M'er's

I'm directing this to you because we believe we have excellent cameras, but are they too excellent? Please consider this. In the February issue of Digital Camera ( vol 5, issue 23) a columnist reviewing a Nikon Coolpix 5000 on p.39 makes this statement,

"... experts generally agree that standard 100 speed 35mm film has a nominal resolution of about 6 megapixels."

I figure that out to be about 2000dpi. I believed that my M6 combined with a 4000dpi scanner could produce about 24 megapixels of data.

( 35 mm film is approx. 1.5" by 1". At 4000dpi that's 6000 dots by 4000 dots or 24 million dots ( data points ) per frame.

If the reviewer is correct, our Leica systems are massive overkill. a 4000dpi scanner is translating approx. 1 "data point" on the film into 4 "data points" in the TIFF file which basically buys us nothing.

Is the reviewer correct about this film? Are all films this good (bad)?

My faith is not shaken, but there is a slight twitch.

Dan

-- Dan Roe (roedj@hotmail.com), April 12, 2002

Answers

I think it is the proverbial bovine feces.

-- Vijay A. Nebhrajani (vijay_nebhrajani@yahoo.com), April 12, 2002.

I have had 35mm slides scanned on high en drum scanners at resolutions of 50MB, TIFF, CMYK, with great quality. a lot more than a 6 megapixel digital file. But remember this 6 MPIX are RGB jpg's that grow to about 18MB CMYK uncompressed.

-- Marco (marco_hidalgo@hotmail.com), April 12, 2002.

I don't agree that the "resolving" power of 100ISO film is necessarly about 6 megapixels. I am not sure as to the size of one grain of B&W film, or the edge accutance of even large grain B&W or the size of dye clouds in color film or edge accutance of that. But with a little math, there is an argument that "resloving" power of ISO film is about 6 megapixels. If you assume that the "resolving" power of 100ISO film is 80 lines per millimeter, that means that there are 6400 "resolution" points in a square milimeter. As I recall, 35mm film cameras frame about 24 X 36 mm. That's 864 square millimeters per frame with 6400 "resolution" points each. That means, by this analysis, 5,529,600 "resolution" points or about 5.5 megapoints. The fallicy of this argument is that lines of resolution in film is made up of many more "points" than 80 per millimeter. If there were only 80 points per linear millimeter or 6400 points per square millimeter, then there would be visibly jagged diagonal lines in the image. So I really think that the resolving power of digital imaging would need to be much greater to match the resolving power of film. I have heard that the megapixel point where digital matches film is somewhere over 12 megapixels. That said, the 12.5 megapixel power of my Coolscan IV can produce some mighty sharp prints up to 12 X 15 or perhaps greater using algorithms of Genuine Fractals. In addition, film simply has a different depth and character than digital stuff. This is not a bad or good thing, just different. Where we will go on this - only the invisible hand of the market knows.

-- Doug Landrum (dflandrum@earthlink.net), April 12, 2002.

On the surface, 6 megapixels sounds reasonable for 35mm film if one makes certain assumptions. For discussion, let's assume that that the lens and film combination is capable of a resolution of 90 lines/mm. At an image size of 24mm x 36mm, that works out to 2,160 x 3240 pixels, or, about 6 megapixels.

The standard line target, however, is made up of black lines separated by white space. Hence, a minimum of two pixels are required to resolve the line - one black pixel for the line, and one white pixel for the adjoining space. Thus, the 24mm x 36mm image area would actually equate to 4,320 x 6,480 pixels at 90 lines/mm, or about 28 megapixels.

Lens resolution will, of course, vary from the 90 lpmm rating, as will the ability of the film to record what the lens projects.

-- Ralph Barker (rbarker@pacbell.net), April 12, 2002.

Scanning a 35 mm silde or neg into Photoshop using a Nikon LS4000 with Silverfast or Vuescan at HR setting (48 bit) produces a file of about 115 meg. Now that's a lot of data with which one can use for color correction and levels. I too have heard that 12 megapix is needed to equal 35 quality.

