View Full Version : Does larger digital file = greater amount of useful data???
There has been much discussion on this forum for Digital vs. film. Below is a l ink to a very reputable site which compares the new Canon D30 digital camera wit h Provia F film.
I was amazed at what I read, and now am re thinking how digital my impac t our future. The part of this review, (and others on the same product) is the file size comparison. For example, a Canon shot taken with the D30 produced a 9 MB file, however the same shot taken with Provia F and scanned at 4000 dpi on a Imacon scanner produced a file size of 34 MB. Conventional wisdom would tell u s the scanned file with almost 4x more data would be noticeably sharper. Howeve r, all the test I read so far revealed the same thing, both prints were equally as sharp!! Or if anything the one taken with the digital camera were even sharp er?
So the looming question on everyones mind now is... does larger files nec essarily equate to more useful information for producing higher quality prints? Why do these film scanned files contain so much extra data - but does not seem to translate this data to the actual printed photograph?
Considering the price of this technology, about $3k usa for the D30 cam era, it may not be long till digital moves faster than we all thought into large r formats. This assumes the relationship between file size (or image capture si ze) is not the benchmark of quality we all once thought. Any ideas of why the s canned film files are so much larger but do not provide a superior end product t hat one would suspect?
In this test, the smaller D30 file was ressed up using software (G Fracto ls I think) to match the size of scanned file...however, I am doubting that is t he missing link, if so, who needs LF files to begin with if we can just ress up smaller digital files.
Richard S. Ross
I think it is a given that leading edge digital image capture is already very close to 35mm film. CCDs are slowly giving way to Si based photodetectors (CMOS technology) as demonstrated by Foveon (http://www.foveon.net/) who have a square 16Mpixel device available. You of course pay the price in storage and energy requirements. It will be a while before these devices become available in larger formats but it will happen. It will probably be longer, however, before the storage and power requirements are brought under control with dense magnetic RAM (or some comparable technology) that can replace disc drives and operate on a few AA's.
As for "ressing up smaller digital files", which I assume means some kind of interpolation (?), forget it. You don't get more resolution this way, you just get bigger mushier images.
For LF field photographers, digital is quite far away. Not so far for studio though. Lets just hope the film companies don't quit too early!
I believe the pro one shot backs (e.g. Phase One) have image sensors the size of 35mm film currently. I've seen some stunning 16x20 prints from 16 Mpixel one shot captures. Even with this smallish capture area, I've seen a pro mount a Phase One on a 6x9 view camera and do product shots. He owns a scanning back, but is finding that 16 Mpixels does the job most of the time.
I would contend that digital capture is absolutely here for much studio based work. For that matter, I know several local pros who travel with their digital rig on location. They setup their computer on a cart, and do their work. They are getting new clients because they are all digital. By being rigorously color managed, they have more control of their image all the way to press. For small prints (up to 8x10) they are using some local minilabs who have digital front ends (e.g. Fuji Frontier). They will typically build color profiles for these minilabs. Many of these commercial clients only want digital files as the end product anyway. The big problem is that it's a significant capital outlay (or lease expense) at this point. Makes it tough for the person getting started in business.
A first explanation would be that in spite of his claims, there is somewhere a weaker link in the chain leading to the print. I'm sure that someone could argue that with a 100M scan on a Tango and a lightjet output, differences might become visible.
A second explanation of the paradox might be that a lot of image information is redundant. Somehow this information is reconstructed adequately from the smaller file by software. Think about a photograph of a flat dark square against a light background. All you need to do to obtain a hi-res image from a small file would be to do some kind of smoothing of the edge.
