Doug, I quote Michael's review of the 1Ds:
"What about a comparison with both 35mm film and medium format? I'm afraid that film has definitively lost the battle. The 1Ds's full-frame 11MP CMOS sensor produces a 32MB file — as big as a typical scan. But this file is sharper and more noise free than any scan I have ever seen, including drum scans. There simply isn't a contest any longer.
The 1Ds also fares very well against medium format. Is it sharper than 645? No, not quite, but really very close. When you add in the extremely low noise of the images compared to scanned film, and add in all of the cost and workflow advantages of shooting digital over shooting and scanning film, in my opinion the 1Ds is to be preferred. I'll gladly take the huge reduction in noise (grain) over slightly lower overall resolution any day of the week. Thomas and Chris basically have concurred with this finding after reviewing many sample images that we shot together."
This says that OVERALL, MR prefers the images from the 1Ds.
Shoot what you want, Doug, but don't put words in Michael's mouth.
When YOU have tested a 1Ds, comparing it to your own MF images, and have published your results, you'll be entitled to your INFORMED opinions and I'll read them. 'Till then, how can you argue INTELLIGENTLY without having done it yourself? One surely can't tell from the medium-sized JPGs anyone including Michael published.
Full Version: Michael's Tests of EOS-1Ds
One feature Michael wished for was automatic horizontal and verticle orientation of the file. Wow, I've been saying that for years. My Kodak 260 (4years ago? about $400) had this feature. I've since owned Nikon 990, 995, Canon D30, D60, A30 and not one of these #### cameras has had it! :p You may think it a small problem but when you have to reorient a few hundred photo's especially at converted file sizes, you can waste a whole lot of time.
Done ranting.
Thanks.
Done ranting.
Thanks.
Hi Dale,
I do not know how you came to 16,72 Mio Pixel but mathematically
36mm divided by 0.006 (microns) gives exactly 6000 pixels on the long end of the 35mm film and 4000 pixels on the short end and this multiplied gives 24 Million Pixels.
Another thing is you would then have optically a resolution of ~ 4233 pixels. More than any film scanner at the momemnt speaking of the Polaroid, Canon or Nikon film scanners.
I do not know how you came to 16,72 Mio Pixel but mathematically
36mm divided by 0.006 (microns) gives exactly 6000 pixels on the long end of the 35mm film and 4000 pixels on the short end and this multiplied gives 24 Million Pixels.
Another thing is you would then have optically a resolution of ~ 4233 pixels. More than any film scanner at the momemnt speaking of the Polaroid, Canon or Nikon film scanners.
Since current sensor pixel pitch appears to be near the limit of the resolving power of our best lenses, it might make sense for the camera designers to turn to exposure latitude for the next set of improvements. Certainly, this seems to be the last factor where film holds the high card.
To increase dynamic range, you need to capture and process more data, resulting in longer processing times and bigger files - not always desireable.
Why not give this control to the photographer? If they provide us with an on-camera "contrast" adjustment we could increase bit depth when we need it.
To increase dynamic range, you need to capture and process more data, resulting in longer processing times and bigger files - not always desireable.
Why not give this control to the photographer? If they provide us with an on-camera "contrast" adjustment we could increase bit depth when we need it.
Similarly, I received notice yesterday afternoon that the 1Ds would be available to me (I'm at the top of the waiting list) in early December, for $7400.
This is about $1400 higher than I think it should be.
Have others received similar notice?
Enjoy!
-- Jim
This is about $1400 higher than I think it should be.
Have others received similar notice?
Enjoy!
-- Jim
I have to agree with Marshall. I'd like to see a comparison made with high detail, high frequency landscape images, expecially with--if possible--prints made at, say, 14x22 or 16x20 and higher, which is traditionally what I've bothered to use medium format for anyway. If I enjoyed smaller prints, I'd be really excited.
On an aside, has anyone here ever made any comparisons projecting digital files from a high end multi-media projector vs. a quality 35mm or medium format slide projector and lens? For my personal use, the 'weak link' in digital is the lack of information regarding the satisfaction one might or might not receive from using the current crop of digital projectors.
On an aside, has anyone here ever made any comparisons projecting digital files from a high end multi-media projector vs. a quality 35mm or medium format slide projector and lens? For my personal use, the 'weak link' in digital is the lack of information regarding the satisfaction one might or might not receive from using the current crop of digital projectors.
Awesome. :D A word not to be used lightly, but I feel justified in using it for the 1Ds.
