LF lenses are superior to some 35mm lenses, in some ways. Not all 35mm lenses are equal and neither are all LF lenses equal. It depends on the design and construction of the lens.

With the digital system I've proposed, there is no noise  or grain to enlarge. Enlargement merely represents a dilution of the resolution that's on the sensor.

In practical terms, if one could manufacture a 4x5" slab of my super electronic material with 4 micron photodiodes capable of generating 256 discrete numerical values, it would definitely be overkill for the quality of current LF lenses, but the camera would nevertheless probably deliver better results at certain apertures, ie. more detail (but less DoF).

I'm assuming that the possibility of this happening is extremely unlikely considering how expensive the 1Ds sensor is. I think it would be fair and reasonable to predict that, in the unlikely event that this could ever happen, the extra cost would never remotely justify the marginal increase in resolution that might be achievable, and therefore no research dollars are likely to be forthcoming for such a project.
Logged
Ray
Guest
« Reply #7 on: February 02, 2004, 07:22:09 PM »
ReplyReply

Quote
The thread has strayed a bit from the original question, but the F828 is exceedingly sharp at F8.
Lin,
I'm glad to hear it! If we strayed a bit from the original issue it's because no-one has come up with the goods.

Thanks for posting that vibrant and spectacular shot of the painting. All we need now is a comparison between the F828 at f8 and the 10D at f13, of a highly detailed subject with the exposure the same in both cases.  Smiley
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5083


« Reply #8 on: February 05, 2004, 12:50:16 PM »
ReplyReply

Ray,
    my example did neglect read noise, so armed with your correction and those "5 electron, 1 electron" claims for read noise, let me try again to seek some natural lower limits on pixel size. My answer will be "about 2 microns, maybe a bit less."

   Some dark current noise figures I have read are 17 electrons for the 6.8micron pixels of the E-1's sensor, 22 for its 16MP 9micron big brother in Michael's MF back, and about 8 for some digicam sensor, the Sony 5MP 2/3" one I think. These numbers seem to scale linearly in pixel spacing, or with the square root of photosite areas, exactly as theory suggests.

   So if read noise cannot be currently be got much below 5 electrons, we could be closing in on the idea that somewhere in the digicam pixel size world, around 3 microns maybe, read noise becomes significant in the deep shadows and sets a lower limit on desirable photosite size, while for anything on the present horizon in the DSLR world (6 microns and up?), read noise can more or less be neglected.

   However, the future is probably brighter: your 1 electron for CMOS figure meshes with reports I have read that there has been significant recent progress in noise reduction in the CCD designs for digicams, and that this is closely related to the trend towards smaller digicam photosites. So maybe we should be cautiously optimistic, and look to 1 electron read noise in the not so distant future. (Do you really believe that CCD's are old hat and that most or all present progress is in the CMOS world? If so, explain it to all the digital back makers, and to Fuji, whose SuperCCD is the high resolution, low noise measurement leader in the APS DSLR field.)

   With dark current noise apparently scaling in proportion to photosite spacing, it can be expected to be more than 1 electron down to pixel spacing of about 6.8/17 =.4microns; smaller than the wavelength of light! So read noise would be practically irrelevant: even at 2microns it would be only one fifth of expected dark noise.

   As an aside, several sources (including Nikon) suggest that below about 2 microns, lens resolution limits mean that you will get little or no resolution gain with further sensor resolution improvements.

  If the limit is 2 microns, I can only dream of 60MP in a 4/3, or 90MP in a Nikon DX, leaving 100MP plus to Canons's 1.3x and full 35mm format territory.


P. S. Can you give me a source for that "Read Noise figures as low as 5 to 11 electrons for CCDs with a full well capacity of 45,000 to 375,000 electrons."?  Because if it means "5 for 45,000 and 11 for 375,000", then there is a considerable reduction of read noise with pixel size, very close to my beloved square root pattern. Remember the idea that, due to the noise reduction effect of downsampling and printing at higher pixel density,

"So long as reducing photosite spacing with constant sensor size causes a noise source to reduce at least in proportion to the square root of signal (i.e. linearly in pixel spacing) there is no loss in image quality."
Logged
Ray
Guest
« Reply #9 on: February 08, 2004, 07:44:08 AM »
ReplyReply

BJL,
Quite amazing really! Now all we need is someone like yourself to translate all this stuff into laypersons' terms  Cheesy .

I'm still none the wiser as regards the image quality limits of the small sensor. It might seem that, as sensor size decreases photon noise increases in significance, as do all other types of noise. The square root of 10 million is a bigger percentage of the total than the square root of 10 billion, and so on.

I've come across statements that nothing can be done about photon shot noise. We're up against fundamental laws of Physics. But that doesn't seem quite true to me. You can always reduce photon noise by letting in more light, ie. reducing ISO, and at the same time effectively reduce all (?) the other types of noise.