I've said it before, and I'll repeat it now.Comparing the grain image of film with a pixel image is like comparing apples with bananas.A film grain image is actually more digital in nature than a pixel. Grain, and/or colour film dye, is either 'on' (developed), or 'off'. There are no in betweeny colours with film; they must be made up from aggregations of silver or dye speckles.Now, even an 8 bit pixel can have 255 different levels of grey, or 16.7 million colours with 24 bits. So a given area of film must contain 255 'grains' to represent the same range of tonal values as an area of 1 pixel.If you do some real measurements of film grain or dye cloud size, and compare those with current pixel sizes, it turns out that a pixel size of 5 microns, (such as is common in consumer digicams) has a greater image capacity per unit area than all but the finest grain films on the market.All that needs to be done is to make larger area CCD sensors.
If you extend that to current state-of-the-art sensors, then area for area, CCD or CMOS devices are capable of better image quality than commercially available film.Why do you think that most astronomers now have CCD sensors strapped to their telescopes, and not film cameras?
Joshua L. Slocum
Pete: You forget that those "255 grains" that film "must have" to equate a pixel are usually considerably smaller than the pixel. That is why film is, often, (not in all cases), more continuous tone in nature than digital. High end cameras not withstanding- and I'm sorry, but I don't think even 16 mp is going to make me give up Kodachrome 25 or my medium/large format gear- digital just doesn't cut the mustard for me. Please don't accuse me of not knowing my facts- I've seen the digital prints from the D1, the Kodak 6mps and Michael's Luminous Landscape page. Strangely, the Provia images still look sharper to me. The bottom line, I guess (and I'm guilty of this too) is that we can argue equipment all we want, but the results count. Having said that, I HAVE YET to see someone show me a digital pic that I cannot match or exceed in resolution or tonal quality with any of my film gear. It's true, digital is a lot closer to 35mm than it used to be, and in high end cases can exceed it, but 35mm is damned good stuff. . people seem to forget that, even though last year they were selling their prints like hotcakes to satisfied customers.
Larry: One question. . why are all these pros "getting more customer because they're completely digital?" Don't tell me it's because clients truly understand the image process-if they did, they'd know that any medium or large format image would exceed their quality standards anyway. I'm almost positive it's because of the magical aura around the "D" word, and has no basis in real understanding of the photographic process. If they did understand it, they'd know it was possible to A) shoot the highest quality film images. B) Make top notch optical prints capable of satisfying the most crucial assignments (it's been done for more than 100 years before digital) C) Develop and scan the film in under an hour, thereby negating the argument that only digital cameras can deliver electronic files. You see what I mean? How many clients, except for the NY TIMES, absolutely must have the finished image immediately and cannot afford one hour's processing time???
I guess it's my basic loathing of the fact that clients are so uninformed and so easily marketed by hype that separates me from the "pros."
I'm not shooting digital (yet), so I can only give you the info as I understand it from some who are shooting this way. One question - if the client is going to press, and needs a quality digital file as part of this process, how is it advantageous to shoot film as an intermediate step (assuming your shooting something where digital is an appropriate tool)? It does provide a tangable reference to the customer, but that is far from the end product being the image in print.
Control (and revenue) of a greater chunk of the process seems to be the first advantage. From what I've seen from these folks, the RGB to CMYK conversion is non-trivial. It appears to be one of the causes for "good" transparencies to print poorly. By being involved from capture to print, the photographer can assure the integrity of their image. These photographers will color profile the printing press being used, on paper stock being specified and target a RGB to CMYK conversion which will yield the best results. The client probably likes the fact that photographs are printing more realistically, and that one person is taking responsibility for that result. That service is of great value to the customer, and they are willing to pay for it. I've also heard that clients like seeing the results immediately on the screen during the shoot. MUCH better than any Polaroid could ever show. Small tweeks can be made in real time.
Shooting film and scanning are also important parts of being digital. As has been mentioned by others, digital capture has a set of limitations where film is often the better choice. However, as a process, I don't understand where your coming from. Quality film scanning is more expensive in time and money than quality digital capture.
Regarding the time equation... The best E-6 labs in this area can get the film back in about 2 hrs with a rush charge. Normal turnaround is closer to 3 to 4 hours. Where do you get 1 hr E-6? Professional C-41 processing seems to take even longer than E-6 (except at minilabs). I've yet to hear of anyone who prefers scanning negs rather than chromes anyway. Professional scanning services are also on a day or two turn around (without rush charges). I believe these clients are often under strict deadlines, so the extra couple days for shooting film + scanning is real money to these folks. The other interesting thing I've observed is that some of the early pro adopters seem to be extremely fine photographers who are located in relatively remote areas. For these folks, the closest professional E- 6 lab might be 2 or 3 hours away. From a practical standpoint, they overnight mail their jobs, so on just film, they are really 2 days for first results. If holdback sheets need to be processed, the first good film might be 4 or 5 days from the shoot. Digital shooting seems to be a big win for these people.
"Any ideas of why the scanned film files are so much larger but do not provide a superior end product that one would suspect? "
Bill, I don't understand. A digital picture is a mosaic of pixels. If you have a mosaic with 400 and the same with 1600 pieces, and you look at them from a distance, they may appear just the same . But when you walk closer you will soon realize one contains many more elements than the other. Similarly the 9 and 34 MB images may look alike when printed at a size of 8x10", but if you enlarge at say 12x16", th ere should be a difference, at least this is what I think. I have read good reports on Genuine Fractals softwar e capability of upsizing an image (and downsizing as well). However I doubt it can transform a 35 mm image into a high definition 4x5 image. To continue with the mosaic comparison, it would be like a mere guess reconstitutio n of a damaged mosaic where parts were missing, compared to the original artist work.
Or shall I sell my LF gear before it's value drops and get the D30? Of course I 'm kidding.
Josh. The problem is that film grains aren't considerably smaller than a pixel.I've recently had cause to measure the 'dye clouds' of colour film, sad man that I am, and even the finest came out at around a micron diameter. Now even if these clods of dye were laid out in an orderly manner (which they ain't), there would still only be 25 of them in the same area as a consumer digicam pixel. Well, OK; there'd be 75 in the depth of the CMY layers, that's still a few short of the 765 grains needed to emulate the colour depth of the pixel.Think about it, a consumer digital camera has a sensor size of about 6mm x 8mm. What sort of quality can you get from an area of film that small?
There is a point of "no greater return" because the device printing the image doesn't output more "information" than a certain amount- feed it more and it just throughs it out. Sometimes it just makes it worse. Yes, the larger you print, the more information is needed.
How do the comparisons look on something like a 20x24 Ilfochrome print? Or on a Fuji Supergloss print of that size? I am sure digital will continue to get better and pixelographers will be coming out of the woodwork luring hot babes into taking their clothes off to model for these guys... some things will never change. Mediocre images will still be mediocre images whether taken with a pixelrecording device or an 8x10 view camera. For now we know silver based negs will last and silver based prints will last & many alt process prints will last. We **think**, based on best guesstimates & accelerated image tests, that pixelography images will also last. But, just as the odd fading with Epson prints suddenly showed up & Ozone was the culprit, who is to say that the pixelographer covering the nude jello wrestlers won't find his images disappearing before his eyes due to the electrical field of the interaction with buttcrack sweat & the green dye in the jell-0? Digital stuff has a bright future but for now it is a "just as good as" game. Given the choice, I like the original better. When "just as good as" goes away & pixelography becomes its own medium and quits trying to be just a copy of what we already have I will be much more excited by it. After all, Madonna is "just as good as" Marilyn, but having seen Marilyn I will keep her memory rather than the imitation available now.
The argument that you need 255 grains to represent 255 shades of grey is pure bullshit. Each grain is efectively one bit of infprmation - it can represent a 1 or a 0 (either its there or it isn't there) sor you only need 8 grains (or there abscence) to represent 255 shades of grey. In addition the positioning of these grains is not locked into a grid as in a CCD sensor.
Powered by vBulletin® Version 4.2.1 Copyright © 2013 vBulletin Solutions, Inc. All rights reserved.