It's not just the resolution, although that is impressive. Its the lack of serious artifacts. Its Foveon like (arrgh, the dreaded F word). I asume this is because of the light touch anti-aliasing filter.
Assuming this to be the case, I would hope (as a Nikon man) that the new Kodak will be similarly free of nasties.
An important threshold has been crossed. A DSLR with resolution surpassing that of 35mm film. We are in unchartered territory. Someone better start redesigning 35mm lenses to keep up.
Quentin
It's not just the resolution, although that is impressive. Its the lack of serious artifacts. Its Foveon like (arrgh, the dreaded F word). I asume this is because of the light touch anti-aliasing filter.
Assuming this to be the case, I would hope (as a Nikon man) that the new Kodak will be similarly free of nasties.
An important threshold has been crossed. A DSLR with resolution surpassing that of 35mm film. We are in unchartered territory. Someone better start redesigning 35mm lenses to keep up.
Quentin
Jim,
This is the whole point of my argument. It's difficult to get the facts. It's a weakness that tends to be ignored. You must know from general experience that you don't promote products by mentioning weaknesses. If a manufacturer produces an audio CD player with a dynamic range of, say, 70db, he's not going to boast about it. The listed specifications at the back of the manual are probably going to ignore this bit of information. It's not impressive. Let's be quiet about it. So is the case with digicams in general. Dynamic Range is an ignored specification. There's nothing to boast about.
When Michael first reviewed the D30, he said he thought there was slightly more shadow detail than the compared Provia F slide. There were later subjective comments, from other sources as well, that the dynamic range of the D30 was somewhere between that of slide film and that of negative (color) film. Pretty vague really. My own experience with the D60 is that the DR is very roughly inbetween that of slide film and that of color film, but probably closer to that of slide film. I would guess it's about 6 f stops as opposed to the 5 or 5 1/2 f stops of the average slide film. This compares with 8-9 f stops for some color negative films and 11 f stops for the best of the B&W films and the best of the MF $20,000 digital backs. There's a long way to go.
This is the whole point of my argument. It's difficult to get the facts. It's a weakness that tends to be ignored. You must know from general experience that you don't promote products by mentioning weaknesses. If a manufacturer produces an audio CD player with a dynamic range of, say, 70db, he's not going to boast about it. The listed specifications at the back of the manual are probably going to ignore this bit of information. It's not impressive. Let's be quiet about it. So is the case with digicams in general. Dynamic Range is an ignored specification. There's nothing to boast about.
When Michael first reviewed the D30, he said he thought there was slightly more shadow detail than the compared Provia F slide. There were later subjective comments, from other sources as well, that the dynamic range of the D30 was somewhere between that of slide film and that of negative (color) film. Pretty vague really. My own experience with the D60 is that the DR is very roughly inbetween that of slide film and that of color film, but probably closer to that of slide film. I would guess it's about 6 f stops as opposed to the 5 or 5 1/2 f stops of the average slide film. This compares with 8-9 f stops for some color negative films and 11 f stops for the best of the B&W films and the best of the MF $20,000 digital backs. There's a long way to go.
Hi Jim,
I do not agree. When you have something to throw away, it is always better rather then making up or inventing something.
When I do prints whatever size, I always scan at highest possible resolution and then either resize up or down.
Just because of this huge discussion in dpreview, on weekend I made some tests to see if the digital really has more detail and for my use I would say digital can still wait.
I took a landscape shot with my Canon 15mm f2.8 Fisheye on Provia 100F (I think not the best example, but i wanted to see what comes out on details) and scanned it at 8000dpi on my Polaroid SprintScan4000 with Silverfast Ai 6.0. Yes 8000 dpi the software interpolates, because the optical resolution of the scanner is 4000dpi.
I do print from 20x30cm up to 60x90cm or 8"x12" to 24"x36".
The bigger you print, the lesser the resolution of the outgoing file can be because the looking distance increases.
Take the huge advertising signs on the Mega-Signposts. Each pixel maybe has the size of 1x1inch square but you see a perfect image because you are 100meters away.
I even do this for my photos in the web. Scan big and then downsize to max. 600 pixels longest side.
So I decided long ago that a ~20' MioPixel DSLR would be the point where I step into Digital Photography until then I begin to save the money for it.
I do not agree. When you have something to throw away, it is always better rather then making up or inventing something.
When I do prints whatever size, I always scan at highest possible resolution and then either resize up or down.
