Michael, Thanks for taking the time to make that comparison. I have been wishing that somebody makes such a comparison for quite some time. A while back when you posted your article about "Pixel Count And Future Imaging Chips". I posted in this forum asking if the 11MP sensor is close to the resolving limits of Canon's best lenses, why are there higher MP sensors made for larger format especially that lenses made for medium format are probably less sharp that those of the 35mm format?
I think comparison article may come close to answering my question. Right now it seems to me that as digital comes closer to being able to closely depict the smallest detail lenses can resolve, it will have far exceeded film resolution/ enlargement performance at any of its formats. At that point a  good ressing up algorithm is all that is needed to get the enlargement size even film large  format can provide.
If we take 80 lp per mm  as the maximum resolution limit of most lenses and apply the Nyquist criteria (two photosites per line pair). A sensor with a photosite size of 6.25 microns is theoretically needed to match this resolution. This translates to 22MP  on a 36*24mm sensors. Again, the statement you mentioned in that pixel count is not likely going to increase by much is reasonable We will probably be able to get this pixel in the next year and half or less. This is good because more energy is likely to be directed to other areas like improving the noise performance of sensors, dynamic range,  frame rate deceasing cost, investigating new technologies, exaimining smaller sensor sizes and better and smaller lenses, etc.
I think next PMA is going to be really exciting.

Michael,

You now have me thinking about moving to a digital system.  I only have two things that are holding me back, cost (as you mentioned) and creativity.  The cost argument is self explanatory, but the creativity aspect is a personal thing for me.  

5 years ago when I first started shooting MF I found that my photography improved significantly.  I attribute this improvement not to a larger negative, but to heavier equipment, fewer frames per roll, and less automation.  My Mamiya MF gear forces me to slow down and think about my photography.  Before MF I had 36 frames per roll and a 5 frame per second motor drive that allowed me to forget the important aspects of photography, metering, the zone system, scanning the entire frame, and trying to visualize the final print.  For me this is a personal issue, I used to shoot 35mm like a machine gun on full automatic.  My Mamiya has forced me to slow down; I've been forced to think and I personally believe that I “see” better.

The piece is a great technical analysis between leading edge digital and MF offerings, unfortunately for me, I must decide between better technical output and "seeing."  For now the decision is easy because of price.  In a few years or less I may be forced to overcome my personal issues and try to continue "seeing" with equipment that will allow me to slip into old habits if I allow it to.

Eric

Am also considering moving from my Leicas to the D1s.  I do large files but would like to see a comparison between the D1s and medium format digital with the kodak back (for untethered working); ideally the H1 or Contax system for excellent lenses.

Michael,

Great article and I totally agree with your findings.  I've been working with scanned images for 14  years...as a professional capturing images for my client's archival, engineering and mapping use (I've been using 360MB b/w drum scans of satellite images for years).  I always have to worry about dust etc...

The quality and depth of digital images coming from the current crop of DSLR's are amazing.  A 180ppi image from a D60 will look better than a 4000 dpi scan of 100VS on my Canon FS4000!  

My main interest is shooting underwater images, large expanses of blue or black in the image....film grain just sucks!   However, for underwater, I still shoot film due to the lack of true wide angle and TTL flash capability. I also put together slide shows or my underwater , it's easier (and cheaper) to use slides than beg/borrow/steal a high quality digital projector....this is where I think a digital still loses out, but for print sales, web use and pretty much any other use of images, digital rules.

Thanks again for a great site!

Stu  

I am curious about the potential of EVF's (meaning the ones that are used like a traditional viewfinder and so can offer a similarly big bright image, not the tiny TV's an the back of the camera), as I see the possibility that they will one day combine the best of the SLR and rangefinder approaches.

1) About the one mentioned disadvantage, focusing difficulty, one potential advantage not discussed so far is the possibility of digital zooming at the focal point for focus check, and also, when using a tripod, scrolling around a highly magnified image to check focus, lighting, color and such at various parts of the image. I can imagine that this could some day work better than most traditional viewfinders for slow deliberate composition. However two-eyed photographers (fans of ground glass viewfinders) might prefer being able to scroll around a magnified image on the "camera back TV".

2) Another advantage only implied by previous messages is the elimination of the extra shutter delay, viewfinder blackout and vibration due to mirror movement. (A few SLR's do this with a fixed half-reflecting mirror, wasting some light though.)

3) Finally EVF's can avoid the undersized and/or dim viewfinder images of DSLR's that use 35mm format lenses but smaller than 35mm sensor sizes (which I think will be around in the lower end of the interchangeable lens digital market for at least a few more model cycles).

4) Relatedly, I suspect that the forthcoming 4/3 system might have to use EVF's to get adequate viewfinder brightess for low light work; can anyone comment on how well the Olympus E-10/E-20n optical SLR viewfinder with half-mirror handles low light?