-- DM (dmaldonado@excite.com), April 12, 2002.

After I posted, I realized that the black lines have to be separated by white lines. Thus for a linear 80 lpm resolution (actually 80 line pairs per millimeter), you would need 160 pixels per mm. So my computation is too small by a factor of 4. That means you would need about 22 megapixels to match 35mm resolution at 80 lpm. I agree with Ralph's analysis. Although I have heard 12 megapixels as the match. That would be about 60 line pairs per mm - or about the resolution that a good autofucus system is geared to hone in on.

-- Doug Landrum (dflandrum@earthlink.net), April 12, 2002.

Ilford FP4 resolves 110 line pair/mm = 24 x 36 x 220 x 220 =42 megapixal

Agfapan APX 25 150 lpmm = 300 x 300 x24 x 36 = 78 megapixels

Kodak Technical Pan 320 lpmm

Even a finger nail sized Minox film 8mm x 11mm with technical Pan film has 640 x 640 x 8 x 11 = 36 megapixels

-- martin tai (martin.tai@capcanada.com), April 12, 2002.

I believe Martin has it correct. Since we all (probably) use decent film in our fine cameras, we can expect to resolve about 200 (just a median number) lines per mm. 90 and 100 seems really low. That's some crap film. 150 - 220 is good. The resolution required to match this is very very high. I can't imagine at all where that 6mp number came from. Sounds like something just plain made up. In order to record the same amount of information contained in a 35mm neg/slid, you'd need much more than 6mp. A hell of alot more. My numbers were like Martin's last time I worked this out. So don't worry Dan. Even when they have 50mp 35mm cameras, I bet some film and a Leica will still take a hell of a better photo. No need to twitch the faith. And don't read that mag! It sounds very poor.

-Ramy

-- Ramy (rsadek@cs.oberlin.edu), April 13, 2002.

Ummm. If you allow 3 bytes (that's only 8 bits) per color channel, you need to multiply by 3. 12 bits (more common in high quality scans) multiply by 4.5. And some schemes use 4 channels, not 3. I fugure. Figure about 78 Megabytes per frame at 150 lines per milimeter, for 4 channel, 12 bits per channel film. It'll happen, but give it a few decades.

-- Tom bryant (boffin@gts.net), April 13, 2002.

I read somewhere, can't remember where exactly, that the resolution of a professional slide film goes towards 1B (yes Billion) pixels.

The 5M pixel camera's provide pretty good results upto maybe A3 format. So if that is the max size you're using, I see no reason not to consider digital. But before you go digital also consider facts like response time, battery consumption, cost, etc etc. As we all have discussed many time before. As for now I still see no reason to trade in my R4s and lovely "new" Hexar AF.

Reinier

-- ReinierV (rvlaam@xs4all.nl), April 13, 2002.

1 billion pixels in professional slide film? That's just 1000 Megapixels. Within reach of Velvia in 8x10, but certainly not available in 35mm due to the wavelengths involved.

Things look a bit different at contrasts lower than 1000:1. At 1.6:1 and similar realistic values, most films can't resolve more than 100 lp/mm, and the wide majority is around 40-50 lp/mm. Still, to say that 'standard 100 speed 35mm film has a nominal resolution of about 6 megapixels' is to lie. As the 1.3 Mp digicams some years back rendered `photorealistic' quality', shouldn't we say that ISO 100 film has a nominal resolution of about 1.3 Mp ;-) ?

-- Oliver Schrinner (piraya@hispavista.com), April 15, 2002.

Kodak supposedly enlarged some 35mm Kodachrome frames to mural size in Grand Central. My question is, how would 6MP digital image look at that size?

I think it will still take a long time before digital reaches film quality because in my mind digital is good only up to a certain level of enlargement, even better than film for lack of grain. But when you overdo it, film is still very much better. I suppose our eyes are somehow used to large grainy images, but over-enlarged 'digital grain' is much harder to accept.

Just my view.

Ilkka

-- Ilkka (ikuu65@hotmail.com), April 15, 2002.