Digital SLR Cameras, Lenses and Film

Thankyou Bob, very informative- and extremely useful for those considering digital and purchasing suitable lenses. Thanks

Good job Bob. This article has given a sound comparison on "normal" scanned image and digital image. I definitely agree that the "quality" of digital image has also met my need.

But I wonder, how do people evaluate the "quality" of an image?

I would like to say it is quite subjective to determine the "quality" of an image. To me, although the "jaggies" on diagonal lines is hard to be seen, I dont like the grainny, and rough feeling appeared in the scanned image, I will say that the general "quality" of this image is not as high as that from EOS 10D. I can accept jaggies when the image is magnified or viewed at a resolution higher than 3000x2000 pixels, and I like the smooth and noise-free feeling that I get from the digital image. In other words, I will say the digital image is of higher quality than the (untouched) scanned image. Again, this is very subjective. Just my 2 cents.

By the way, if I were to judge whether I will use film or digital to take pictures, will be to consider the usage of the final image. If I will post them into the 'Net or I wanted to make special effect on them and become the desktop wallpaper in my PC, I will use digital camera. And if I will print them and send them to my grandma who have no idea of what a computer is, or if I find a nice little Micky-Mouse blank photo album, then I will use film camera!

Very good article. It doesn't try to be exhaustive, but takes some common questions and answers them well. The comparison between the digital and film images of the "real-world" target was particularly nice, although I'd like to stress that one image doesn't tell everything. Bob obviously has experience shooting with film and with the Eos10D, which matters more when making the comparison than a few images on the net.

Bob,
  From looking at some of the 10D images it looks like in-camera sharpening has left a halo around the high-contrast parts of the image - for example the middle picture of the three when comparing film and digital.
  How much does this artificially boost the 10D images compared to the film-derived images?
-P.J.

Yes, if you are all willy-nilly over sharpness. Shoot with 4x5 or 8x10 film...

I'm not sure what an "artificial" boost in sharpness is. If it looks sharper, it is sharper, even though resolution isn't affected. Most of us sharpen both digital images and scanned images fot the maximum sharpness that doesn't show artifacts. You can see the artifacts in the images here because you are looking at the individual pixel level. You wouldn't see the artifacts in a print. If you can't see them are they really there? It's a bit like the old "if a tree falls the the forest...." question!

Actually film also shows "artificial" sharpening to some extent too, via adjacency effects, where edges show higher acutance than they really have in the image produced by the lens, due to uneven action of the developer at edges in the image. In fact in B&W work developers may be chosen especially for their adjacency effects in order to increase sharpness. So boosting of sharpness isn't only a digital thing.

Note also that digital cameras have built-in enhancement algorithms that reduce the appearance of noise in smooth areas. This can be done using software such as NeatImage for scanned film shots, which should in my opinion be done to the film images before a comparison is made.

Neatimage is great on some shots, not so great on others. However it takes about 2 hrs to process a single 4000dpi scan on my machine (550 MHz) - after I've setup the noise parameters which may take 5-10 minutes. That doesn't exactly speed up my workflow! Even with a state of the art 3GHz machine it's still going to take 20 minutes or so, or 12 hours for a 36 exp roll of film...

There's no argument that you CAN get better results from film, if you pick the right film, scan it optimally, process the scan to reduce noise etc., etc. It's just the amount of time and effort (and expense)that takes is generally not worth the possible marginal quality improvement for the majority of my applications.

As I said, if I wanted images I could blow up to 20x30 or more, I'd shoot slow speed film in a 6x9 or large format camera, where the extra effort would really payoff.

Scanned 35mm film of the right emulsion may indeed provide more detail, but the examples here show wider color gamut that direct image capture provides. The color of a film scan is limited by the dyes used in the emulsion, not the color gamut of the scanner. Color detail, even with bayer sensors, is far better on a direct capture from a competent camera than from a film scan. I find a direct capture allows much more excavation of shadow detail than a high dmax scan from a transparency.

Although the detail isn't there, the smoothness of tones that you get on a large print from a 6 Mpix capture has a feel akin to a print made from large format film. At 4000 ppi or higher, even a fine grain film is having grain resolved by the scanner. On a big print, the absolute detail is slightly better on my scans from Velvia or Provia, but grain is also obvious. In many cases, I get better 13x19 prints from the 6 Mpix digital camera image.

It's hard to compare absolute resolution between film and CCD emulsion, simply because you're doing it on a computer screen, which is CCD-favorable. I've had quite a lot of experience with my S2 Pro in the past year or so, and although digital images contain much detail (i'd say a 6mpixel image would print nicely upto 16x20), most are still easily distinguishable from their film counterparts.

CCDs resolve more shadow detail, but do miserably with highlights. There's not even one DSLR out there, regardless of chip size or manufacturer, that can handle highlights well. This is not because of limited latitude- it's because CCDs have yet to incorporate a design that would know how to digest highlights.

As for grain, lack of grain is somethimes favorable, sometimes not. I often intentionally added noise to my DSLR snaps just because the photo seemed unrealistic to me. Sometimes, lack of grain draws unwanted attention to even-toned surfaces. All in all, IMHO DSLRs work better for portaits than landscape photography.

B/W photography is possible with a DSLR, but film would bring out better textures. It's also hard to rely on one channel only, since current Bayer CCD designs leave you with 1/3rd or 1/4th of actual resolving power of the CCD.

Digital is another medium to master- quite different than film, and with great potential. I'd like to see more articles as this, Bob. Good work!

<em>since current Bayer CCD designs leave you with 1/3rd or 1/4th of actual resolving power of the CCD</em> <p> I'm not so sure that's realy true. For example the 10D has a 7.5 micron pitch pixel array. That's 133 pixels per mm. If we assume the usual Nyquist limit (which may not be valid, but at least should give an upper bound), that would suggest maximum resolution of 67 lp/mm. <p> 1/3 of that would be 22 lp/mm and 1/4 would be 17 lp/mm. The resolution of the 10D sensor is demonstrably better than that. I'd estimate the native resolution of the Bayer interpolated image to be somewhere around maybe 55 lp/mm, depending exactly how you define "resolved" and the orientation of the test pattern. <p> I don't know the details of the Bayer interpolation algorithm. It's possible that for pure red and pure blue bars of the right orientation resolution could be less than for black and white bars, but in practice, on real world subjects, Bayer interpolation really doesn't degrade resolution too much. <p> If you want to use the "F" word (Foveon), it doesn't show 3 or 4x the resolution of a same size Bayer sensor. It does show a bit more, just how much more depending on which test you pick.

Very nice article!

I agree expecially on the part where you explain that for the most part of uses, digital quality reached the necessary quality we need..

Shoot on something different from a 35mm frame if you need absolute quality..

I think that 35mm film has his strenghts in other hands instead that in those of quality, nowadays...

Ciao!

Two hours to process an image is a lot. On my one-processor 2.0 GHz Pentium IV, a 4000 dpi scan takes about 7 minutes to process with Neatimage. With a dual-processor setup it would be only a few minutes.