5) About power consumption; I get good battery life despite using my LCD for most photos, by being careful about turning it off when not in use; wouldn't that be true with EVF's too? For example, have them activated by a half-press on the shutter release and turn off automatically after a brief period of idleness, as with most film camera electronics?

i guess, what i tried to say was that i would like to see a comparison between high end digital backs, latest high end digital SLR and drum scans....if the digital SLRs are better then scanned MF, how good must the digital backs be?

Michael,

I enjoyed your ultimate shootout article and agree with your results. But one aspect of digital that has aways nagged me is projection. I'm currently able to give awesome 645 slide shows with my Rollei MF slide projector with my glass mounted slides; however, any digital slide shows I've seen given with an InFocus or Epson LCD projector look like a cheap TV compared to a properly projected MF slide.

Do you know of any projectors that can do justice to high end digital? Or because these projectors are mainly aimed at the corporate market do you think that digital projection is really something that will never take off, and that the digital print is really the main thing, so to speak?

I know my fixation on projection is a bit off topic, but a proper slide show can be quite magical if done properly and done for an audience that cares. It's something I'd hate to give up.

Doug:

I saw somehwere at Rob Galbraith's site a while back a comparison someone had done with a 1Ds shot of a house and a 4X5 shot of the same house taken with Velvia that was as close to whatever lens had been used on the Canon. Whew, out of breath. Anyway, the result was similar to Michael's Ultimate Shootout against the Pentax 67. I don't remember what 4X5 capable scanner had been used or at what resolution.

I do remember that in comparing the two, the 4X5 was ever so slightly sharper or more crisp. By a hair. In grain however, same story as with Michael's tests. Even 4X5 had a lot more grain or noise then the 1Ds.

The best way I can describe it is that the 4X5 won on sharpness by only one point, while the 1Ds won on grain(or lack thereof)by two touchdowns and extra points.

My real world results with a Rollei 6008i system (w. Schneider lenses) and a Canon 1Ds, printing on an Epson 2200 have largely mirrored Michael's.  Here's the caveat -- "for color prints."  Here's my question for Michael or others using a 1Ds and medium format: are you satisfied that the 1Ds equals or surpasses film (even 35mm, for purposes of this question) in terms of tonality and subtle gradation for black and white prints?  There's just something about the converting color digital files to black and white, using any technique, that just doesn't do it for me.  The resulting prints generally look like they came, at best, from chromogenic black and white film (which I don't like).  I've started using imaging factory's Convert to B&W Pro, and am getting better results.

These issue for B&W are really what has kept me from selling my medium format equipment.  Gratnted, it has stayed int he camera bag since I got the 1Ds, but I worry about that one prize-winning B&W shot that would have been spectacular using the Rollei but only mediocre w. the 1Ds.

Dear Michael,

First I want to say, how much I admire your work and enjoy your site.

Somehow, after reading your shootout, I had the feeling, its pure goal was just to show how good the 1Ds is. Everything is so straight. Your 1Ds Field Report seemed to me to be more "neutral".

In that report you end up with "in terms of resolution, our final conclusion is that while it's very close, medium format still has a slight edge in this area". In the Con's corner you mention grain, sharpness, equipment. Even in your conclusion, there's no talk of a margin, it sound's more like a equilibrium thing to me. Your shootout starts with "The 1Ds surpassed my Pentax 645nii system by a considerable margin". Hm.

In the Field Report you compare the Data itself, which IMHO is the fairest way. Now you compare prints! Their quality is not limited by the available data but by the output medium (printer). The only conclusion can be, that a print in the size xy from a 1Ds meets the quality of a MF's (if it is not limited by a scanner, as shown in the drum scan section).

I don't understand your point of view. Heck, when I watch an image appropriately sized to my 20" CRT, a coolpix 990 meets MF! It is undoubtly remarkable, that the 1Ds is capable to produce high quality A3's. It leaves 35mm in the dust. In concerns of grain it rules. Mobility and variety of equipment is hardly competable. And up to A3 they deliver the same quality.

In pre-1Ds days there was a point, where you said (going for pure resolution): Ok, for this size prints I have to leave 6x7 and go for 4x5" (or whatever). And this is the size you have to compare. If the 1Ds rules there, it rules 6x7. Otherwise there is a region, where the 1Ds just doesn't meet 6x7 (which would be no blame!).

By the way, you confused some readers claiming about missing window bars in the grain section crops. The crops are different. If you overlay, you see that the right building's windows are quite different.

Best regards

Dennis.

Please excuse obscure expressions, I'm just a Kraut.

Mike,
Good morning.  I agree completely with your article.  I've shot 35mm, 4 x5, and panoramic 617 for years.  I've had the D30, then the D60, Now it's the D1 and the EOS1Ds.  I got the 1Ds at Thanksgiving and have shot all my assignments with it.  If I do a 4 x 5 shot , or a panormainc layout, I did the same series with the EOS1Ds to keep some comparitive notes.  I got interested in the digital side about 4 years ago, since I also shot Canon 35mm, with most of the L series lens.  I also  do my own large format printing on the espon 1280, 2220 and the new 7600.  The testing methodology and printing series, along with the drum scans are very similar to what you'd done.  I can also do a very nice 24 x36 with a some USM and it is very difficult to see the differences between what was being done with a drum scan vs. a worked raw file.  I just recently sold my 4 x 5 system, and also sold off my EOS1V bodies.  Haven't shot film in about 7 months, and the digital side - whether anyone want to believe it or not - is truely hard to beat !

It can only continue to get better.  Anyone who doesnt believe it, should rent a high end digital for an upcoming assignment, and then take a long, serious look at the results.  If they dont come to a similar conculsion - they will be passed by their competitors, becasue of their lack of acceptentacnce of where the market is really going.
Thanks, Tom

I read Michael’s Ultimate Shootout article with great interest and agree with his conclusion that, for his workflow, the D1s is clearly superior to medium format Velvia scans. The ensuing posts on the discussion forum confirm Michael’s prediction that no matter what he does, some nit picker will suggest some additional tests.

However, I think that David Littleboy’s observation that the tests were limited by the resolution of the Epson 2200 printer is very well taken. Michael’s friend, Norman Koren, has performed an impressive scientific analysis of printer resolution:

http://www.normankoren.com/Tutorials/MTF3.html

His research on the Epson 1270 printer shows that for 4000 dpi Provia scanned images, the printer’s “Response above 190 lp/in is mostly aliasing—garbage”. He concludes that there is no reason to send more than 300 dpi to the printer. I downloaded his resolution target for 4000 dpi Provia scans (since this is what I use) and printed the targets on my Epson 2200 and observed the same findings. I encourage readers to do the same. To give the medium format scan a proper evaluation not limited by the printer, I think that Michael should print the Provia scanned image (cropped, of course) so that the image sent to the printer at 200 dpi, well below the limitation of the printer.

He then should res up the Canon image and print under the same conditions. Since I don’t have the 1Ds or drum scanned medium format images I can’t rise to Michael’s challenge to back my assertion with hard data, but I think that my impression is backed by the above considerations.

It would also be useful if he posted cropped images from the two cameras at full resolution so that we could examine the images at 1 to 1 pixel representation on our own computers. Viewing of images at screen resolution of less than 1:1 can result in severe aliasing, which is easily seen when viewing Norman’s chart with Photoshop.

The most important conclusion that I reach from Michael’s tests is that perceived sharpness is more related to contrast and low noise than resolution. The 1Ds must give very good MTF at its relatively low resolution (Nyquist frequency of 57 lp/mm). Scientific analysis of image sharpness is very interesting, but the theory must be confirmed by observations such as Michaels.  

I wish I could afford the 1Ds, but I will have to stick to 4000 dpi scans of Provia or Velvia for the time being.

Bill Janes

Hope you can find time to read all this:

I did think that was an excellent article, you must have very similar objectives as myself. I use 6x9 format and would welcome a camera such as the Canon EOS-1Ds if it gives a superior image. However I really have trouble understanding the physics of this, not that I don't agree with your results. How can in just a few years, film, with all it's improved grain/sharpness that has gradually improved since the 80's suddenly be bettered 'over night' by digital sensor technology? I can't imagine how you can make a pixel smaller than a grain on film, or maybe this is not necessarily the case? Tends to make the film companies look a bit stupid.

Talking about stupid, how am I going to look now, with my trypod and huge camera set up on location, only to be bettered by some 'kid' next to me with the latest 'teckno toy'. I don't really have an argument in favour of 120 anymore. Of course exactly the same thing happened with the Reel to Reel tape recorder. A Revox was a 'God like' object of desire for it's quality of build and sound, only to be bettered by some black box with a 'CDR' label on the front. Now 'overnight' nobody has heared of a Revox. It's a shame this happens but it IS progress and in the case of the Canon EOS-1Ds  if the end result IS far better then my 120 cameras then there is no argument.

I take Landscapes myself and must say it is the skies that are always grainy/patchy/disappointing when scanned. This wasn't something I got with an enlarger for some reason ( I'm using the Minolta Scan Multi Pro). Digital is beautifully clean and noise free, and if this is a measure of quality by itself, a 3 or 5MP camera is better than 120 film. It's a bit like CD, who cares if the actual sound may not be a good as a vinyl LP, the lack of noise is such a quality in itself that it becomes the dominant quality. But the Canon seems to score on both counts, lack of noise and better resolution.

I find it fascinating that we are now looking at the limit of the lens resolution rather than the sensor being the limiting factor. Of course 135 always had the highest resolving lenses. 120 and 4x5 had respectively lower resolution as in theory it wasn't necessary to have the resolution of 135 as the enlargement factor would be less. I still think there will be an advantage in having a larger or higher MP sensor even if the lenses resolve the same. More pixels will improve the apparent sharpness as it improves the edge acutance. The contrast between light and dark will be improved. But what is the limit?

To my mind we almost need to determine what is an absolute limit the healthy eye can resolve at any distance to the paper, to establish an absolute resolution goal. To me you can't beat a contact print, and a 8x10 contact must be the ultimate print in terms of sharpness and quality. I can't believe you could better that or want to. I somehow don't think the new kid on the block, the Canon EOS-1Ds could produce a print this good? (or could it?) If it doesn't, we need to work on why it's not so good (including printer technology) with an aim to produce something that is as good as this reference.

I tend to print up to 24x16 and I have now just realised from your articles, that my goal will never be realised (or can it?). I would like to look up close at a 24x16 and see sharp petals on flowers in the foreground or basically sharp small detail (say sub 1mm). I now realise this is impossible to achieve as a lens could never be this good. I would like to see a 24x16 of  'contact' print quality.

How do we improve the Canon? I would like to see a rangefinder design. It would be smaller, have superior lenses with better distortion figures. Who needs an SLR now? The lenses need to be bigger and heavier and be compromised in design because of that mirror. Who needs an optical viewfinder anymore, where you have to squeeze your head up close to stare sideways into a hole, with your nose squashed against the back?

Nobody seems to have mentioned a huge digital advantage. Perfect 'film' flatness at last! This has cured the problem totally and must really help to improve the technical geometry/distortion figures especially if architectural photography is your thing.

Another BIG advantage, again not mentioned. A 135 camera has inherently far better depth of field for any given f stop than a larger format camera and that always will be the case. This is a quality which will always better medium format or larger format cameras (excluding movements). So from this point of view it is great to stick to 135 format or smaller. Chromatic aberration was always more of a problem for 'sub miniature' camera lenses and so was micro contrast. Reducing the image by so much onto such a small 'sensor' does bring it's physical problems. I would imagine most of these dilemmas have now been overcome by todays lens manufactures.

I would now like to see Foveon do a full frame sensor. Their design seems the way to go even though the Sigma SD9 has not had great reviews, early days still?

I see dirty sensors being a problem with all that lens swopping going on. Film at least provides a fresh clean 'sensor' with each shot. Also how long will a digital sensor last? 2 , 3 years? Pixel dropout? How about upgradeable sensors?

Anyway great stuff - Rich

Michael,

I thought the review was very interesting, as far as it went.  The only thing missing is comparison of larger print sizes.  The size of the 1Ds print was apparently 16 inches wide by 10.7 inches, assuming no cropping of the frame.  What about 12X18 (I know, the aspect ratio of the Pentax 67 would not allow printing the
whole photo on Epson 2200)? 16X24?  20X30?  The real question for many is, "At what print size (if any) does the Pentax 67 print show more detail than the 1Ds?"  I realize that you did the comparison based on print sizes you use, but it sure would be nice to have the comparison of large print sizes.

Michael,

Just wanted to thank you for the review, I greatly appreciate it.  Money's a little tight for a 1Ds right now, but as soon as it's practical, I will leave film behind entirely.  (Your review was not the only research I did, of course, but it was the final nail in the coffin.)

Joe

On [1] Pixels do not need to be as small as grains, since a single pixel conveys a lot more information than a single silver grain, which is genuinely either black or white, no shades of gray (one bit output!). For example, a medium size digital photosite (6.8 microns) can produce an output signal of 0 to 50,000 electrons with an error of about 20, giving about 2500 distinguishable levels of gray; several thousand silver halide crystals would be needed to give the same tonal gradations, based on what fraction of them get exposed to silver grains, and I have read that not than many crystals fit into a square of the size of that photosite.

On [2], there is a traditional guideline: the definition of 20/20 vision is resolving features down to an angular size of one minute of arc, or in other words, resolving details whose size is a bit less than 1/3000 of the viewing distance. So about 1/300 inch at 10 inches, or for readers outside of the US, 1/12mm at 25cm. Probably related to this is the resolution standard of about 12 pixels/mm required of digital files for input to some high quality commercial typesetting equipment.

Of course some people have better than 20/20 vision; maybe up to twice as good.

Robert queried the viability of electronic viewfinders in cold weather.  I've used EVF's of various kinds at minus 30 with no ill effects, other than my breath freezing on the VF : )

Seriously, I don't think cold weather is a problem.

There are many advantages to EVF's.  They're seldom too dark, unlike optical units.  I don't like what I see in the few digicams that use them, but they have a lot to offer if implemented well.

You see exactly what you're going to get.  Colour, noise, exposure are all WSISYG.  

You can have a wide variety of additional information on tap.  A continuous, onscreen histogram, for example.

Focus can be difficult.  Especially with critical DOF adjustments and wide angle lenses.

They also consume power.

