Hi,

Please comment on this one instead. I realized that some important points were lost or misrepresented in the small extract quoted in the previous post.

Thanks, Vikram

[Start of copied from LUG]

Subject: [Leica] digit stuff for those interested From: Jim Brick Date: Tue, 06 Nov 2001 08:32:40 -0800

----------------------------------------------------------------------

I was told, off line, the following:

"Fact is, your bringing up digital 4x5 lenses, when talking about a Leica, is just ludicrous and you should know it."

Since I wrote a response, I decided to share it with the group.

The parallel is that the "film" 4x5 lenses are too sharp, their MTF is too high for digital sensors, just like MTF of Leica lenses is too high for digital sensors. So Schneider (and others) designed a set of lenses with a MTF that is matched to digital sensor spacing. And wrote a white paper explaining the facts and why they had to do this. Which is why Leica will not mount an M lens on a digital camera. They have designed new lenses for their partnered (Panasonic) cameras. The Canon and Nikon SLR lenses that are used on their respective cameras are not redesigned, but the cameras contain low-pass filters to "dumb down" the lenses before the image gets to the sensor.

The bottom line is that all lenses are equalized by digital sensors. A leica lens is no better or worse than a Sony or Olympus or whatever lens. The things we buy Leica lenses for are lost in the digits. In film, a 1 micron square silver halide grain contains 20 Billion silver halide molecules, each capable of being hit (exposed) by a photon. It only requires three being hit to produce a developable speck. A digital sensor pixel (the minimum recording spot) is 5 microns square (25 sq. microns vs 1 sq micron) and will ultimately report a light level of 0-255 (256 levels) for this whole vast area of 25 sq. microns. This is why Leica lenses out perform most other lenses on film, but are no better than anything else on pixels. And why film can record deep shadows and bright highlights in the same scene. Digital sensors cannot. All fine detail (Leica's strong point) is completely lost. Digital cameras are digital cameras. Their integration into a film camera body by Canon, Nikon, Kodak, & Fuji is simply to give professional photographers a known base to start from. The professional level digital cameras from Olympus and others that don't look like traditional SLR's and have non-interchangeable zoom lenses produce photographs equal in every way to the SLR interchangeable lens cameras. They just aren't "familiar" to the pro photographer.

Astro photographers have the same problem. But instead of dumbing down the lens, they shift the sensor half a pixel in four quadrants, take four exposures, and then analyze the result with software to pick up points that fall in between pixels and to differentiate double stars from single large objects. A static CCD sensor cannot record these (and other) phenomenon. And without either dumbing down the lens MTF or taking multiple exposures and processing the results via software, serious aliasing occurs that is not fixable with software without producing other artifacts. None of these problems occur using film. But film has to be processed and scanned to get the image into a computer for analysis.

Basically, digital cameras are digital cameras. The digital sensor is the great equalizer. All lens/camera brands sharing similar sensors and price will perform equally. Only the post processing software can make a visual difference. Lens performance is completely lost. And it will remain this way as long as the 5 micron square pixel is the smallest obtainable.

Jim [End of quoted material.]

-- Vikram (VSingh493@aol.com), March 15, 2002

Answers

Vikram;

Thanks you very much. I use digital for eth things I can not do with film, such as instant reviewing and Internet usage. I use film because it can do what difital can nopt-detail shadow detail , texture etc.

Long live the Memory of William Henry Fox Talbot and Captain Pizzarelli.

CHeers

-- richard ilomaki (richardjx@hotmail.com), March 15, 2002.

Vikram:

My understanding is similar to what the author has written, but I was under the impression that a useable or effective grain size for film is about .0001 (1/10,000th) of an inch or the equivalent to a 150 megapixel image. This is significantly larger than what the above autor has stated, but still MUCH smaller than the best current cmos or ccd.

But, if what he says about all lenses being equalized by digital sensors is true, wouldn't it also mean that using the same scanner to scan a 35 transparency shot with a Nikon and another with a Leica will produce images that are identical to each other?