I think the use of a medium or large format camera to get high quality is not that great an option today. Neither of my favourite colour films (Royal Supra 200, E100GX) is available in large format, and only one of the two is available in 120. I will rather stay with 35 mm until the dust has settled and it becomes clear what exactly is possible with digital sensors.

One thing that I'm worried about is the people candid - the exposure latitude of negative film and top-quality analog printing is a quite efficient combination for the people photography that I do. I don't have to stare at histograms, play with laptops and memory cards while shooting 7 rolls of 135-36 during an event. It obviously depends on the lab, but I've struck gold in that department. And you can always scan the results and do what you like with the images. I haven't seen any digital printing method that would hold a candle to the 4x6's that I get from C-41 with traditional means. It's kind of frustrating that there is no way that I can reproduce the quality small print without sending out the negs. Why would I want high quality small prints? When a presentation consists of several pictures together, putting them in an album is difficult if they're 11x14's...

Bob,

It's not the apparent lp/m ratio that I'm concerned about... B/W digital photography is sometimes handicapped because we normally can view and print the images in 8 bits per pixel. When you got 3/4 channels, it gives you sufficient gamut, but when you got only 256 levels of gray, you'll probably be experiencing strange rendering effects in similar-tone subjects.

Moreoever, as you've mentioned, the Bayer algorithm normally performs well (when the image does lend itself to it) with retaining the original resolution... But when the original image does exhibit Moire patterns after been 'Bayer'ed, the b/w transposition (especially for b/w images based on a single-channel) is catastrophic.

Foveon: I'd wish one of the big names would've supported F-technology. Just a wish...

Ilkka - 7m is pretty fast! I've just been using the "free" version of Neatimage to play around with. I didn't think the "pro" version was supposed to be any faster but maybe it is. Are you using the pro version?

Otherwise it's hard to reconcile 7m vs 120m with only a 4x increase in CPU speed unless your machine is somehow a lot more efficient than mine (cache?). I have 512MB RAM, lots of free disk space and nothing else running on the machine. Win 2Kpro OS.

Maybe some PC expert could shed light on this. Or maybe there's a new version of Neatimage I should try.

I was struck by how much better the resolution chart taken with the same lens looks in digital than in scanned Kodachrome. Most of the film partizans argue that scanning at 4000 dpi doesn't get all the information. So the question arises: if you look at the Kodachrome slide with a microscope, is it actually resolving more than the digital sensor???

The article is well executed and I enjoyed reading it. Applause to Bob!

I'm a little taken aback by the large bubbly grains of the Kodachrome25. Anyone know what's the specs of this discontinued product in terms of RMS grain size? Is the grain the same size as Velvia (or Provia?) ?

Very interesting and informative article. I also found the included Photomicrograph very interesting. I have long had the impression that film contains more detail than a 4000 DPI scanner can reproduce. I also suspect that film scanners exaggerate grain, and that as a result a large print will look better if printed on a traditional enlarger (although I have not tested this).

But digital is clearly the way of the future, and some day even medium and large-format photography will be obsolete. Today however, one of the major drawbacks of consumer DSLR's is the small sensor size and the resulting focus multiplier. I do quite a bit of work at wide angles, and find the 1.6x multiplier to be really inconvenient.

Photomicrograph of K25 test slide

This isn't meant to be a test of Kodachrome 25 or how well it scans on an FS4000us, so any data on those subjects shouldn't be used in that context.

The FS4000US scan at 4000dpi does not, and should not be expected to, reveal all the detail that you can see in the slide. Under a microscope you can see finer detail than a 4000dpi scan can reveal. The very BEST a 4000dpi scan could do is 78 lp/mm and in practice it won't quite resolve detail at that level. A good lens will put close to 100 lp/mm on Kodachrome under optimum conditions (tripod, perfect focus, mirror locked up etc.).

Here is a photomicrograph of the same section of the K25 slide. It reveals a little more detail than the 4000dpi scan. Not a whole lot more, but certainly a little more. It's possible a better lens would yield even more resolution and it's likely that the test pattern if shot on Technical Pan B&W film would yield significantly more resolution.

The real question is, however, what happens in the real world, where you're not using mirror lock up, you don't have a prime lens set to f8, your subject is moving, you may or may not have a tripod and you are shooting ISO 100 or faster film. Under those conditions it's quite possible that a 4000dpi scan will extract all the useful information out of a slide or negative.

Bob, thanks for the concise and accurate article. Your comment that the digital versus film decision depends on the workflow of the particular photographer is a key to understanding the issue and to coming to a conclusion that works for that photographer.

I am a hobbyist and use film because the images appear more pleasing to me compared to digital, especially color range. Also, digital can't touch black & white film in my opinion. Furthermore, I don't take the film/slide/print image digital because I too can't afford drum scans and, if I could, digital output is not as pleasing to me. Unfortunately we can't compare a film based slide to a digital image online which makes my workflow defective in the online sharing of images category.

Thus, in my opinion, if one wants a slide or print, then film is better. If one wants a digital file, then digital may be better depending on the tastes and workflow of the photographer. I say "may be better" because, for photos presented on photo.net, the photos taken with a digital camera are usually distinguishable and less desirable than photos recorded on film, to my eye. However, this may result more from average years of experience in the two crowds than the camera.

Great article. I often think of comparing digital and film to comparing CDs and LPs. CD wins on numbers but some people prefer the sound of LPs. The same is true with digital compared to 35mm where I find the "feel" of films like Velvia and Provia very pleasing.

That is not to say that digital isn't very good, I would switch to a 10D in a second if it made financial sense. But the camera + cards + very wide angle lens + other accesoiries will cost me 300+ films to earn back and shooting only 50 rolls or so a year, this would take me 6 years to earn back. Darn.

Great article, but I would like to point out that while comparing the sharpness of digital vs film isn't exactly apples vs oranges, it's at least tangerines vs oranges. So in some ways these comparisons aren't important. Too many people are too obsessive about the digital vs film debate (and shapness in general).

Digital will probably take the market for non-art applications. Eventually it will be sharper/easier/quicker/cheaper than film for just about everything. But it will probably never have the "chi" for artistic work that traditional processes have. Even if it's in the form of "negatives" made from digital images, the printmaking process will always be around, if only as an art form. People are still making tintypes and using wet plates.

For example,the "sharpness" of an email I type is way better than my horrible cursive, but if I'm writing a love note, it's going to be in my sloppy writing. And even when digital has a much higher resolution than film, I'm still going to be taking pictures of art-decco buildings with my vintage TLR on Tri-X and printing on fibre, because that's the look/feel I want.

Too many photographers get caught up in the technology and forget about the art.

Bob, I'm using the pro version, with 512 MB RAM. My best guess is that Neatimage uses Wiener filtering and that's a linear operation, which Pentium IV with RIMMs does very fast indeed with the appropriate programming.

I didn't notice any significant difference in the processing time by the demo vs. the pro version.

It's interesting to see the photomicrograph. The difference is quite clear. I'm also scanning at 4000 dpi and I'm happy with the results when the slide is properly focused (especially since I don't want to deal with larger than 120 MB files in photoshop!). Still, 4000 dpi is enough to see significant differences in image quality between lenses.