Peter

I don't have a lot of experience with projection. I gave a seminar earlier this week and used a rented Epson 1200 lumen projector at 1280 PPI for about 50 people, and the images were very nice. But, not up to those from a high quality slide projector.

It's something I plan on exploring more closely in the months ahead, as I have a number of seminars coming up and I'm planning on buying my own projector at some point.

Michael

It struck me that there is a major flaw in the ultimate shootout: at 255 dpi with the 1Ds, that maxes out the resolution of the Epson 2200. Basically, there's no way to put more information on the paper. So, a priori, there's no way that MF will win.

(Also note that 255 dpi from a 1Ds is slightly smaller than 11x16. Were you are printing on standard A3 (11.75" x 16.5") paper with a small margin???)

It is pretty amazing that the 1Ds produces definitive photographic images at 11x14. I was never able to get decent 11x14s from 35mm (Plus-X) in the darkroom.

However, the claim that the 1Ds has more resolution than 6x7 seems seriously problematic.

If you printed a 4000 dpi scan of 6x7 at 450 dpi, you'd have about the same size image (about 18" in the short direction) as when you print a 1Ds image at 150 dpi. My experience is that 450 dpi printed film scans are a lot sharper with a lot more detail than 150 dpi printed digital images (including dowloaded 1Ds samples).

Try Calculations in PS

The last article of Michael " The Ultimate Shootout " is very interesting but

I remember when we were discussing about the dpi resolution for going to the printer the values which were mentioned were laying around 300dpi.

Now suddenly 255dpi ist enough and 1Ds is better than MF. Hmmmmm, interesting.

I have made trials and found out that even printing with 150dpi is enough so the 1Ds will manage without any problems prints in size of up to 18 x 27" without any problem.

I am still reluctant to get the 1Ds for Landscape due to the missing detail resolution.

Just my thoughts to this.

PD It is just so that I am little bit dissapointed how fast Michael is coming to those results. I always saw how he took some more time for this, but now he is very fast. Sorry  

David,

I much enjoy your posts on that plebian news group, RPD.

But I disagree with your statement that the Epson 2200's
resolution is 'maxed out'.

I thoroughly tested my Epson 1270, and I found that details
(the 'fineness' of lines, the saturation of colors in
tiny details) increased all the way up to ~600 ppi, when
examined on Epson Premium Glossy Photo Paper.  I examined
the prints with a 6x magnifying glass...

I don't pretend to know what algorithyms Epson uses in
it's printer driver... but I know what I saw in dozens
of test prints.

I printed 2240 x 1680 digital images, and drawn bitmaps
containing single-pixel wide features.

Michael,
Truly amazing results! Once again, it goes to show you really have to do the tests to get an appreciation of the significance of any quality differences between different systems. Quite often, as I've found myself when testing lenses on my D60, differences that one expects to be there are so trivial they're no worth bothering with.

One thing that does concern me a bit, the enlarged sections you've chosen for comparison have been taken from the very edge of the frame, and the short edge at that. I would not expect with such highly regarded lenses there would be any fall off at the edges, resolution wise, at F8 but I'm a bit worried there might be.

If that detail you've isolated had been in the centre of the frame, would we have perhaps seen greater differences?

Michael,
Perhaps I have only a limited understanding of sharpening and accutance and such, but I have a question about the 1Ds test images.  They seem to show more outlining around the high contrast edges than do the 67 images.  This would normally indicate to me that relatively more sharpening has been applied for the 1Ds (higher Radius setting in PS).  

This isn't a criticism, but more of a request for clarification.  Would the 1Ds images have shown as much resolution and detail if lower Radius (and/or Amount) settings had been used?  It's a critical question for me, as in my own work I am bothered by the outlining (or halos).

-Montelle

Michael, excellent article, as always. I very much like the way you compare prints, which, at the end of the day, is what matters. I was equally surprised when a friend preferred a D30 print to a Hassy-film one (both at 8x10"). My suspicion now is that a "nicer" print has much more to do with smoothness, tonality and gradation, rather than being "tack sharp". When we talk resolution numbers I am also driven to a similar conclusion. Excellent human vision, without a loupe, considers 5 lp/mm or 250 dpi as tack sharp in a hand-held print, or as sharp as it gets - hence all the stuff regarding printer resolution. The same cognoscenti then start talking of Depth of Field. DoF calculations are generally based on a Circle of Confusion of (if I recall correctly) either 0.25 mm (200 dpi) or 0.3 mm (only 150 dpi) in an 8x10" hand-held print. This means that these same people are willing to consider a great deal less than "tack sharp" resolution as "in focus"! Eureka! 250 dpi is at times an overkill, provided things stay smoooooth. No wonder, provided things are not pushed too far, most people would seem to prefer smoothness, tonality, gradation (the digital strong points) over a grainy, even if sharper image from a large negative. There will, of course, always be a print size beyond which the larger negative will win. If we refer to the CoC business one would peg the cross-over point at an enlargement that yields, say, only 150 dpi (for eg portraits) or 200 dpi (for eg very fussy landscapes, group shots, etc) in the print. That translates to the cross-over from a 1Ds to larger negatives/transparencies of any size that even hints at graininess at that enlargement at  18x27" for portraits and 13.5x20" for the extra fussy stuff. Quite in line with your findings. It's the CoC business that confuses things   "Tack sharp" is almost a theoretical concept when it comes to choosing the "nicer" print, IMHO anyway.

You are right that you don't need an images resolution to be so high all the time. The more the better, from an editing point of view, but for just printing, you can get away with less.

I worked for years in electronic prepress. The way we worked it, optimal file resolution doesn't have to do with the printer's resolution, but with the printer's screening (halftoning) technology. In the offset printing world, you usually print photos at 150 or 200 lines per inch (lpi, or linescreen). Linescreen refers to the density of the screen -- the higher the number, the finer the halftone. The imagesetters creating the film had resolutions of 2,000 or 3,000 or higher -- but this just meant that we could get finer and finer halftone screen.

If you had a perfect image (outputting at 100%, with no crazy contrasts or ramps of color), you could technically get away with its resolution being identical to the linescreen: 150 lpi only needs 150 dpi. But this was a perfect ideal that never really existed! Photos always had to be scaled or rotated, or th colors didn't hit the screens just-so. So we would always up the resolution to handle all the unpredicatable aspects of the real world. The 300 dpi rule of thumb comes from 2x the average linescreen of 150 -- these seemed to work the best, and handled most situations. It also gave you enough detail if you had to remove some dust or tweek the image a bit.

I have no idea what the screening technology of the Epson printers is, so I personally just stick with the 300 dpi as my target. If I can get it, awesome. If I have an image with more, I often downsample it to save storage space. If I have an image with less (as long as it is between 200 and 300), I leave it alone. I'll only upsample if my prints are giving me problems (jaggies, etc.)

all i really want to know is: why anyone would even bother with any format film, scan(drum or other..) if there are pro-backs out there (kodak pro 16mp and the new sinar 22mp) which should blow the 1Ds (and probably similar pro14n) completely away. yes they are more expensive, but we are at a level (5000-8000 for the cameras, or 100/drum scan) where 10000 for a back would be reasonable (i am talking pro shooters, catalogue can easily run 100+ rolls of 120 a day,so within a month the costs are justified) and there is still the option of shooting film and polaroids, because yes, many clients still want that, quality or not. personally i have always preferred the digital workflow and the control it offers....i would really like to see these latest 11-14mp cameras compared to digital medium format backs. the lens quality alone should make a huge difference in sharpness, crom.abr, detail...that should make 4x5 drum scans look bad! or not?

Well, if Michael's sources are correct you may see a Foveon full-frame sensor in about a month  :) ! He reported a rumor that a "major manufacturer" was going to introduce a DSLR with a full-frame 10MP Foeveon sensor at PMA in March.

Raghu

I agree about projection - I've projected some of my 645 images and they are indeed magical (technically speaking, rather than critiquing!) and you can wander up to the screen and keep seeing more detail.  I just loved to get lost in them.  If we take Michael's conclusion to mean that the 1Ds resolution and quality matches the 6x7 in most aspects, we can say that we need a projector with 11 MP to project what's contained in a 6x7 image.  I have an HDTV (Marquee 8500LC) which has 2500x2000 addressable pixels but the best it can actually do is a little below 1600x1200 and still keep the pixels sufficiently distinct to make a reasonable case.  There are some that actually do resolve close to 2500x2000 using 9" CRTs and superduper components &c &c (they are also a pain to colormanage, BTW) and they run around $30-60K and I'm sure in the stratosphere are even better units that likely cost 100s of 000s of $.
Toshiba has just released a rear projection TV (52 inch I believe) that has three DLP chips each of which is a 1920x1080 array (the maximum resolution in the ATSC format)  So we've got about 2 MP each of RGB which might be considered around 4MP when compared to a Bayer array.  This Toshiba unit is listed at $9k and is a bargain compared to other HDTVs.  The HD folks are wondering if/when a front projector with such a 3chip array will be released which would beat the pants off an InFocus or Epson unit, not just for resolution but for photographic images (the other two are for Powerpoint and such and don't do well with HD signals). While I think these units will do wonders for these new digital images, they won't have the detail capability that the big slides will; I do think , however, that they will still have a major wow factor nonetheless, given the quality of some of the best HDTV pictures I've seen.

Things are indeed converging but it will be quite some time before mere mortals could afford projectors that go significantly beyond the resolution of the Toshiba or similar units. Eagerly waiting a professional review of the unit and news of a front projector form.....