Do you know of any fundamental limits? Perhaps photodetectors that are smaller than the wavelength of the light they are intercepting? But I'm not even sure about that. Doesn't light have a weird habit of turning into a particle when we try to catch it?  Cheesy
Logged
Howard Smith
Guest
« Reply #10 on: February 09, 2004, 03:26:26 PM »
ReplyReply

I am not trying to be difficult, but you say some interesting things, like small sensor size will be optimal for "a great majority of photographers" and that larger sensor will never drop below a price suitable for "professionals and very serious amateurs."  I think those very same statements can be made for film.

I am jumping to the conclusion that a small number of photographers (call them pros and very serious amateurs for lack of a better term) may be willing to spend more for a digital camera with a larger sensor.  But why?  If there is no advantage, even a small one, why pay more?  Most pros aren't that stupid, or they would be starving.  But if a larger sensor is at least SOME better, then it is better, regardless of cost.  Small becomes only a better value for the dollar, not a better camera.  Just like medium format is generally more expensive than 35mm, but for some, it is worth the expense, and the bulkier cameras are not a problem.  Same for larger formats.  It's not for everyone, but the images large format can take are better than 35mm.

I will readily acknowlegde that it is not all wrapped up in image quality.  I use medium and large format film.  One reason is I like just like to, even if a 35mm Canon was just as good or better.  I like to fiddle with the manual controls of a 4x5.  I like the kerplunk my Hasselblad makes when I trip the shutter.  It's not a tiny beep.
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5083


« Reply #11 on: February 10, 2004, 01:29:30 PM »
ReplyReply

Quote
In 3 m sensors, even if you leave the read-out-error and thermal noise aside and thus go down to abt. 100 electrons or 300 photons (giving a 10/1 S/N-ratio), you'll have roughly 500 distinguishable steps of intensitiy (by dividing the [18,000;300]
The noise level is (at minimum) square root of electron count, so at 100 electrons, the photon noise level is 10 and the next distinguishable level is about 110. Moving up, the noise level grows with the signal, so the distinguishable gaps spread.
   Optimistically considering photon noise only, about ten levels couldbe distinguished in the lowest f-stop, 14 in the next, 20 in the next, and so on up to 56 in the topmost of the six f-stop range; at most about 160 levels. Almost but not quite enough to use all the 256 levels available in 8-bit output; the 2.7 micron photosite really looks as if it is close to the edge, until the highlight capacity of such small wells can be improved.
Logged
Digi-T
Full Member
***
Offline Offline

Posts: 149


« Reply #12 on: February 18, 2004, 03:43:23 PM »
ReplyReply

You're right, we're not listening anymore so please keep going :laugh:  Woo hoo, page 11.

T
Logged
Howard Smith
Guest
« Reply #13 on: February 19, 2004, 06:20:31 PM »
ReplyReply

Ray, the brown moth thing is bunk.  It was later discovered that the spotted moth does not rest on tree truncks or branchs.  The moths used to support that rot were dead and glued or pinned to trees to prove the theory.

Thank who?

Sorry I ot you and BJL confused.  I'm also confused about why you care about the "Christian movement."  I am laboring under the opinion you are an atheist.
Logged
Howard Smith
Guest
« Reply #14 on: February 20, 2004, 08:20:41 AM »
ReplyReply

Ray, like I said, the gurus of evolution aen't about to let there followers know their theory is in real trouble.  So they don't advertize their failures.  I don't think there was any intention of fraud when the peppered moth story came out.  The only raud is contiuing to peddle it as true when the flaws are known.

I am sure that believers in God are not foolish like atheists.  For a second, let's assume I am wrong - there is no God.  What harm have I done myself or anyone else by living the life of a believer?  But the long term harm for the atheist and the agnostic is unspeakable if they a wrong.  When you meet God, are you going to be provocative and deny Him?  Might as well.  It will too late to embrace Him.

John, if you find a "church" that is "all right," give me a call.  I want to join.  Don't let misguided theologians run or ruin your life.   Jesus did not teach theology, but the nature of God, man, life and the world and the relationships betwen them.  If you study the Bible, you will find the people he had real problems with were the church.  I think Jesus is vy unhappy with churches today.
Logged
Ray
Guest
« Reply #15 on: January 22, 2004, 07:04:02 PM »
ReplyReply

There's an interesting example on dpreview of the F828 purple fringing issue at various apertures. It's called the foil 'torture' test.

At f2, PF is just horrible; at f4 somewhat reduced; at f8 it's virtually all gone.

It seems that, for those who are particularly concerned about this issue of CA, there is a solution. Just use f8.

Okay! It's understood you don't want to use f8 all the time, but the fact that you have that option to slay the dreaded Purple Fringing ogre in one fell swoop is surely significant.

The only good reason I can think of why this solution might not appeal, is the possibility that resolution at f8 is compromised as a result of the very small 2/3rds sensor requiring a greater degree of enlargement for any size print.

In other words, the very fine Zeiss 28-200mm zoom does not have optimum performance at f8. Maybe! Could someone please do me a favour and find out if this is true, because the foil 'torture' test images at dpreview 'appears' to show better resolution at f8.