Bob, I recently went from the free version of NeatImage to the pro version. The pro version takes about half the time to process an image. This is consistent over colour digicam images at a range of ISOs and also B&W film scans (XP2 and N1600). Several other folks have made the same observation.

Good article, well thought out. Thanks,

I noticed the white halos around contrasty edges on 10D test images. Where they come from?

Has anybody tested Canon 17-40 f/4L on 10D here? I'm confused as I was thinking about buying one but have seen some not so good comments about it and people saying that in fact Sigma 15-30 is better? 16-35 is out of my reach for now but used 17-35 could be an option (but t's focusing distance is twice of 17-40...).

Best,

The white surroundings to black lines are the result of edge contrast enhancement, ie. a sharpening filter operation. Due to the low noise in digital captures, you can apply it quite freely without the increase in grain that film images show (I assume the camera does it automatically to some extent). Of course, if you first run Neatimage, then you can use wild sharpening on film scans as well. Still I often prefer the original version with grain and no sharpening at A4 size, larger is another matter.

On my machine, processing the same file with demo 2.2 vs. pro 2.3 takes exactly the same time.

Another positive aspect of having better lenses: pixels will inevitably get smaller, and you won't need to buy new lenses to match the increased sensor resolution. Buy quality optics once and use it all your life.

I think two hours is too much. Something must be wrong with the computer. In my PC, Neatimage Home Edition v 2.3 takes 8 minutes to process a 3976 x 2651 pixels image. 1.4MHZ Athlon - 512 MB RAM - 120 Gb HD in RAID.

Regards,

The "problem" appears to be that Neatimage is optimized for the Pentium instruction set and my machine has an AMD K6 CPU. The FPU in the K6 is known to be slower than the Pentium FPU when the code is optimized for the Pentium. I actually get FASTER processing using a SLOWER Pentium II. With a 330 MHz Pentium II it takes about 40 minutes to process a 4000dpi scan of a 35mm slide. On a small test, the 330 MHz Pentium II was 30% faster than my 550MHz K6. Perhaps my 2 hour processing was on a particularly difficult scan. I'll run the test again.

Whatever the cause, a PC upgrade wouldn't hurt!

Great article, Bob! And thanks for all your recent effort to make Photo.net interesting again!

Well, I will agree with Bob totally if the monitor is the ultimate output medium. But bear in mind that monitors and prints will give u variation in colour.

With slides, u are seeing the 'true' or 'master' colour for other output medium to follow. And your image will appear more vibrant on a light box compared to a monitor screen.

True, but then viewing slides on a light box isn't the normal final use of slides. If it is, then there's no contest. A slide under a high quality 6x loupe on a light box blows away any monitor screen image.

As for color, well, that's another issue. While a slide may give you a "master" image, it's not an absolute color reference. Shoot the exact same scene on Kodachrome 64, Velvia, Astia and 100VS, then tell me which slide is the "master" slide when it comes to color reference!

Just a side note on image projection because slide shooters enjoy giving slide show: I know that there are some high end projectors for computers / laptops.Do these projected image provides better resoluton than the computer screen? And how do they compare to the images from a slide projector?

More on NeatImage...

A quick test between an 1.5 GHz Athlon with DDR333 memory and an older, 450 MHz Celeron with SDR100 memory gave running times of 10 min. for the Athlon, 40 min. for the Celeron (the image used was a crop from a scanned 6x6 neg, the amount of pixels processed was higher than a 4000 dpi scan gets from a 24x36 mm frame). This was using NeatImage demo. Memory size doesn't seem to matter as long as the program fits in memory (for winNT4 256 MB was enough, while winXP wanted over 400 MB for this image).

So I think that a fast CPU (Athlon or Pentium 4) with fast memory are the keys to good NeatImage performance (incidentally, these two help photoshop performance too.)

Bob and Ilkka,

On the performance of NeatImage on PCs. There are many factors contributing the performance of a particular piece of software on a particular machine, not just CPU speed. A few of these have been mentioned, such as type and speed of memory and the way a particular processor configures its pipeline for certain instruction groups. However, there are many other contributing factors as well. Such things as the chipset used on the motherboard, the operating system, the disk drives and the disk controller, other interfaces and their drivers and, in fact, how all these operate together in various combinations. Much like with photography the best way to decide on the right computer is to try it out. Just like with photography though, this is difficult to do. There are not a lot of computer shops that will allow you to install Photoshop on their systems and do some benchmarking. There are a few though. In the new digital age it seems that building the right relationships is still as important as ever.

.. and the 28-105/3.5-4.5 are all good lenses
I don't own this lens but have 24-85 and it is not good - very soft at 85 mm, distortion at 24mm, average sharpness between them. Yes, it is light and inexpensive, but no good in terms of optical quality.
Acccording to photozone 28-105 is not better than 24-85 and all tests I saw online said that 28-105 is just an average consumer soom like 24-85 or 28-135 are. Still from time to time I hear the myth that 28-105, 28-135 and 24-85 are good (not cheap nor light but *good*!). Quite strange.

Bob, Great article and thanks for pointing out the Canon 10D actually has more density than many hgher MP cameras. In one of the sections you showed the 70-200 at 187mm to show a similar FOV for the 10D and film, since there really is no magnification factor, only really a cropping due to the smaller sensor of the ccd, aren't the left and rightmost images comparing the slide and 10D the most valid? thank you again for this and other articles Brian

Peter, it's quite clear from Bob's images that the 10D cannot match the resolution of film if the reduced field is compensated by increased magnification to yield a print of the same size. (Compare the middle and right images.) It also indicates that if the pixel density of the 10D is retained, a full-frame digital body would show significantly better results than K25.

There is no comparison. Rather there shouldnÂ’t be. Just because you can print Digital pictures or just because you can scan film doesnÂ’t mean you compare them both. The reasons are, that their purposes are different. The evolution of technology has given us the flexibility to use different mediums in realization and storage of pictures. First choose this medium of presentation, and then choose the technology to capture it. We sometimes get lost in the obvious and go in search of the oblivious. If you print a digital picture and compare it with a film print - of course film print is better. If you scan film and compare it with a Digital shot - of course Digital is better. When what you want is the best, why compromise to compare which is better than what?

What is important is to understand the "cost" of convergence. The problem with that is “we need to have the cake and eat it too”! So how do you solve the problem? Get 2 cakes! Eat one and keep one. Get 2 cameras - one film and one digital (when you can carry some 10 lenses - can't you carry one extra tiny camera?). Yes, no kidding - its cheaper and better. Before we play the pixel game - Lets ask ourselves a question - What is the final size of your digital picture? And at what resolution? And what is the size of the monitor and its pitch that you or your friend might use, to see it? Then ask the question, "So how much resolution is that?" - You will end up with an answer that that will lead you to conclude that a tiny digital camera less than $400 is worth its while. Do all else later - thanks to Photoshop. And, don’t throw that 'ol 35 mm away - its still got some shows left when you want to print.