Andy

Marshall:

Wasn't that test on the Imaging Resource site?  I saw it as well.   Sure film is film no matter what size.  Surely the shadow detail advantage of the 1Ds would be there up against 4x5.  But again 4x5 would allow even larger prints up to say 40 x 50.  I'd hate to pay for a framing job on a print that large.

Whatever we may say Michaels test is quite an eye opener.

Here is the address for the D100/1Ds/4x5 comparison on outbackphoto.com.

http://www.outbackphoto.com/reviews....ds.html


This is the address for William Castleman's Mamiya 7/D60/1Ds/35mm review. Castleman's website seems to be down today.

http://www.wlcastleman.com/equip/reviews/Eos1Ds/index.htm


And finally, here are some remarks from Jim Collum, who did the 1Ds/4x5 comparison on outbackphoto.com, about medium format.

http://forums.dpreview.com/forums....3918312

Eric,

You have encapsulated my thoughts almost perfectly.

For those of you not aware of the book Chased by The Light, by Jim Brandenburg, I would suggest at least reading a synopsis of the idea and content. For 90 days Jim would only shoot one photograph. No second chances or additional exposures. The results, in my opinion, were breathtaking.

Daniel

Herein lies the problem with writing about the tests that I do. It's impossible to cover everyhting in an article. I do test shots at various apertures, make prints of the center and edges at each of several apertures, do the test with different lenses, etc etc.

But, there is only so much I can write and so many images that I can reproduce. It seems that no matter what I do there will be someone that says, "if only you'd done such and such, it would be better...". I do what I need to do to satisfy myself that my results are what I need to know. I report enough so that people can understand what it is I'm trying to accomplish. But I obviously can't satisfy everyone's interests and needs. That's why I suggest that people try and do these tests for themselves.

Don't trust what people write or say (including me). Don't trust the "theory" of the experts. Simply trust your own eyes.

Michael

Michael

I found looking at the sample imgages from the Kodak Pro Back, a 16mp medium format back, and comparing them to sample images from a better light scan back quite informative.

Both sets of sample images are on the Kodak website, the pro back samples being here:

http://www.kodak.com/global....4&lc=en

and the Betterlight scan back samples (not full size, but with mouseover enlargements) being here:

http://www.kodak.com/global....4&lc=en

I regard the "standard" Betterlight 6000 back (itself the current equivalent of the Dicomed Field Pro scan back that I own) to be roughly equivalent, or slightly better, in terms of resolution to large format drum scanned Velvia.  The Betterlight images are I think clearly superior (not surprisingly) to the 16mp Pro Back plus.  The Betterlight not only generates a larger file in absolute terms, but it is does not suffer the pixel averaging that goes on it a mosaic sensor such as the Pro Back (or 1Ds for that matter).  A mosaic sensor would probably need double the dpi of the scan back to match the scan back's quality - a very large file indeed.

So, what does this prove?  Not a lot, save my personal view that current 16mp backs are not as good yet as scan backs, and we still have a way to go to match 4x5 film.  The limiting factor in film performance is I believe partly scanner resolution and quality.

Quentin

Bill,
I think it's always been the advice on this site to set the image resolution at 240ppi or 360ppi, partly because these numbers divide nicely into 720 and 1440. I'm quite prepared to accept any conclusion that Norman Koren draws that anything in excess of 300ppi serves no purpose. If the native resolution of the image is greater than 300ppi (or 360?) then one should downsample to 300ppi. If the native resolution is lower than 240ppi, one should upsample to 240. Upsampling to a higher res than 240 serves no purpose if there's no actual detail there at that resolution. The purpose of the upsampling to that minimum of 240 is to produce a 'smoother' result with no hint of 'jaggies'.

I recently carried out a little experiment printing at Super A3 size a cropped D60 image. Without resampling, the image was 133ppi. Normally I would set the resolution at cropping to 240, but if I'm saving in GF I'll leave it at 133. I was curious as to how the two prints would compare, one at 133ppi and the other upsampled to 240ppi.

Well, big surprise! Without the aid of a loupe, I couldn't tell the difference between them. With the loupe, I could detect a faint line structure in certain parts of the 133ppi print and occasionally jagged edges in other parts which weren't there in the 240ppi print.. I think it would take someone with exceptionally good eyesight to choose between these two prints on technical grounds if they were hanging on a wall side by side.

It seems to me, when one considers all the post processing in Photoshop that one often does, changing levels, selective sharpening, hue, saturation, colour and contrast etc etc., then the choice of ppi for printing (whether it's 150, 180, 200, 240 or 300ppi) is the LEAST significant of all the changes one is likely to make. Having said that, don't misunderstand me. I'm not advocating downsampling an image with a native resolution of 300ppi to 150ppi because it doesn't matter.

I would now like to see Foveon do a full frame sensor. Their design seems the way to go even though the Sigma SD9 has not had great reviews, early days still?

Rich,

I have a SD9. The Foveon sensor meets its goal in terms of sharpness and its ability to resolve detail. It's a 3.54MP sensor that captures as much detail as a 6MP sensor. The only thing 'wrong' with the chip is that it works best at ISO 100 and has limited long exposure capability due to noise. The image quality itself has never been questioned. Popular Photography (March) has a review of the camera. Shutterbug as well. You might want to check them out. In the next year or so we'll know if Foveon can increase the size of its sensor. By the way, if you're interested, I can e-mail you some of my SD9 images, if you want to have a look at them in Photoshop. Let me know. Actually, they are all available at www.pbase.com/chromelight

Feel free to download them. No sharpening or post processing was done. They aren't great photos. I was just tesing the camera.

To Doug:

You may be right. It might have been from the Imaging resource site. It was from Rob Galbraith's forum at least that I read about that test. It may have been a link to the IR site that I saw at Rob's site.

To RD:

Thanks for those links again. I remember reading the articles at a couple of those sites. I see a pattern here in the test results.

All of my evaluations are done on unsharpened images. I sharpen them for web display because otherwise they are too mushy. It's a lot of work to get the USM right for the web and sometimes when the job is done it's too much on one and too little on another. You should take my written descriptions as being more accurate than the small files that I prepare for the web.

People who pursue photography as an art might generally want prints of higher quality than your typical catalog.  If all you do is commercial work, who cares?  I'd shoot with a 1Ds for the convenience of a 35mm system.  If the client wanted shots from a 22MP Sinar back, they'd get that -- for a price.  I'd rent one to do the work, and if I ended up renting it often enough I'd lease or buy it.  If they wanted me to dig a hole and paid for it, I'd do that too.  Who cares?  It's a business.

I'd love to test the 1Ds against 4x5 but part of the reason I am even using a 4x5 camera is the price ot the 1Ds.  The 1Ds body alone costs more than my Arca Swiss Metric Field 4x5 and three lenses put together.  I just can't afford one.

Dear Michael,
I fully agree with your saying that digital is better then film. As a matter of fact I have been trying to convince my peers of that since I had my D1x more than a year ago.
However, I would suggest that before you conclude that MF has been bettered by 1Ds try a comparison between files from 1Ds and a MF digital back.
I happen to have a Contax 645 and a Kodak DCS645C and I think the prints I got from it definitely have a different qaulity as compared to those from D1x( sorry I have no 1Ds to compare with but now with the D1x I can get 10 mp files and judging from the 1Ds images I've seen on the web, D1x is not that far off).While 35 images are always flat a MF image always pops out at you and you can almost feel the texture.I think photograhy is not just about resolution ( the resolution of 1Ds is indeed the same as 645C after cropping ), just like in HiFi, if only the frequency response matters then all systems will sound the same.
I am not trying to say that my 645c is better than your 1Ds but I wish you would make a comparison before proclaiming that MF is past.
CMMa

Michael,

Great stuff, as always.  You say goodbye to medium format at the end of the article - are we to take that literally, or will you still use medium format?

Regarding 4x5, I don't have a 1Ds, but I have downloaded and even printed a couple of tests shots others have done, and I agree they are impressive.   My own view (as someone who drum scans his own 4x5 Velvia) is that 4x5 still has a reasonable quality lead, similar to the lead it has over, say 67.  A LF tansparency, drum scanned, still has a richness of tonality, and rendition of fine detail not yet reached by a DSLR like the 1Ds. But it's still a stonking good peformance by the 1Ds.

I also need to be slowed down, and using LF or Medium format does this.  I feel a bit silly mounting my Fuji S2 on a tripod in comparison.  

Quentin

Rich,

Many good points and the one that interested me the most was why we still need/want D-SLR's. I imagine that with the current bodies, AF was easiest to achieve by keeping the mirror and the AF mini-CCD that is used for focusing only. It would be nice to find a way to get rid of that.

EVF's are improving and though they may never be the same as an optical viewfinder they may be good enough. As always, I wonder about cold weather performance as I live in Canada.

In P&S digicams, my understanding is that the main CCD is used to set focus first and then the "shutter" is fired (electronically). Thus, there is an upper bound  on the speed of AF, that may be a burden to those to whom this matters. On the other hand the Contax G2 is an AF rangefinder (no mirror) so there is a way to accomodate AF and get rid of the mirror and prism.

I would be happy with a MF stripped down K-mount digicam with 35mm sized sensor for the DOF reasons you mentioned. I rarely need speed of picture-taking, as is true of probably many others. To do MF you need either a rangefinder body or SLR with ground glass focusing. Relying on the LCD for visual focusing is iffy. Of course, maybe I need to join the modern era and simply accept AF but I don't want to buy new lenses.

It will be interesting to see what variations on the theme appear in the next few years.

One advantage of EVF's I forgot to mention, and it's a biggie, IMHO.

They're orientable.  No more back-breaking stooping to peer through a keyhole on the back of your ten-thousand-dollar DSLR.

Peter