As most of us know, 35mm lenses tend to have their best performance at f8. Occasionally, a lens will have its best performance at a bigger aperture, especially really expensive telephoto primes, but when this happens that extra performance at, say f4 is very marginal (compared with f8).

I would expect the performance of the Zeiss zoom to be best at around f2.8. But it would have to be significantly better than at f8 to make f8 an unusable solution to PF, especially considering the tremendous DoF advantage at f8, equivalent to almost f32 on a 35mm camera.
Logged
Scott_H
Sr. Member
****
Offline Offline

Posts: 331


« Reply #16 on: January 24, 2004, 08:40:13 AM »
ReplyReply

I could see it being useful, just as I could see the lack of background blur at large apertures being frustrating.  It depends what you are trying to do.
Logged

Ray
Guest
« Reply #17 on: January 25, 2004, 10:46:15 PM »
ReplyReply

Okay! Let's try to make this crystal clear. Manufacturers of digicams, more often than not, give us a 35mm equivalent for focal length because that's what so many of us are used to, but they neglect to give us a DoF equivalent.

I don't know why this is. I can only assume it's because DoF is considered to be more esoteric than focal length and most buyers of P&S digicams wouldn't have a clue what DoF means.

On the other hand, to be fair  Cheesy , it might be because the aspect ratio of the 2/3rds format is different from 35mm, so the 35mm DoF equivalent is not one easily defined figure. It depends on where you've positioned yourself at the time the photo is taken. Are you cropping the 4:3 format of the F828 to equal the 3:2 format of 35mm, or not? Do you want that extra horizontal expanse or not? If you don't, you can get closer to the subject, or use a slightly greater focal length setting on your zoom, and that affects the DoF.

Either way, the DoF equivalent at any given aperture is significantly greater for the smaller format. For equal horizontal FoV it's about 4 F stops, and for equal vertical FoV it's about 3 1/2 F stops, in the case of the 2/3rds format versus 35mm.

I'll use the greater figure for ease of calculation. I'm using a 200mm lens on my 35mm camera and I'm shooting a scene with some interesting stuff in the foreground which I want to be sharp. My lens is likely to be sharpest at f8, very nearly as sharp at f11, a bit off at f16 and decidedly fuzzy at f22.

I have a problem deciding whether or not to use f11 or f16. I know I'm going to blow up this photo to A3+ so I decide to use f11 and the hyperfocal distance principle (bring focussing back slightly from infinity), but I also know that F11 is not going to give me sufficient DoF to get everything sharp that I want to be sharp. I'm between a rock and a hard place.

But hey! Maybe there's a solution. I pull out the F828 from my bag. This lens is sharpest at F5.6, equivalent to f22 on my 35mm camera. Wow!

Maybe the F828 lens is also equally sharp at f8. That's equivalent to f32 on my 35mm camera. Extra wow!

On the other hand, maybe the performance of this Zeiss lens really sucks at f8. I just don't know. The dpreview foil test gives the impression that resolution is actuall better at f8, but I find this difficult to believe.

So, I'm hoping some of you brave owners of the F828 will risk personal trauma and do a few experiments at various apertures and let me know what the results are. I know it's a lot to ask. If you discover that performance at f8 is really lousy, this could make you angry and depressed for days. It's a great risk, but in the greater interest of scientific objectivity, which we can all benefit from, I hope at least some of you will oblige.  Smiley
Logged
BJL
Sr. Member
****
Offline Offline

Posts: 5083


« Reply #18 on: January 27, 2004, 09:39:22 AM »
ReplyReply

Quote
Yes, depth of field calculators do use different size circles of confusion for different formats. You should read the fine print to understand why. The reason is smaller formats are usually enlarged more than larger formats.
Good, it seems that we are all now in complete agreement on the practical point, which is that one DOES have to take format size into account in order to make a useful DoF calculation, because DoF calculations for different formats are and should be done with CoC values adjusted roughly in proportion to the format used.

   (I clearly have read the fine print, since that point about using different magnifications for different formats is one that I have been making repeatedly from the beginning, and that Ray has also discussed.  I do not understand why you think you need to explain it to us at this stage!)
Logged
Howard Smith
Guest
« Reply #19 on: January 28, 2004, 02:36:19 PM »
ReplyReply

There is an article on photodo that attempts to demostrate that 35mm can be as sharp as medium and large format.  In that demonstration, each format is tested with a "normal" focal length lens for each format (50mm, 80mm, and 150mm).  This gave each example the same "perspective."

The lenses were stopped down to an f/stop to produce the same depth of field for each focal length.

The same film (same ISO, not varied) is used for each format (T-Max 100).  Shutter speeds are set for the correct exposure (varied, not constant).

In the comparison proposed above, it is suggested the ISO be adjusted so that shutter speeds would be constant.  That seems to favor the Sony since the higher ISO for the D10 would be presumably noisier.  Why not keep ISO the same (same "film type) and let shutter speed take care of exposure?

The result was the 35mm as sharp but grainy compared with its bigger brothers.
Logged
Pages: [1] 2 3 ... 6 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad