A Site for Photographers by Photographers

Community > Forums > photo.net > Digital Imaging > How many pixel are equivalent...

How many pixel are equivalent to 35 mm film

Sandra S , Aug 24, 2000; 01:22 a.m.

I had a discussion with a friend yesterday and he said if DI reaches 9 million pixel it will have the same quality than a fine grain 35 mm film (let´s assume Velvia). What do you think, is this true? I don´t want to start a DI-traditional photography war here, I´m just interested in verifying his statement (I don´t do any DI and I don´t intend to).




    1   |   2   |   3   |   4     Next    Last

Michael Hohner , Aug 24, 2000; 01:28 a.m.

A good film (as Velvia) and good lenses can resolve about 120 lp/mm, i.e. 240 pixels/mm. This means a 35mm frame has a resolution of 36*240*24*240=4976640 (almost 50 MPixels). You can use Moore's Law to estimate when this resolution will be reached. My estimation is 4 to 5 years.

Michael Hohner , Aug 24, 2000; 01:30 a.m.

Typo: The final number should be 49766400, i.e. really 50 MPixels

Mark Wilkins , Aug 24, 2000; 01:37 a.m.

A common resolution to scan 35mm color stills for motion picture visual effects work is 4092 x 2728, which is about eleven million pixels. However, many slow B/W films can achieve effective resolutions much higher than that. Also, even with color images, a 4K scan is not held to be the extreme upper bound in quality, just a good compromise between size and subtle digital artifacts for critical work.

A bigger issue for the hypothetical 9 megapixel camera is that it would have to measure at least 12 bits per channel to come near the dynamic range of film.

An eleven megapixel digital camera with twelve-bit-per-channel output that looked like the color space of film might just get me to go all-digital.

Daniel Lakeland , Aug 24, 2000; 01:50 a.m.

You've already heard what the number of pixels is in terms of lp/mm, (about 50M) here's the next set of problems for DI.

Digital camera sensors currently only sense one color per pixel, and at 8 bits per pixel. Film has CMY info at all points on the film. Slide scanners scan all colors per pixel, but in general for a digital camera (ie. no film involved) you will have to create a sensor with all 3 RGB color sensors at 12 bits per sensor to get the same color information as film.

THEN... you have to deal with diffraction limitations. The point spread function of the digital optics has to be 1 pixel wide, otherwise your extra pixels are just giving you redundant information. (in other words, a point in your subject gets spread out by diffraction to be a little blob on the film (or sensor), which means that you can't just make smaller optics in proportion to tiny silicon wafers, unless you can make the optics sharper.

If you assume that 35mm resolution is optics limited (ie. that the film itself can resolve more lp/mm than the optics can) then you will need a CCD sensor the exact size of a 35mm frame before having this many pixels actually helps your resolution. The problem is that the probability of a flaw destroying your CCD in manufacturing is basically proportional to the area, so the yeild is like 1/A, and costs go up in proportion to the area. The real question is when can they manufacture a CCD with 150M or so pixels, with 12 bits per pixel of sensitivity, that covers 24x36mm of physical silicon wafer without significant flaws at low enough cost to make it feasible to sell to consumers? My guess is not in the next 20 years.

Oh, and then I haven't even mentioned charge bleed between pixels (when one pixel gets charged up, it leaks charge to adjacent pixels reducing effective resolution, and blurring your image slightly).

Then what would you do with the approximately 450 megabyte image file you would get from EACH SHOT???

George Hager , Aug 24, 2000; 02:23 a.m.

I just followed a post on another group by an optical physisist for a large internation electronics firm. He concluded about the same as the earlier posts. It would take somewhere around 405 Megs of pixels to match film. I didn't follow everthing he said, but most of it seemed very reasonable. Maybe some secret new technology will come to our aid but it doesn't look too good right now. (Use of a beam splitter like in high end video and three chips could help.)

Hey, on the other hand, if you are just making 4x5 prints and stuff for the web (like 99% of the people), what is available is more than good enough. And it is a lot of fun to play with.

Joseph Albert , Aug 24, 2000; 03:03 a.m.

Take a look at my contribution to the end of this thread, for a discussion of another problem that needs to be solved by digital photography technology, and why your current optics may not be that useful if and when CCDs get as large as current popular film formats.

I have no expertise in this area, the above-referenced posting was just a re-hash of stuff I read in an article, so if anyone has more detailed input, I'd welcome it.

Mark Horton , Aug 24, 2000; 04:42 a.m.

Since you ask how many pixel are equivalent to 35mm film would it fair to answer the question based upon what we perceive visually rather than pixel count vs. film grain? If so, then the following news release from Hewlett-Packard Laboratories may be of interest to those of us that are digitally inclined. "For the first time, digital photography will be superior to film,"

Joseph Albert , Aug 24, 2000; 05:37 a.m.

The HP press release is certainly interesting, but it appears to be mroe of a point&shoot "what you get is what the camera gives you" type of tool rather than an artistic tool at the control of the person behind the camera. If this technology is successful and percolates into serious cameras, I can see that they will be pretty complex to operate.

Greg Poulsen , Aug 24, 2000; 10:01 a.m.

All the above are interesting points, but there is one other point that is becoming increasingly important. We don't view 35mm material directly very often. The quality that we see is limited by our display technology, which introduces additional artifacts into the final image (whether it is a slide projector and lens, the enlarger lamphouse and lens, the scanner, or whatever). Digital images have the advantage of eliminating much of the "additive error" problems of analogue processes. This is ultimately the thing that makes digital audio and video so pervasive now.

So, although 35mm film and lenses have a very large potential information content, the processing, transferring, storing, duplicating, and display preparation all take a profound toll on analogue images. This will make the "real-world" point of parity for digital images MUCH lower. IMHO, good digital images (with 12 bits of depth per channel) of 60-80 megabytes (not megapixels) are fully equivalent to any 35mm in terms of useful information. By the way, a good 80 meg scan of a 4x5 neg blows away the info on any 35mm.

Best, GP

    1   |   2   |   3   |   4     Next    Last

Back to top

Notify me of Responses