D30 Vs. Provia 100F
This subject is now featured inVolume 1, Number 1of The Luminous Landscape Video Journal.
The Test and The Methodology
Before you can evaluate these results for yourself, and understandmyinterpretation, you need to understand the methodology that I used and what I was trying to achieve.
For the past year whenever I’ve seenNikon D1output I’ve usually been favourably impressed, but no one has ever been able to show me a side-by-side comparison between it and film. Therefore one of the tests that I wanted to do more than any other for myself when theCanon D30arrived was to mount the same lens and shoot the same subject and draw my own conclusion.
The images you see below were taken on a moderately bright overcast day. They show the skyline of Toronto and were both taken with the Canon 100~400mm f/4.5L lens. The two cameras used were the digitalCanon EOS D30at ISO 100, and the other was theCanon EOS 1VusingFuji Provia 100F. I used this film for the test because it is arguably the finest grained, sharpest ISO 100 speed film available. It also is the film that I use 95% of the time.
The lens was mounted on a tripod and the bodies were interchanged. The lens was zoomed to accommodate the difference in magnification ratio between the two system (1: 1.6). As you can see, the film-based image is slightly larger. I wasn’t as close in matching as I’d wanted to be. But since the advantage is to film, if there’s any bias it’s in film’s favour.
I was obviously concerned that I not introduce any unnecessary biases into the comparison since the film would have to be scanned, and both it and the D30’s output processed inPhotoShop. The scan was done using anImacon FlexTight Photoat 3200 DPI. This produced a 34MB file. (The Imacon is one of the highest quality scanners available and retails for USD $10,000). The scan was done using the software’s default settings and with no sharpening. Only anAuto Levelswas done.
The D30 image was shot in RAW mode and therefore noWhite Balance, colour enhancements orSharpeningwere done in the camera or prior to loading into PhotoShop. A roughly 9MB file was produced.
Both images were loaded into PhotoShop and Unsharp Masking was done usingNIC Sharpener ProinAutomode. I find that this program produces superior results to using PhotoShop’s Unsharp Mask function and it also allowed me to be removed from making subjective sharpening decisions.
Again, to remove subjectivity as much as possible the only other adjustment made to either image was to use PhotoShop’sLevelsinAutomode.
It should be noted that both images are shown here at 6" wide (9" wide on their stand-alone pages) and that the Provia image is slightly higher because 35mm film has a different aspect ratio to the D30’s imaging chip.
So that’s the methodology. It seems to me to have removed as many subjective variables as possible and even weighted the comparison slightly in film’s favour because of the slightly larger magnification (I’m a photographer, not a scientist). Finally, I should mention that the evaluations below have been made by looking at 8.5" X 11" and 13" X 19" prints as well as on-screen. Critical evaluations were also performed by my friend and associate Chris Sanderson an award winning cinematographer and film director. (We agreed in every aspect).
Both images taken with the Canon 100~400mm f/4.5L IS zoom at 1/250th sec @f/5.6
Canon EOS D30 Canon EOS 1V on Provia 100F
D30 Detail Provia 100F Detail
And The Winner Is?
No one will be more amazed at the conclusion than I was. The D30’s digital image actually was better in almost every respect.
The first thing I looked at was, of course, resolution. As can be seen by the detail blow-ups immediately above, the D30 image shows finer detail. It easily does so on an 8X10" print as well. Look at the central row of windows and also at the windows in the brownish building at the left.
Though Provia 100F is rated as an extremely fine-grained film at this magnification the grain is just starting to show. It can be seen in the smooth areas of the central light blue building. The D30 has no grain of course, and even in an 8" X 10" print look much smoother as a consequence. In fact it’s only because the D30 image has absolutelynograin that any is discernable in the Provia print.
In both prints the white tennis dome in the lower right hand corner displays identical clean detailed white. (The prints show this much better than the low-res JPG images shown here; trust me). Looking at the shadow details it appears to me that there is about a half stop more detail in the D30 image. I hope to do more definitive testing in this area soon.
This was possibly the biggest shocker. I’ve been shooting Provia 100F and its predecessor Provia 100 for years. I’ve always regarded it as a moderately low contrast, neutrally balanced film. I know its characteristics and have done many hundreds of scans from it.
The D30 blows it away in colour accuracy. I know that this is hard to believe, but in this test at least (soft light) the colours are richer and truer from the D30. If you know Toronto you’ll recognize the gold coloured building on the far left. It’s theRoyal Bankbuilding and it’s colour is very distinctive (they used real gold in the glass). In the D30 print the colour is exactly as I visualize it, while in the Provia 100F print it is less saturated and inaccurate.
You’ll also notice the reddish wall of the warehouse and the red brick of the building just above the tennis dome. Again these are more saturated and closer to reality than the Provia image.
D30 Detail Provia 100F Detail
Here are sections froma larger image(on the main review page) that again shows that the D30 is in fact sharper than Provia 100F on 8X10" prints. These 72 dpi JPGs, even though I’ve posted them here with minimum compression, may not show this very well, but the difference jumps out on viewing a print.
Again there is better detail in the shadow areas and the colours from the D30 have a broader and more subtle range of tonalities.
I was not prepared for this result. While I expected that the D30 would account itself well I never anticipated that it would actually produce an image that in most ways is superior to film. I’m drawn to the unavoidable conclusion that the Canon D30, when shooting in RAW mode, is able to produce comparable images to Provia 100F scanned on a high-end scanner. Now, ain’t that a surprise?
Where Did The Extra Resolution Go?
This is a puzzler. The D30 produces a 9MB file. The 3200 DPI scan from the Imacon a 34MB file. Common wisdom has it that a scan of somewhere between 3000 and 4000 dpi will capture just about all of the information there is on ISO 100 transparency film. How then can a file that’s almost 4 times smaller produce a print that subjectively, at least, appears as sharp?
I don’t know, but I do know what I see. I’ve made prints of the above images on 13" X 19" paper on anEpson 1270printer usingPictorico Hi-Gloss FilmÃ¢â‚¬Â¹ the sharpest paper available. Both images are roughly 11X15" (allowing for the different aspect ratios) and were both printed at 264 DPI. (The D30 image was ressed up using Genuine Fractals Pro to match the Provia scan’s size). Besides its lack of grain and superior colour saturation and shadow detail Ã¢â‚¬Â¹ none of which are affected by image size, the D30 photograph looks essentially as sharp as the film scan even though at this size the resolution is interpolated rather than actual. The results speak for themselves.
I’ve been writing for magazines and technical journals for as long as I’ve been a photographer; more than 30 years. In all that time this is one of the most controversial opinions that I’ve published so I’m concerned that I understand what it is that I’m seeing. After spending a number of additional hours examining images on screen and on paper, and sharing the evaluation with both pros and laymen, I’ve come to the following preliminary conclusion.
It is inescapable that the D30 produces sharper, better looking images than the scanned film combination at sizes up to about 10 X 13". Larger than this 35mm wins, but it isn’t till above 11X15" or so that this starts to become obvious. Most lay observers can’t see the difference. (The D30 image is, of course, ressed-up inGenuine Fractalsfor sizes above 6X9").
What appears to be happening is that the degree of superiority over the film/scanner combination is about 20%. It is only when the D30 image is res-up beyond this amount that the playing field levels.
I have received quite a number of emails and seen message board comments calling my conclusions into doubt, for various reasons. Let me start by saying that I’m the first to admit that I’m a photographer, not a scientist. But, I do understand the scientific method. (I have 8 granted U.S. patents in the field of telecommunications and computer interface methodology, so I have some experience in this area). Having said that I will also add that all that I care about is how an images look on a print. As a fine-art landscape and nature photographer what counts most for me is what will appear on gallery and purchaser’s walls. Theory is secondary.
One comment that I’ve seen is that my comparison is flawed because my prints were made on anEpson 1270inkjet printer and that if they had been Cibachromes done on an enlarger the difference would have been the other way. I disagree. I have been a Cibachrome / Ilfochrome printer for 25 years and have taught workshops and written articles on it. I closed my darkroom 2 years ago because I believe inkjet prints to be superior in almost every respect. Many professional and fine-arts photographers believe similarly. Ciba prints Vs. inkjets is an old debate that I’m no longer interested in.
Another reason to dispute this position is that the superiority of the D30 image is clearly visible on-screen, even before a print is made. This visible difference then carries over to prints. In any event what we are doing here is a comparison, not a measure of absolute goodness.
Another objection is to the scanner used. Again, this is a pointless debate. TheImacon Flextight Photois one of the most highly regarded scanners on the market and is used extensively around the world as a viable alternative to commercial lab drum scanners. Is a better scan possible? Yes, almost certainly. Will 99% of all photographers ever have such a scan made? Unlikely. Is the Imacon better than almost every other desktop scanner under USD $10,000. Most would agree.
The point that I wish to stress again is that what I have attempted to do is compare D30 output with a well scanned transparency. Not to see which can be blown up to 30 feet. Not to compare Ciba to inkjet. Not to argue the merits of various scanners, printing papers or other methodologies. My tests, evaluations and opinions are done under conditions as good as if not slightly better than most photographers would use if they were doing them. I’ll leave absolute scientific rigor and exactitude to others more qualified and especially those more interested in such minutia. As for me, I’m going out to take pictures.
Don’t forget to also read myfull reviewof the D30 as well as other related articles.
There has been much heated discussion on various forums about these results. Because of the limitations of web display to a large extent one has to take my word about many of these conclusions.
One contributor to the Canon Forum onDPReviewhas posted the following comment;
Just back from Vistek in Toronto where I saw first hand Michael Reichmann’s
infamous prints; the ones discussed in the ‘digital vs film’ thread.
They are all you need to see to confirm the significance of digital and the D30.
They are unbelievable. Big beautiful prints,17 inches by whatever. There’s a
print of the fall leaves, (the one he posted in his comparison), that is so
sharp, so clean and the colors are so good that no professional
photographer could ask for more out of an image. Forget film versus digital
and just look at them as photographs. I guarantee you will say, " I want
whatever it was that took that picture and whatever made that print.".
Too bad they don’t sell the talent of someone who can actually take those kind
I appreciate the compliments and hopefully this will add some third-party objectivity to the discussion. If you are inTorontoyou are welcome to visitVistekon Queen St. East and ask to seeBrian Georgein the second floor digital department. He’ll be pleased to show you the prints.
Brian Taoof Toronto has visited the dealer where I bought my D30 and has evaluated the prints, and compared them with my conclusions. Brian has published his commentshere.
The only thing that I would add to what Brian and others see atVistekis that these prints are roughly 12 X 16" images on 13 X 19" paper. If you look at 7 X 10" prints theD30is visibly superior, even under a loupe. What we’re loosing in the larger prints is inevitable no matter how good a jobGenuine Fractalsdoes in ressing-up a file. Next year’s 6 Megapixel camera with theCanon’sCMOS chip and electronics will definitively push digital over the top.
October 27, 2000
As this is written I have had theCanon D30for just 4 days. This page has been online for less than 3 days. During that time it has received more than 16,000 visitors Ã¢â‚¬Â¹ which I find rather astonishing. I suppose it shows the incredible interest that the photographic community has in this remarkable camera.
Also to my amazements there has been a huge amount of discussion on various forums and message boards around the world debating my findings. Most of it has been thoughtful and intelligent, some of it has been virulent, and as always a small amount has been just plain misinformed. I have also received close to 300 private emails on the subject during this time and have done my best to respond to them all.
Since keeping up with the ongoing discussions is an impossibility, and I’m obviously drowning in email, I’ll use this, my own soapbox, to expand my perspective a bit and hopefully to serve as a reference point for various questions and critiques that I’ve received.
Is the D30 better that film?
What do I mean by "better" and what do I mean by "film". That’s what the debate is raging on about. It also has to do with my testing methodology. Why didn’t I use a different lens? (I have). Why didn’t I compare to a traditional chemical print? (Because digital prints are better). Should I have used a smaller aperture? (I have in other tests). Did I use mirror lock-up? (Yes, in other tests). What type of loupe did I use? (Schneider MC 4X Aspheric). Why didn’t I have a drum scan made? (Because for this purpose theImaconis just as good).
One more thing. Some have said that what I’m comparing theD30to is the combination of filmandscanner. Yup. That’s what I’m doing. Arguably the best desktop scanner on the market and the sharpest finest-grained ISO 100 transparency film. This is what I use. This is what I’m comparing to. I’m not seeking absolutes. I’m trying to determine how what I’ve been doing till now compares with this new method. Anything else would be meaningless Ã¢â‚¬Â¹ for me at least.
I’ve said this before elsewhere, but it’s worth repeating for anyone coming to the party late. I’m a pretty good photographer. I’ve been at this for over 30 years and have made my living at it for many of those years. I’ve been writing articles for magazines and technical journals for most of that time. I’m a decent darkroom worker and have taught courses in Cibachrome/Ilfochrome printing. I’m a competent digital image processing person. I write articles on PhotoShop technique and teach its use by photographers. I’m using state-of-the-art cameras, lenses, scanners, printers, computers and software.
Why am I telling you this? Because I lack the ability to sit down and show you my prints personally. You can see some of what I see on the relatively low (roughly 72 dpi) resolution of your monitor, but I have no idea what you’re seeing because of this limitation and other variables. I do know though whatI’mseeing. I also know thateveryonewho I’ve shown my prints to thus far, layman and photographers agrees with my evaluation.
All this is lead-up to my current opinion; unvarnished and without reservation.Prints that I can make from D30 images are better than the prints that I can make from 35mm film.Period.
TheD30, whether because of its unique CMOS imaging chip, electronics, software, magic fairy dust or whatever, allows me to to make prints that are superior in apparent resolution, colour saturation, shadow detail and overall image quality than what I am able to produce from film. ( I should also make clear thatD30images are the first digital camera images that to my eye don’t look digital. They look like film.)
Quite a statement, isn’t it? But, after just 4 days and a few hundred test images, that’s my opinion. Others may find differently. So be it. But I can only tell you that if I were to head into the field today with my Canon gear and wanted to come back with an image from which I could make a print for gallery exhibit or sale, in a size up to about 10 X 15", I would use theD30in preference to theEOS-1Vusing Provia 100F. That’s where, as the current saying goes, the rubber meets the road.
Will further experience and testing change that opinion? Possibly. As for others, they’ll have to make this evaluation for themselves.
October 30, 2000
Harold Merklinger, a well regarded expert in the field of optics and author of several books on the subject has weighed in with his opinion on this issue. I have reproduced this comments onthis page.
December 26, 2000
I have now shot more than 2,000 frames with theCanon D30in the past 2 months. Whereas my observations above were based on a few hundred frames and less that a week’s use, I have now used the camera extensively on assignment in the field.Please see my portfolio fromIndian Country.
With this experience under my belt I can say without reservation that my initial impressions hold firm. I can produce higher quality images from theD30than I can from 35mm scanned film (using the best desk-top scanner available).
As far as I’m concerned the case is closed. But, if you have a cogent case to make otherwise or have had a different experience, I’d love to hear from you.Michael
Over the past several months I have received numerous emails from photographers who have purchased and now use the D30. Not a single one has contradicted my findings. Here is one such letter…
I am a portrait photographer who’s been in the trenches for over 25 years, and would like to comment on your D-30 review. You state that you have concluded you can get a better 11×14 (approx.) landscape from the D-30 than you can get from Provia.
I bought a D-30 myself and am in the process of comparing 11×14 prints from it to 11×14 prints from my favorite color negative film, Fuji Reala. Skin tones are a particular problem for all photographic materials. The problem is not just overall color rendition, that is relatively simple to fix, if annoying. The problem is that there are subtle variations in color that must also be rendered faithfully for the whole to be believeable.
I’ve made the prints, and will be showing them to colleagues for their opinions, but I’ll tell you mine now: I like the D-30 images better. To my eye, the film prints look muddy in comparison. It’s the grain, of course.
– Michael Corbin –
The following letter sheds some additional light on the subject.
On the subject of digital vs. film, "quality" pixels: I can’t remember the title or author, but remember reading a book which gave resolution figures for commercially available film. As I remember, the highest claimed resolution was for Kodak Tech Pan (exposed ASA 25 and processed in Technidol) said to be 200 lines/mm. The highest claimed resolution for color film was way less, less than 100 lines/mm. More typical films, like ASA 100, 200, or 400 color film had claimed resolutions like 50-60 lines/mm. As another point of reference, my HP S20 scanner (which is very poor, by the way), scanning 35mm color negatives at 2400 dpi, easily resolves the grain itself on Fuji ASA 400 film. My experience (admittedly limited) leads me to believe that typical consumer/prosumer films around ASA 100 might be capable of a maximum resolution of 1250-2000 lines/inch or 2.3M – 6M "pixels" per 35mm frame. I have no way to test this, but Tech Pan should be capable of about 38M "pixels" by the same logic. I note that this is similar to the figure given by another of your correspondents, Harold M. Merklinger. One way of thinking about the relationship between digital pixels and film "pixels: Suppose we were to use the resolution figures given for film as a model for "pixels". In this model, the grain effects are noise. This noise, being essentially random and molecular in scale, is in some sense analog noise superimposed on our signal "pixels". Digital noise is fundamentally different, and follows the same pattern with any digital signal, the value of a pixel is absolutely different from the real value being represented. The digital values have been quantized, noise being "incorporated" into the signal, and some new "quantization" noise having been introduced (reality is not a step function). Based on these assumptions, digital and film pixels could be viewed as equivalent when the signal-to-noise ratio for each medium was equivalent. One measures the observed noise with a digital camera, as Phil Askey does in his reviews, then one adjusts the supposed "pixel" size of film of like ASA until the measured signal to rms-noise ratio is the same. This feels intuitively right, because our ultimate objective is to see the image to the exclusion of the noise. Your comparison between the D30 and film amounts to exactly such a comparison because, as you note, the rendered detail was at least equivalent yet the observed noise was less for the D30.
February 12, 2001
It’s nice every now and then to get a pat on the back, but more importantly an affirmation of ones position.Dale Cotton, a contributor to theCanon SLR Talkforum ofDigital Photography Reviewhas done just that regarding this report.
The current (March / April 2001) issue ofAmerican Photomagazine will be of interest to anyone following the ongoingCanon D30vs. film debate. It features a comparison review byJonathan Barkeybetween theFuji S1,Canon D30and ISO 100 color negative film.
It’s the latter comparison that interest me most because Jonathan’s results essentially mirror mine done several months ago between the D30 and ISO 100 transparency film. In his review the D30 prints (dye sublimation) are described as "absolutely grainless" while regarding the C-print equivalents from film he writes, "the film image appeared sharp but with pronounced grain". He continues, "Everyone who viewed the prints side-by-side preferred the digital images in terms of color, contrast and saturation".
Do I detect a trend?
A year after the D30’s announcement and some 8 months after this article first appeared the debate has died down somewhat, and the supporters are outnumbering the naysayers by a considerable margin. Here’s a recent letter from a photographer in the Canary Island.
"Hi Michael, I’ve come across your article on the D30 versus film, I agree absolutely with all your conclusions, and I can add something. I own a drum scanner capable of 10,000 ppi and 4.6 D and also made tests with Provia 100F, scanning at 5000 ppi (the theoretical limit of film resolution). At that res the grain shows clearly so the D30 prints is vastly superior. The most approximated look to a D30 file is a 4×5 transparency drum scanned to match the size of a print from the D30+GF Pro (30×40 cm). So I too am impressed by the performance of the D30. It seems that all we used to know about photography no longer applies !!"
This subject is now featured inVolume 1, Number 1of The Luminous Landscape Video Journal.