Wake up world!! Photography is art. ItÂ’s all about capturing the moment, the mood, and the very essence of time, that lets us reflect and document the past. Remember technology is just an enabler - itÂ’s just the means to get there. Not the end in itself. So donÂ’t go talking to your friends about your new 5MP - 20MP camera. Talk to them about the mood you captured - that lovely girl with a curl, that awesome shot at the ball game etc. Spend time, capturing time - so you can share a thought, or re-live a moment. That is what photography is all about.

First off, nice article.

Just to let everyone know, but I had a client order a 20x30 print from a recent event. I shoot with the EOS 1D(4.13MP, but larger sensor :-). A little color correction, dust spotting and very minor unsharp mask, set it to my lab(whcc.com) and WOW, this print looks amazing. BTW, this was a full frame in-camera large JPEG shot taken with a very nice f/2.8 L lens and 1.4X TC.

This was the largest print I've ever made. Largest digital prints before were a few 18x24's and 16x20 from some Fuji 200 color neg scanned with a SprintScan 4000, same lab as above. Largest print from neg (Fuji Reala 100) was a 16x20 from a local pro lab. I'm going to have to say that the 20x30 was the nicest large print. Don't get me wrong, the others were great as well, two are hanging on my living room wall and the others were sold to clients. Ironically all of these shots were full frame, so that helps.

Bob- Do they still make Tech Pan 32? I haven't seen it in years :)

http://www.dpreview.com/reviews/canoneos1ds/page20.asp

On this page are presented two photos:

1. Canon EOS 1DS 11MP

2. Canon EOS D60 6MP

taken of the same scene, minutes apart, with controlled: exposure; focus; framing; BUT shot at different focal lengths (due to the 35mm-format crop factor in the D60) and using JPEG Large/Fine file format.

Download both pictures (due to copyright restrictions I will not present crops from these two pictures here).

Resize the 11MP 1DS picture to 3078x2048 (the D60 image is 3072x2048).

Compare the two pictures.

Unfortunately the picture is made up mostly of large areas of no detail. But there's some fine detail on the bottle labels, the apple and the watch face.

A good area to compare is the "yellow" side of the apple and where it meets the Baileys bottle, specifically the text "(Bail)ey & Co., Nangor Road", and the rural scene on the label. Another good area is the Martini bottle's main label (which is arguably in critical focus with both cameras).

In my opinion there are three reasons for differences in detail:

1. different focus plane/depth of field, and the fact that the camera was in slightly different positions when the two pictures were taken (parallax differences can be observed)

2. bayer sensors lose approximately 45% of the nominal pixel count of the picture, through the bayer process, i.e. a 6.3MP bayer picture is approximately equivalent to a 3.4MP full-colour-pixel picture.

3. noise reduction by the D60 smears low-contrast and low luminosity detail.

You can test 2. by resizing the 1DS image in Photoshop:

a. size to 2278x1516

b. size to 3077x2048 (1 pixel of width lost in rounding)

(Both resize operations using bicubic resampling.)

Anyone prepared to post a more tightly controlled comparison?

There is a real side effect of the "lost" resolution that bayer sensors incur. The approximately 45% pixel-count resolution-loss in a bayer sensor is approximately a 25% loss in linear resolution. In stating this, I'm ignoring the results that a comparison of resolution test charts produces (about a 23% loss, approximately a 4.8MP picture), since digital cameras' interpolation algorithms are usually optimised for these charts. Real world detail is not normally black and white...

Simply put, this means that you need a lens with an excess 33% linear resolution to achieve the desired resolution of your digital camera's sensor.

Or, in other words, to achieve the 67lp/mm that the 10D's pixel-pitch implies, requires a lens capable of 90lp/mm.

If you're shooting a resolution test target with a Canon DSLR then I estimate an extra 14% linear resolution is required, i.e. a lens capable of 76lp/mm on the D60.

Of course, if you're using a camera with a Foveon-type sensor, there is no appreciable resolution loss caused by the sensor's colour-interpolation (since there is no colour-interpolation). So, a 67lp/mm lens on this camera is approximately equivalent in resolution to a 90lp/mm lens on a 6.3MP bayer sensor camera.

According to (let's hope I've got the right lens...):

http://www.photodo.com/prod/lens/detail/CaEF28-70_28LUSM-64.shtml

the lens used to make these two pictures has a weighted MTF of approximately 81 for the 1DS picture's focal length of 70mm, and approximately 83 for the D60 picture's focal length of 43mm.

This would imply that the lens is limiting the D60 in this test.

For comparison, the 1DS pixel pitch implies a native 57lp/mm, which requires the lens to be capable of 76lp/mm, which it just achieves.

Of course, all of this is ultimately theory - more tests needed!

Bob, do this again in two years. (Anybody want to place any bets on the outcome?) There are a lot of new people coming along who are asking not only "What should I buy?" but also "What should I be learning?" The time invested in digital processing is going to give increasing returns. That invested in film processing is likely to give diminishing returns, at least at the 35mm level. All bets are still off at MF and LF for now, but look down the road--and not necessarily too far down the road. Film will perhaps always be around, but it is going to be increasingly inconvenient, and probably increasingly expensive, and pretty soon that increasing expense will be very difficult to justify on any grounds, artistic or otherwise.

Time spent on learning photography should be spent on learning your subject matter and to use the visual language, not on "film processing" or some such silly thing. That's done by the lab you know. Even when shooting film, you use photoshop to do what you'd do with digital captures so the difference is just the image capture medium. To me, it's absurd to move to a new technology before it's mature & stable enough. The time for some people has come, but many things that I'm used to shooting would require a prohibitative investment in equipment to do in digital at this point. Yet in three years things will have changed. That doesn't mean that I won't shoot film. I probably will. I'm in no hurry - I have a working technique for doing what I like and need to do.

Jawed

Just about any lens not made of plastic will easiliy resolve 100 lp/mm at f8. In fact most lenses will resolve closer to the theoretical limit of 200 lp/mm in the center of the frame. The reason they don't put that on film is that film isn't very good!

So resolution is very much sensor limited rather than lens limited.

Actually it's the MTF near the resolution limit of the sensor that counts when it comes to image quality near the resolution limit, but that gets us into a whole other area.

Theoretical analysis is fine (if you get it right), but it's no substitute for doing the actual tests!

But as you decrease the pixel dimensions, the signal-to-noise ratio goes down quickly. If we want to retain high-ISO operation (many people do, including wildlife photographers!), the current pixel density is likely to stay. Of course, they may make higher resolution sensors for landscape and studio applications, but those will likely have a low volume & remain very expensive.

Really well prepared article. It has high value for instructional purposes.
There's seems to be one area which I was hoping this article would talk about: can DSLRs capture enough dMax for B+W images to rival scans from B+W negatives.

Blah blah...

The scanned Kodachrome is nowhere near the result I get with my "standard" slide film Provia scanned with a Nikon LS-4000.

I think that by choosing a slightly better film than the Kodachrome and a better scanner than the Canon the difference will be much bigger than suggested in this article.

The DSLR still lags in the areas colour correctness, contrast and resolution. (But the gap is closing fast).

My Canon Digital IXY S-400 is amazing!

///Magnus

Gosh, Bob, where to start? You equate 67 pixels/mm to 67 lpmm. But it takes two sensors or pixels to image a line, one for the black dots, one for the white space. So 67 pixels/mm is at best 34 lpmm resolution on the DSLR. Similarly, if the FS4000 scanner is 78 pixels/mm, it is equivalent to ~39 lpmm (not 78 lpmm). So your calculations and comparisons and conclusions are based on wrong and inflated digital resolution numbers, right? Here are the simple facts: Kodak has estimated that 35mm mid-speed films are equivalent to roughly a 24 Megapixel sensor. Pop. Photo's lab tests suggested real world values of 40 megapixels for 35mm films in their tests. Fine grain films like Kodachrome 25 have much higher resolution, and should score above 100 MP equivalent. An AFIP researcher's peer reviewed paper determined one fine grain 35mm film as being 124.76 MP equivalent.

So any consumer or amateur "tests" that conclude that a 6 MP DSLR equals 35mm film are simply wrong, based on these facts from film makers and lab tests. Kodak etc. are right, and you-all are wrong ;-)

The only thing you are "testing" is your scanner and your technique. Your scanner is acting as a ~40 lpmm low pass filter, blocking all high frequency image data from film. You concluded that K25 roughly equaled a 6 MP DSLR image. But Kodachrome 25 is more like a 100+ MP equivalent sensor, per above sources.

So you have lost 94%+ of the image quality of film in your scanning process (from 100+ MP to 6 MP), right? This is equivalent to taking a 6 MP DSLR image and downsizing it to 640x480 pixel webcam image size (also ~94% loss). That's a pretty big error! It is also the same 16X factor by which kodachrome 25 is superior in resolution etc. to a 6 MP DSLR ;-)

Let's move on to why digital cameras can't use the full high resolution and fine contrast response of even so-so 35mm lenses. Due to anti-aliasing needs, DSLRs have a low pass filter between the lens and the sensor. This low pass filter limits the high frequency data seen by the sensor to ~50-55 lpmm equivalent.

So right away, any claims of DSLRs imaging above 55 lpmm (i.e., 67 lpmm here) are problematic, right? On the other hand, lenses using velvia and other fine grain films and good technique can do much better than 50 lpmm. Velvia is rated at 160 lpmm (1000:1 contrast) and 80 lpmm (1.6:1 contrast). So fine grained film has hugely more megapixels at 100+MP equivalence than any prosumer 6 MP DSLR sensor.

Most 35mm lenses can easily deliver 200 to 450+ lpmm aerial resolution (at least centrally to your 1.6X sensor). The low pass filter is going to reduce that to circa 55 lpmm and below before it gets to the DSLR digital sensor.

This is like two racing cars, one able to do 200 mph and the other 450 mph, but both are stuck behind a police car in a 55 mph zone. As long as the police car (low pass filter) is there, you aren't going to exceed 55 mph (55 lpmm), regardless of how fast either car is ;-)

Similarly, the sensor never sees the high frequency data from the lens, due to the low pass filter. So the sensor doesn't care if the lens is a really high resolution zeiss or leica optic delivering 450+ lpmm, or a so-so third party lens barely capable of delivering 200 lpmm aerial resolution. Both are delivering waaay more than the 50-55 lpmm limit imposed by the low pass filter. Everything above the low pass filter limit is chopped off and never seen by the sensor. So the high resolution and high contrast benefits of the pricey lenses are lost on the digital sensor, due to its 50-55 lpmm low pass filter limits, right?

The anti-aliasing low pass filter is the physical basis for the statement that high frequency response lenses (i.e., high resolution, high contrast) are NOT needed with digital sensors. Designers of digital lenses such as Schneider have noted this fact too. They have "dumbed down" their digital lens design resolution and contrast response to better match the digital sensor requirements, while improving other digital lens parameters that can make a visible difference. These factors include color fringing, as you noted, plus cost and distortion and flat image planes and so on.

I was hoping that this would be a tutorial on those digital optics topics, and how they differ from film oriented lens designs. Instead, it was another erroneous effort to prove by amateur "tests" that a 6 MP DSLR is as good as fine grained Kodachrome 25 film. Kodak says that's wrong, and Kodak is right ;-)

Similarly, the low pass filter in DSLRs blocks high frequency data from the lens. So selecting a lens for its high resolution and fine contrast is wasted effort on current DSLRs with a 50-55 lpmm low pass filter limit. The lens resolution isn't the limiting factor here, it is the low pass filter.

Note that I am not saying you won't see any differences between lenses. You may be able to see some DSLR image improvements from higher cost lenses which are better corrected for color fringing, distortion, flatness of field, and other parameters. But the major $$ spent on higher resolution and fine contrast response for film lenses is going to be blocked by the low pass filter.

That's why Schneider and others are designed a new generation of digital lenses, trading off high frequency response for improvements such as flatter field curvature and lowering lens aberrations which can and will make a visible improvement in DSLR image performance.

The flip side is that today's high dollar lenses, which have been optimized for high frequency response (i.e., high resolution and fine contrast) on film are a less than optimal match for current DSLR sensors. So don't expect that your lenses designed for film camera use will work as well as future digital lenses designed specifically for digital systems taking into account digital sensor limitations.

A final point derives from the observation that even mid-speed 35mm film is at least 24 Megapixels equivalent or better, per Kodak. I have to assume that scanners will get even better, given that they have a long way to go as I've shown above ;-) So today's film images may become tomorrow's 24 MP or 40 MP or even 100 MP images as scanner technology improves. Unfortunately, today's 6 MP image will always be a 6 MP image at best. So film offers higher quality today, and more benefits for future digital images as scanner quality improves.

For those who want to challenge my analysis, you have to start by explaining why Kodak, Pop Photo, AFIPs, and other sources have all derived megapixel equivalents for film much higher than 6 MP etc. You also have to explain why higher resolution or finer contrast lenses can make a noticeable difference despite that low pass filter in there. Amateur "tests", especially any using scanners with film, have a large burden of showing how you did not lose 75-95%+ of the film resolution data, as here, to reach any conclusions about today's DSLRs being as good as film ;-) grins bobm

Bob:

Your article is very good, specially if you are determined to prove that for your case digital is better than film.

At least for me, the photographer's issue is the lens, their cost, their weigh, and the fact that the older one gets the worse it gets to hand hold them and lug them around. The logic of compromising on the camera body, which has the least impact on the photographic result whilst keeping your huge and expensive lenses, sound a bit strange to me. If you are happy with 11"x8" prints, I guess you could also use cheaper lenses than the beauties you own (300/4 L and the 400/4.5). Other combinations, film + 75-300mm vs digital + 300/4, could give similar results in the smaller prints you are happy with. The difference is that probably the second combination of equipment would be much easier to handle, carry around, and much cheaper.

Thanks for your article.

Claudio

I think it is important to keep in mind that what matters is not whether a digital sensor can capture all the information on a low speed, high resolution transparency or negative, but whether an image captured by a digital sensor can produce a print that matches the quality of a print made from a transparency or negative.

Once you scan the film or enlarge it with an enlarging lens and print on some sort of paper, much lower resolutions are observed. The question then becomes, "what resolution digital sensor is needed to make digital prints that rival a print made from a given piece of film whether the film is scanned and printed digitally or printed directly as an Ilfochrome, silver, or C print?

I don't know the answer to this, but I'd be amazed if it were anything close to 100 megapixels. I'm guessing that around 20-25 megapixels will pretty much obsolete a 35mm SLR, and 40-50 megapixels will do in medium and large format cameras/backs. But it's just a guess.

Hi everyone,

This article is ultimately very practical. People have brought up some very valid technical points about how Bob's test was "unfair". However, living in Jamaica, it is VERY expensive to get film drum scanned, anywhere for that matter. I look at this article from the point of a home viewer. I believe that film scanners do give film justice, however image quality is very subjective. In my opinion film still has some kick left in it, as I can vouch by the very high quality of the slides my FM2 produces. Borrow a good film camera, then a good digital. Shoot to your hearts content with both and then compare the results, choose the one you like. All this tecnical stuff I have read gives me a serious heaadache. We must remember however, digital IS advancing, but so is film. Kodak's new Ektachrome E100G film is an example of this with almost nonexistent grain and amazing resolution. It blows away anything I have seen digital produce, but I am sure in X amount of time after I have written this digital will up the ante again. Then again a new wonder film will come out. BTW, if you want ultimate print quality, nothing beats a cibrachrome print made from a slide.

Robert Monaghan,

Bob does not need to explain why better lenses give better results on DSLRs even though there is an optical low-pass filter. He already explained that convolutions multiply MTFs and a better lens will always improve the result. In fact this is so obvious that I feel that Bob did not even need to mention it and give it so much space in his article. I feel that 35mm film is at best a 6000x4000 equiv. You could call that 24MP, but most cameras seem to count color matrix photosites, so perhaps it would take a 72mp camera to match film in resolution although the film would be really gritty at that point. But the EOS 1Ds does seem to produce great results without the problems of scanning, with the exception of highlight detail. But hey, slides also have bad highlights. Can I produce an image with slide film better than an EOS-1Ds? Of course, but it is mostly not worth it. I no longer have any interest in images that are not digital in the end. So with that being the case, scanning many images is too much trouble. I just shot 14 rolls of film and am scanning them at 4000dpi. It will take me over a week to scan them, even though I did not cut the negatives. I put in a roll, and wait 3 hours. Put in another, and then go to work, and put in another before bed. I am done with 35mm film. I still except to use 4x5 and maybe 6x9.

Hi every one !

about the comparison between digital & film : why do every one compare them on the computer, after having the film scanned? If the final material is the print, why not take a good picture of a landscape with beautiful colors (a real pic, not a chart test pic...) with a DSLR, and a Film SLR (vith velvia for instance) at the same time (same light) with different focal lengths to have the same field of view on both supports, and then print your DSLR file onto a big print, print your velvia slide onto a reversible paper of the same size (cibachrome, fuji supergloss, etc..._am I right about the names?_) and compare them. For that matter, you can scan both prints with a good flatbed scanner. I would be very interested to see the results with this protocol (I don't have a DSLR, so I can't do it myself). What do you think about that?

I'm addressing this response to Robert Monaghan, who maintains an excellent website on image sharpness, because it contains too many misconceptions to go unchallenged. Most readers who've compared EOS-10D images to the sharpest 35mm film images already realize that Bob M's comments don't add up.

To start out, the pixel spacing of the EOS-10D is 7.4 microns, or 135 pixels per mm. The Nyquist frequency-- the highest spatial frequency for meaningful data-- is therefore 67 line pairs per mm (lp/mm), not 34. The lowpass filter (lpf) in the sensor is designed to block most energy above this frequency, which could result in artifacts like Moire. Similarly, a 4000 dpi scanner operates at 157 pixels pixels per mm, for a Nyquist frequency of 78 lp/mm. Since a 35mm image is 1.6x larger than the 10D image, it has a simple pixel advantage of approximately 1.86. Digital cameras do as well as they do because film loses a lot of detail.

Now as for the estimates by Kodak, Pop Photo and others that it takes between 24 and 100 megapixels for a DSLR to equal film, I want to see the exact references to I can analyze them-- examine their assumptions and reject them. Hearsay won't do! The 11 megapixel EOS-1Ds is a good reality check. I'd challenge anybody to produce better 35mm color images than the 1Ds. There are reasons why photographers use medium and large formats, and they have to do with the limitations of film.

Kodachrome 25 (discontinued) performance is in Kodak Technical Bulletin E-55. The true indication of its sharpness is MTF-- Modulation Transfer Function-- contrast at a given spatial frequency relative to low frequencies. It is 50% at 43 lp/mm and 10% at 77 lp/mm. These numbers are good, but not extraordinary. Fujichrome Provia 100F has a 50% MTF frequency of 40 lp/mm. Other well-known slide films (Velvia, Ektachrome, etc.) are similar. Negative films, especially B&W, tend to be better. MTF curves are available at manufacturer's websites. They tell the story about real film resolution.

The ratings Bob M cites of 160 lpmm (1000:1 contrast) and 80 lpmm (1.6:1 contrast) have limited meaning, especially for 1000:1 contrast. They refer to "resolution" defined as just noticeable contrast difference-- at most 1.05:1 contrast on the film, maybe as low as 1.02:1. So Velvia at 80 lp/mm would have MTF of 0.05/0.6 = 8% MTF at 80 lp/mm. Velvia specs aren't on Fuji's website, but this is about right for Provia. The 1000:1 contrast may be of some interest to astronomers, but is of no value for terrestrial images. It corresponds to MTF of about 0.05/1000 = 0.005 percent. Any texture would be absolutely flattened at this spatial frequency!

The 200-450+ lp/mm aerial resolution of lenses is similarly impractical-- it's like rating a maximum car's speed in RPM. Sorry, but the rubber has to meet the road, and most lens manufacturers publish MTF charts. These charts typically display MTF percentage for for 40, 20, and 10 lp/mm. MTF at 40 lp/mm is considered to be an indication of lens sharpness. I admit I'd like to see MTF at higher resolutions. It's extremely low at 200+ lp/mm.

But the biggest issue is 4,000 dpi scanner performance. Bob M estimates that 4000 dpi scanners lose 75-95%+ of the film resolution data. I won't bore you with the details here-- it's on a web page-- but my analysis shows that 4000 dpi scanners can capture about 80% of the detail on film (linearly). That would be 64% area information. Not too bad. I welcome the new 5400 dpi Minolta 5400 scanner-- it will be a modest improvement. More importantly, I found that a print from a properly sharpened 4,000 dpi scan is slightly sharper than the best traditional darkroom print, which has to pass through an enlarging lens, and can't benefit from sharpening. I have a portfolio full of prints to prove this. I can easily fool people into thinking they're medium format.

The bottom line is what your eyes tell you, and my eyes tell me that the 10D is a damned fine camera. I can make 13x19 inch prints almost as sharp as my best 4,000 dpi scans (but grainless), and every bit as sharp as the best prints I made from 35mm in the traditional darkroom. I wish I could afford the 1Ds, but then I'd need a 24 inch Epson 7600 printer to take full advantage of it. One day... My thanks to Bob A for putting up this excellent page.

Hi Norman! there is an issue when everyone tackles this matter : partiality. when someone has invested money in DSLR, he tends to prove to himself and the others that this medium is far better than film. And your word only adds to this certitude. you say at the end

>>I can make 13x19 inch prints ALMOST as sharp as my best 4,000 dpi >>scans (but grainless),

4000 dpi doesn't deliver the detail of film, this is obvious (the scan/photoscope comparison and your link describe it well)

The thing is that you all raise figures of MTF and lines/mm where we just want to see real comparisons of real scenes, under fair circumstances (that is the best way to print the 2 medium if the print is your final material)

if the best way to print them for you is through a 4000 dpi scan, then you would do well better with a 5400 scan, etc...

if you shoot film, you are not limited by the current scanner technology, you'll be able to scan it with a better scanner later on if you wish. If you shoot digital, you're limited by the current ccd/cmos technology, and your images are frozen.

and remember that the best way to see a slide is with a projector, which is a completely other world (you won't contradict me on that point) and digital stays on screens and paper.

but I'm falling the other way, defending film against digital.

I'd simply want to see real examples of texture (hair foliage far away, road textures, etc) taken on film and on DSLR (with different focal lengths to have the same field of view) to see the different rendition. An objective test would require to try honnestly to get the best result from both mediums.

I don't want to be polemic on the subject, I'd like to know the pros and cons of each medium from objective minds, that's all.

to conclude, let's take pictures with what we have, and enjoy it, there is no need to battle in line to convince everyone that the other medium is INFERIOR to the one we use. the 2 mediums are different, comparisons are interesting to know which one we prefer, if we want to change/when, but don't take this battlefield position, please.

peace & love my friends ;-)) , and happy shooting above all !!

cheers !

request to bob : could you put a sample of bark from the 2 sample pictures you have? it would be interesting to see their texture. Thanks!

There are a number of sites that show "real" photo comparisons between film and digital --http://www.sphoto.com/techinfo/dslrvsfilm.htm is Steve Hoffmann's site with Canon 10D, 1Ds, and film comparisons. 1Ds blows away a Nikon Super Coolscan 4000 ED of Provia 100 ASA, and even the Canon 10D appears to match or exceed the film scans.

I'm a Nikon user (since 1956, as well as a Linhof & Bronica user) and have a newly acquired D100 that is providing image quality I wouldn't have believed from 6 MP. I've been scanning film for years, at up to 4000 dpi, and the D100 is providing image quality pretty near the best scans I've done, and better than many scans. As Steve mentions on his site, with digital, there is one lens system to contribute distortion and aberrations, with scanning, there are two, not to mention possible film flatness problems both in-camera and in-scanner.

Norman Koren's comments above are accurate as always (and his images, especially his large prints viewed in person, are testimony to his mastery of the medium). The comments by Bob Monaghan puzzle one for their lack of accuracy. One should check facts before attacking someone else's presentation. And finally, Bob Atkins is to be commended for his excellent presentation in this forum.

thanks for the link, very interesting. but it is a comparison between DSLRs and scanned film, not really between DSLR and film... ;-)

thanks for the link, very interesting. but it is a comparison between DSLRs and scanned film, not really between DSLR and film... ;-)

Since I shut down my chemical darkroom several years ago, I won't be doing any comparisons between optical prints from film and DSLR prints in my own lab. I have some 11x14 and 16x20 optical prints that I made with my Durst L1000, CLS 450 colorhead, Schneider/and/Nikkor enlarging lenses from negatives made with my Linhof Kardan 4x5. They are very sharp. And I have made prints of the same negatives scanned last year, and printed at the same size using one of my inkjets. Note, the scanner is an Epson 2450, so the amount of image information I've scanned is less than possible with a really top-end scanner. And yet the scanned-film prints look better overall than the optical prints.

Unless you use a magnifying glass to see the ink droplets, the inkjet prints look pretty much identical even on close inspection (actually, the color is better on the inkjet prints). From those tests I was convinced that a good inkjet print looks like a good optical print. And I was comparing a scan made using the Epson 2450 to a print made using a top enlarger with excellent optics.

This still doesn't address optical print versus DSLR. I've also made prints from my D100 enlarged to 12x18, and they look excellent. In fact, the lack of grain, and the purity of color, produce prints that look better than many of my enlargements from 35mm shot on Reala 100 and Supra 100. And those look really good.

I have access to a couple of Noritsu printers (digital and optical), and plan to make some tests where I'll run prints on the same printer from film and compact flash (same image taken using the same lens) -- but this comparison still will use scanned film (which is what the digital Noritsu uses). I can also use the optical Noritsu to print the film-based images. Unfortunately, this wouldn't really show what a good optical print can look like since these automated printers use very diffuse light to limit effect of scratches (the digital Noritsu uses Digital ICE).

My advice would be to get a good Nikon or Canon 6 MP DSLR, shoot pictures, print them, share them, and enjoy them. After 40+ years of photography, including running a commercial business (photography and custom color printing), I'll never go back to a chem lab. I will continue to use film, especially in large format, but scan it. And I'll be shooting the vast majority of my work digitally.

I just wanted to add a comment about the 10 D, and digital in general. Right now I'm unsing a Canon EOS 33 wich is a fantastic camera. With Kodak Supra 100 and my EF 50 1,4 and EF 85 1,8 I got razor sharp and ultra fine grain prints made at my local Fuji Frontier lab. The 10 D seems to be a digital version of the EOS 33. The thing about digital that I like is that I can see the result right after I get home, no lab anymore, no cost for rolls, no waiting. I did read this whole thing about digital not beeing as good as film. So I wanted to know if the 10 D is good enough and downloaded some some original, unedited full size images from the web and gave them to my frontier lab to make prints. I just got them back. I am completete blown away! It's even better than in my wildest dreams!!! It's true, unedited shots look a bit soft on your screen, but that's because you can do all the sharpening later. I you take these "soft" pictures to the lab, the come out RAZOR SHARP! Belive me, the quality, is 10 times better than film! This is even more amazing if you know that the sensor of the 10 D is a lot smaller, so the pictures have to be enlarged more to get a print of the same size. And they look even better! The colors... everything. But the most important thing: The grain! There is NO GRAIN! Kodak Supra 100 had a little bit of grain, alsmost perfect. But it's such a low quality compared to the 10 D. You can stick your nose right on the print and you'll see no grain!!! Amazing. Now, I'll sell my EOS 33 and get the 10 D. I can't wait... what a quality...

Great article, Bob. I too have discovered that, since I began shooting with a 6 MPix DSLR, I have barely touched my film cameras. For my style of photography and printing and considering my workflow, the finished quality I get with digital is at least as good as what I used to get scanning slides.

I occasionally do want ultra-high resolution for very large format prints, but lugging around a large format camera isn't very practical for me, and I can't really afford to have 4x5 transparencies drum scanned. So, I do what a number of other photographers are doing these days - I stitch a number of 6 Mpix frames together to make a huge final image. This really only works for static subjects, like landscapes, but these are what I used to use my 4x5 camera for anyway. A brief article on creating ultra high-resolution images via image stitching can be found on my web site.

Very interesting Bob. From a repoduction point of view I would agree with you. Having spent the last 25 years of my life in prepress I can tell you that 2 years ago I wrote digital images off. I knew I could get a seriously better result from scanning a tranny than a digital image. Not so now!

How times have changed. If I want output an image say at A2 size, a good digital shot (maybe colour corrected in CMYK) will give just as good a result as a sharp 5x4 tranny......... if not better!

Our company now only drum scans images that are a)supplied as tranny/print or b)needs to be output AO+. Most of our clients now supply digital images placed in their jobs. Not so good for us as we don't do as many scans as we used to but we can get through our workload much faster. As long as the photographer and designer know what they are doing, the printed job will be just as good as if we drum scanned the images. Let's face it, I can output an image on our epson 9500 plotter at A2 size that I shot on my pocket size Canon S50 and get just as good a result as a drum scan from a 2 1/4" tranny. Why bother with film?

I now swear by digital. I am indeed a convert, and things will only improve. If you are still a believer in film for hi-res reproductions it's time to open your eyes and smell the coffee. Digital imaging is here and film WILL BE GONE sooner than you think.

I agree with the commenters who say the most meaningful comparison is to use actual prints (I have no desire to get a PhD in digital imaging...I'd rather put my energy into creating beautiful images that people want to display TYVM!). Bill Brandt once wrote the following about photography, "It is the result which counts, no matter how it is achieved." Therefore, I think a more meaningful comparision is between scanned film and digital capture, because most magazines and calendar companies these days work with scanned transparencies, and a lot of larger display prints (e.g., corporate and govenment art from medium and large format film) is made this way too now rather than with the traditional enlarger.

I'm coming from the perspective of a 35mm transparency shooter (Velvia 50 mostly) that I submit to magazines who turn the trannies over to their prepress houses for quality scanning (not always a drum scan unless necessary) for 300 dpi reproduction using CMYK offset printing (not a home inkjet printer). Currently I'm considering retaining my Canon 24-70 f/2.8L and 70-200 f/2.8L lenses and trading in my EOS 1V (that I initially bought to shoot sports action for magazines) for an 11MP 1Ds (since the 16.7 MP 1Ds MK II is financially out of my reach). However, of late I've really gotten into scenic images, and my goal is now to make prints up to 16x20 (and possibly 20x24 on occassion) as well as increase my calendar work. I know one top pro who routinely makes up to 30x40 gallery prints of sports action shots from his 1Ds and 300mm f/2.8 IS L prime (one of Canon's sharpest lenses), and they sell like hotcakes. However, Arizona Highways magazine doesn't accept digital capture at all, saying that output from an 11MP DSLR isn't sufficient for landscapes, which they opine is still the principal domain of 4x5 (although they will accept "exceptional" medium format and even 35mm trannies, though mostly of wildlife in the latter case).

Contrary to one reviewer above, I am one of those photogs who shoots all my static images with a tripod, cable release, f/8 or f/11 (to minimize image degradation from diffraction), a lens hood and MLU when appropriate, since it is necessary to wring every drop of resolution out of 35mm as possible since I'm competing with pros shooting 645 MF or 4x5 LF. In the alternative, I'm giving serious consideration to trading my Canon setup in for a 645MF system that would allow me to shoot scenics with 120 roll film now (again mostly Velvia 50 using every image acquisition optimizing technique above), and switch to a digital back later once the technology matures and prices come down from the stratosphere (6x7 SLRs are too heavy to drag into the field). I'm even thinking about getting a lightweight 4x5 view camera as a third alternative (and keeping the Canon system and upgrading it to digital later when prices come down to more reasonable levels).

I think it is important to say that film and digital will probably always look "different", but this isn't necessarily a bad thing - who says that how the world looks as interpreted by the medium of Velvia 50 (around since 1989) is the end-all? What is of more concern to me is that while digital capture has definitely "left the ground", it is hardly a mature technology like film, so that in 5 years 11MP capture from a 1Ds may be what "entry level" DSLRs will be using (we can only hope :) Any thoughts on how to further approach making this decision would be most appreciated.

wow..what a lot of noise. First off you have to judge using actual shots of the kind you usually take to see if you think film is better than digital.

Now I had a brand new Nikon N75 and 3 inexpensive Nikon lenses. 50mm 1.8 28-80mm 3.3-5.6 G 70-300 4-5.6 G all brand new.

I recently bought a D50 Body.

I can't explain why BUT the stuff I am getting from the D50 is much much better that the film shots I was getting.

Could it be the drugstore processing & scanning?

I use the same CVS to make prints from my digital files?

Anyway you just can't beat digital for so many reasons..BUT the main

2 reasons are 1- you can see what you have imedicately and reshoot if necessary. and # 2- you can just keep shooting without fear of spending to much money on wasted shots....

There may be a place for 35mm film cameras, but I think that place might be along side a stack of 45rpm & 78rpm records and some 8 track tapes.

Ever hear the 'bo-zo's' that claim that the sound from the old records and amplifiers using vacuum tubes is better than solid state equiptment.....sure..sure...if you like scratches I guess?

I have to say the 10D shoots are very sharp and detailled, wouldn't have thought. But there are three reasons why I still use a film camera.

First: I hate that crop factor. I wish there were more lenses which are adapted to the smaller sensor sizes of digital cameras.

Seccond: I don't want to view all my pictures on the computer. I like slide shows but for digital photographs you need a beamer or a really big TV. Both cost a lot of money.

Third: For reprints I still need a lab. I could print out my favorit pictures at home but that costs more and the quality is not quite as good as the reprints from a lab.

Wow, this thread has lasted more than three years, and the debate of digital vs film is still raging on. For the folks who were converted to digital back in 2003, are you still using the same camera you had back then? Don't you think the best picture you took back in 2003 would look even better if you use today's DSLR? You won't have that problem with film, will you?

I was 35mm in the sixties and only a fool would argue between DSLR and 35mmSLR. Each has its place and each has very good points for their use. However with CS2 a darkroom has become outdated and quality control is easier then it has ever been. Being able to dodge and burn in are things of the past with all of the adjustments that you can do with CS2 and now CS3 is coming. Printers are catching up in quality and having to put your picture where others can see them even 35mm has to do digital.

Do anyone had experienced that to photo the color "dark purple" the digital cameras never take it right when seen on printed or on lcd screen, it appears washed out... no matter how you adjust the iso, white balance or shutter/aperture settings.... why?