Just because of this huge discussion in dpreview, on weekend I made some tests to see if the digital really has more detail and for my use I would say digital can still wait.
I took a landscape shot with my Canon 15mm f2.8 Fisheye on Provia 100F (I think not the best example, but i wanted to see what comes out on details) and scanned it at 8000dpi on my Polaroid SprintScan4000 with Silverfast Ai 6.0. Yes 8000 dpi the software interpolates, because the optical resolution of the scanner is 4000dpi.
I do print from 20x30cm up to 60x90cm or 8"x12" to 24"x36".
The bigger you print, the lesser the resolution of the outgoing file can be because the looking distance increases.
Take the huge advertising signs on the Mega-Signposts. Each pixel maybe has the size of 1x1inch square but you see a perfect image because you are 100meters away.
I even do this for my photos in the web. Scan big and then downsize to max. 600 pixels longest side.
So I decided long ago that a ~20' MioPixel DSLR would be the point where I step into Digital Photography until then I begin to save the money for it.
I downloaded a couple of the SD9 pics as I did with the 1Ds last week. I haven't had the time yet to work with the SD9 samples, but I will say upon first look, that the color is very good in all of them. The blue sky in the SD9 sample looks more saturated than in any of the others. But again, the color in all the SD9 shots is good.
I haven't applied any sharpening to the SD9 pics yet, but the in-camera sharpness, even before USM is very good. The images look clean and clear and pretty sharp already. As good as most I've seen before USM or NS Pro.
I'm not saying any of this because I'm getting excited about the Sigma/Foveon; the 1.7X magnification is completely unacceptable to me and therefore I won't be buying it. But I will give credit where credit is due.
As for day to day use of the camera, that will have to wait until test reports. How good or accurate is the metering, AE, AF, Ergonomics, Reliability? All those are unknowns so far. Doesn't matter, I'm a Nikon man. But based on the image quality of the samples I've seen from Phil and Kumio from the SD9, I do think the X3 is a very good chip. If Foveon could just make a much bigger version of it and provide it to Nikon.
One thing I noticed about the Foveon, like the 1Ds CMOS and Nikon CCDs, you can still easily see dust spots. At least when they're stopped down to f11 or further.
I haven't applied any sharpening to the SD9 pics yet, but the in-camera sharpness, even before USM is very good. The images look clean and clear and pretty sharp already. As good as most I've seen before USM or NS Pro.
I'm not saying any of this because I'm getting excited about the Sigma/Foveon; the 1.7X magnification is completely unacceptable to me and therefore I won't be buying it. But I will give credit where credit is due.
As for day to day use of the camera, that will have to wait until test reports. How good or accurate is the metering, AE, AF, Ergonomics, Reliability? All those are unknowns so far. Doesn't matter, I'm a Nikon man. But based on the image quality of the samples I've seen from Phil and Kumio from the SD9, I do think the X3 is a very good chip. If Foveon could just make a much bigger version of it and provide it to Nikon.
One thing I noticed about the Foveon, like the 1Ds CMOS and Nikon CCDs, you can still easily see dust spots. At least when they're stopped down to f11 or further.
I wonder if these new DSLRs as the new Canon and Kodak will enter into the medium format quality, at least for medium sized prints. If thats true, why bother carrying around a heavy film based 6x7?
What do you think?
Sergio :
:
What do you think?
Sergio :
:
Michael:
I don't know if you'll see this, but your time exposure photo of the Milky Way over the lake at dusk with the 1Ds was fabulous!! Gorgeous picture had it been taken with any camera. There are probably some astronomy clubs near you that would enjoy that picture and it may also have stock potential.
It would be interesting to print a 13X19 of it not only for the enjoyment of it, but also to see how much if any noise it shows from the long exposure. Thank God and Canon for CMOS!
I don't know if you'll see this, but your time exposure photo of the Milky Way over the lake at dusk with the 1Ds was fabulous!! Gorgeous picture had it been taken with any camera. There are probably some astronomy clubs near you that would enjoy that picture and it may also have stock potential.
It would be interesting to print a 13X19 of it not only for the enjoyment of it, but also to see how much if any noise it shows from the long exposure. Thank God and Canon for CMOS!
QUOTE (Guest @ Oct. 07 2002,06:02)
What's more, it is almost certainly an engineered inhibition. CCD sensors can now achieve 14 stops range, and do so completely linearly. I don't know any reason why CMOS should be worse, so it sounds like Canon have knobbled the range.
Tony Sleep Photography tonysleep@halftone.co.uk ::
Tony Sleep Photography tonysleep@halftone.co.uk :
:Tony,
It never occurred to me that the limited dynamic range of the D60 and 1Ds (and other DSLRs) might be a deliberate engineering limitation to make the image look more 'punchy'. I tended to assume that the limitation was raw computing power. If you want high dynamic range, you not only need the sensors to be capable of it, but perhaps more computing power than can be fitted into a 35mm size body without slowing down the whole process. Typical MF backs which, until recently, needed to be tethered to a laptop at least, were capable of 11 f/stops. The 35mm format has to be more flexible.
Dale,
I gathered you are an 'arty type' from your excellent introduction to composition. Nevertheless, you seem to have a good understanding of the technical issues as well. The notion that, because 35mm lenses cannot resolve finer than 6 microns (whatever circumstances of f/stop and contrast ratio etc apply!!), therefore any digital sensor with pixels of a similar dimension has reached the limit of 35mm lens technology, is clearly false. It's a fallacy.
If a lens is capable of resolving a particle 6 microns in diameter, you need a digital sensor containing pixels considerably smaller than 6 microns in order to 'extract' the most out of the lens.
There's a 'law of diminishing returns' that applies, as with scanners. 2700 dpi gets you most of the detail available on 35mm. 4000 dpi gets you a bit more. 6000 dpi gets you an even smaller bit more, as does 8000 and 12000.
To get the relativities in perspective, if the 3MP D30 had been designed as a full frame digicam with no multiplier effect, it would have been a 7.8MP camera (assuming the pixel density was not diluted). The 1Ds at 11.2MP does not have a significantly greater pixel density than the D30, and has less than the D60. I think we can safely assume that there's still a fair amount of improvement to be achieved before we reach the limit of 35mm lenses.
QUOTE (RS Foto @ Oct. 11 2002,13:29)
Why do we assume that the resolving power of the lenses is limited to XYZ microns?
How have we measured this?
How have we measured this?
Rainer,
Good question. And I certainly don't have the answer. I've been running with this 6 micron limit because it came from an Olypus representative who was promoting another format. I figured it would be a very conservative figure and close to real world results as opposed to laboratory tests with high contrast ratios which, I believe, would show resolution limits far smaller than 6 microns.
I would surmise (as you've gathered, I'm no expert - just fishing in the dark for the few facts available) that Kodak have omitted the anti-aliasing filter because aliasing and other artifacts are reduced as the resolution of the sensor reaches and exceeds the resolution of the lens. I would predict that the 14n will be relatively free of artifacts at F/8 and greater, but not with a good lens at large apertures. This is one advantage of using sensors that exceed the resolution of the lens, and AA filters as you know reduce ultimate resolution somewhat and therefore are best avoided if possible.
Hi Michael,
Thanks for the first really independent test of this camera. Also for the shots.
While I have read with much interest everything the comparison of the 1Ds with the provia 100F film what I did not like is that you presented a 10% bigger sized scan of the provia film. For the 645 you reduced it to a similar size of the 1Ds but the scanned crop of the 100F 35mm you left it on 553pixels wide.
I allowed myself for comparison purposes to set all 3 images in one row for comparison puposes.
http://rainerehlert.com/fotos/1Ds.jpg
After seeing the 3 shots together my decision to buy a 1Ds has never been nearer than now.
I will keep attention to you further test and if it is possible it would be nice if you could later on provide some RAW files to experiment with them. Will it be possible to open those RAW files with Photoshop 7.0.1?
Thanks for the first really independent test of this camera. Also for the shots.
While I have read with much interest everything the comparison of the 1Ds with the provia 100F film what I did not like is that you presented a 10% bigger sized scan of the provia film. For the 645 you reduced it to a similar size of the 1Ds but the scanned crop of the 100F 35mm you left it on 553pixels wide.
I allowed myself for comparison purposes to set all 3 images in one row for comparison puposes.
http://rainerehlert.com/fotos/1Ds.jpg
After seeing the 3 shots together my decision to buy a 1Ds has never been nearer than now.
I will keep attention to you further test and if it is possible it would be nice if you could later on provide some RAW files to experiment with them. Will it be possible to open those RAW files with Photoshop 7.0.1?