-- Jack Flesher (jbflesher@msn.com), March 15, 2002.

Jack,

I have followed this discussion with much interest when it comes up on the LUG for some time now and certainly have respect for Mr. Brick's knowledge in this area.

As I recall his "day job" is in the arena of software for digital imaging and thus he is very knowledgable in the physics involved with sensors and film capabilities.

As with all Leica forums the "digital M" question re-occurs with regularity. Jim will let it go for awhile, then come on to remind everyone that with the current technology available to construct sensors, one will not experience the true capability of Leica lenses with an image on current sensors vis a vis current film technology.

Not to say that won't happen someday, but we ain't there yet.

To address your question of scanning a negative or tranparency I suspect he would answer something like this (and my apologies to him were he to feel I'm speaking out of turn).

It's comparable to the old computer phrase "garbage in, garbage out". If I photographed a scene with two lenses, one of which was superior to the other, I would expect to see the difference in the negatives or transparancies. When scanned digitally the scanner would also see that same difference and foregoing any post scan manipulation, the results would be different also.

But would exactly the same difference result? No, simply because the image created by the "better" lens (at least in MTF terms) has "more to lose" in the process than does the poorer lens.

Think of the question in reverse. Would the "poorer" image become better in the scanning process, becoming as good as the one from the better lens? Not likly with either current or future technology.

Jim's just saying that digital can't equal the performance of todays lenses and film. He would also say I'm sure that they're getting closer as time goes on.

Leica will probably introduce a digital camera (other than their S digitals and the latest Digilux), most probably in the R line, where body space is more readily available. In another 18 years, perhaps the technology will have advanced and they'll come out with the M8! If the past is prolog however, Leica will wait another decade or so beyond that.

Best,

Jerry

-- Jerry Pfile (JerryPfile@msn.com), March 16, 2002.

Another reason lens manufacturers are re-designing their lenses for digital use is to reduce colour fringing. Because the larger digital "grain" (in relation to film grain) exaggerates this effect. I understand that's one of the reasons why the latest Nikon wideangles contain ED glass as well as aspherics.

For the record comparable 120 lenses have inferior MTFs to 35mm lenses, and 4x5 lenses have inferior MTFs to 120 lenses. However, to get a 240mm wide output 4x5 needs to be enlarged about x2, 120 about x4, and 35mm about 6.6 times. This fact of course hides fact that 35mm film is actually sharper that 4x5 film and 35mm lenses have the best MTFs

-- sait (akkirman@clear.net.nz), March 16, 2002.

The quoted piece is full of technical errors and half-truths. The low pass filter mentioned is not designed to 'dumb down' the lens but rather to correct for aliasing artifacts (moire patterns) due to the configuration of the color filter array. The multiple exposure technique described is designed to reduce noise by oversampling, not to 'pick-up points that fall between pixels'. Most CCDs used in astronomy have pixels with 100% fill factor, there is no dead space between them. The arguments on film superiority in terms of MTF and dynamic range are also completely unfounded. In fact, in the field of scientific astronomy, film has been long replaced with CCD sensors because of their superior performance.

-- Bert Na (bergna@yahoo.com), March 16, 2002.

Ive heard that the new digital lenses from Schneider and Rodenstock are better than their film counterparts necessarily because digital sensors require a better lens performance than film requires.The image circle is less but the overall lens performance is better in the digital versions.Some film users are very happy using the digital versions because of their extreme sharpness.

-- Emile de Leon (knightpeople@msn.com), March 16, 2002.

Dear Bert, could tell us the size of the actual ccd tablets used in astronomy. I think the importance of reproduction ratio to resolution is being missed here.

-- sait (akkirman@clear.net.nz), March 16, 2002.

Sait and Bert,

With regard to the sharpness of 4x5 lenses and their relative 35mm counterparts, I think you are absolutely right. I'm a photographer, and think that contrary to popular belief, there is a lot more 'slop' in 4x5. Just because Sinar makes a camera that encourages anal retentiveness doesn't make the format more precise. It follows that a Leica would have to be more exact in its manufacture in order to maintain a useable image. A front standard of a 4x5 could be slightly off without drastic image problems, but take the same amount of 'slop' in 35 and you'd have a big problem. I regularly use both formats for professional use and the sharpness of 4x5 has much more to do with the film size and its relative enlargement...

At any rate, I too am interested in the discussion of astrophotography and its digital aspects. Sorry I'm not a scientist here, just an image maker...

-- Carlin (carlinm@abac.com), March 16, 2002.

I believe there is only 1 reason why manufacturers are introducing new lenses for digital interchangeable-lens SLR cameras: the magnification factor of the digital sensor has created a need for affordable lenses wider than 20mm. Heretofore, 35mm-format lenses in the 14-17mm range were hideously expensive because they had to be designed with a corrected image circle large enough to cover the 24x36mm frame. Now, the manufacturers can make them with a much smaller usable image circle, which is less costly. You do not see lenses in the 35mm and up range being introduced as "digital lenses". However, given the image circle situation, and the obvious commitment of the major Japanese manufacturers like Canon and Nikon to digital, if my intended use was full-frame 35mm I would test any newly-introduced lens *very carfeully* before buying it.

-- Jay (infinitydt@aol.com), March 16, 2002.

CCDs commonly used in scientific astronomy typically have 13 to 15 um pixels and the array sizes range from 2048 x 2048 to 4096 x 4096 (there's also a 2048 x 4096 chip). The active areas are up to 31 x 31 mm for the 2k, and 61 x 61 mm for the 4k chip. In comparison, the current CCD on the Hubble is only 800 x 800 in array size, but the new Advanced Camera for Surveys instrument that was just recently installed on Hubble has a 4096 x 4096 active area that's based on two 2048 x 4096 CCDs.

-- Bert Na (bergna@yahoo.com), March 16, 2002.

Here's what I think is really funny. Two years ago on this very site are a number of posts saying things like "in two years film will be dead" -- "digital taking over" -- "we need a digital M". All of which kind of makes me believe that the posts in this thread, which say exactly what was being said two years ago (or seven years ago in PC Magazine) are, just as someone above stated 'farts in the wind'. Again I'll bring out my standard reply to the CD analogy - 20 years many were saying the synthesizer would revolutionize music and replace instruments. I go to our local symphony 4 times a year and have yet to see a damn synth on stage. Boy they must be behind the times!! Sure digital is coming. Sure it will replace a lot of film applications. But as I've said before - with a few billion rolls of film sold last year alone (mostly to Gramps and Grandma and old aunt Mable - who sure aren't racing to buy a computer or a digi-cam) film isn't going anywhere fast. To whoever above mentioned the lack of supply of the different types of film - more new films were introduced last year than old films discontinued - please check your facts.

-- Bob Todrick (bobtodrick@yahoo.com), March 16, 2002.

"...I was under the impression that a useable or effective grain size for film is about .0001 (1/10,000th) of an inch or the equivalent to a 150 megapixel image."

This may be true, but each of those grains/pixels has a color bit depth of one, on or off. A CCD gives color bit depths of 8, or 12. It takes a lot of little film grains to create the impression of continious tone, fewer CCD pixels.

I find this discussion interesting but ultimately pointless. I frequently visit a photo gallery which exhibits traditional and digital prints side by side. The digital prints look, for the most part, better, at least to my eyes. There are legitimate reasons for prefering the look of film, but theoretical arguments about pixels and grains don't change the fact that digital has passed film in many ways and is progressing faster. It's great that we can use both, and a pity that Leica seems to be left behind.

-- Masatoshi Yamamoto (masa@nifty.co.jp), March 16, 2002.

Huge numbers of digital users ,on hearing the rumour of the release of the sony moonraker digital banal,have joined the lemmings in Canada,where they have been confirmed as suitable members.

Get real ,digital is far from the quality of print.Read Love Digital Monthly even they will admit this fact.Take some real world photos,if you want a toy,try a toy shop.

-- allen herbert (allen1@btinternet.coma), March 16, 2002.