Michael:
I came to this site late Friday evening, actually early Saturday to read the next installment of your 1Ds diary and didn't see any updates. I was disappointed. But I figured you were probably tired after a full day of shooting with the new camera and maybe printing some 13X19s. Right? I'll look forward to the next update later today.
I came to this site late Friday evening, actually early Saturday to read the next installment of your 1Ds diary and didn't see any updates. I was disappointed. But I figured you were probably tired after a full day of shooting with the new camera and maybe printing some 13X19s. Right? I'll look forward to the next update later today.
Ray,
Digital cameras respond differently with regard to dynamic range. My subjective impression is that their dynamic range is somewhere between negs and good transparency film *but* with one caveat: easily burnt out highlights. You have to err on the side of underexposure. That means shooting Raw. Seeing a burnt highlight, it would be easy to conclude, I think wrongly, there was a problem with dynamic range.
There is another factor here. The scanners a lot of people use to scan their transparencies have very limited dynamic range. Forget the marketing nonesense that would have you believe xyz brand of $200 flatbed has a dynamic range of 4.2, or whatever. I read some tests recently that showed even some quite expensive and well regarded scanners had limited dynamic range. So, in comparison, a DSLR can really win out. Film is, if you like, being held back.
Finally, most DSLRs are 12 bit. That's pretty good, but 14 or 16 bit would be better.
Quentin
Digital cameras respond differently with regard to dynamic range. My subjective impression is that their dynamic range is somewhere between negs and good transparency film *but* with one caveat: easily burnt out highlights. You have to err on the side of underexposure. That means shooting Raw. Seeing a burnt highlight, it would be easy to conclude, I think wrongly, there was a problem with dynamic range.
There is another factor here. The scanners a lot of people use to scan their transparencies have very limited dynamic range. Forget the marketing nonesense that would have you believe xyz brand of $200 flatbed has a dynamic range of 4.2, or whatever. I read some tests recently that showed even some quite expensive and well regarded scanners had limited dynamic range. So, in comparison, a DSLR can really win out. Film is, if you like, being held back.
Finally, most DSLRs are 12 bit. That's pretty good, but 14 or 16 bit would be better.
Quentin
There have been recent discussions about the usefulness of 11 MPs if one doesn't care about making large prints.
I contend that more pixels are only better if an image must be "ressed-up" to print at the desired PPI. For instance, if printing a 4 x 6 inch print at 360 PPI, a 3 MP D30 image has sufficient pixels to avoid interpolation altogether.
But printing that same image at 12 x 18 inches requires significant up-ressing using Genuine Fractals, bicubic interpolation, or some other process.
If one were to compare an 8 x 12 inch print using a 3 MP image and an 11 MP image, I would expect visible differences - assuming the image contained fine details like the texture of distant foliage.
I wonder if Michael could conduct a test - or make a full-size 1Ds image available for others to do the test.
One could print a full-size 1Ds image at 360 PPI. Then down-sample the image to D30 size (~3 MP). Then use Genuine Fractals to up-res the image to its original size, and print again at 360 PPI. I'd like to compare the fine details between the two prints.
Has anyone done such a test?
Enjoy!
-- Jim
I contend that more pixels are only better if an image must be "ressed-up" to print at the desired PPI. For instance, if printing a 4 x 6 inch print at 360 PPI, a 3 MP D30 image has sufficient pixels to avoid interpolation altogether.
But printing that same image at 12 x 18 inches requires significant up-ressing using Genuine Fractals, bicubic interpolation, or some other process.
If one were to compare an 8 x 12 inch print using a 3 MP image and an 11 MP image, I would expect visible differences - assuming the image contained fine details like the texture of distant foliage.
I wonder if Michael could conduct a test - or make a full-size 1Ds image available for others to do the test.
One could print a full-size 1Ds image at 360 PPI. Then down-sample the image to D30 size (~3 MP). Then use Genuine Fractals to up-res the image to its original size, and print again at 360 PPI. I'd like to compare the fine details between the two prints.
Has anyone done such a test?
Enjoy!
-- Jim
There's one thing that worries me about these comparisons. Most of the images are taken with different lenses, are they not? Now I know if one is comparing a Canon DSLR with a Nikon or SD9, there's basically no alternative. The lens mounts are different. But what about comparisons between, say, the D60 and 1Ds. We still use different lenses to compensate for the D60's 1.6x factor. The Canon 100-400 IS Zoom set at 400mm or 160mm does not have the same performance (necessarily) as it does at 250mm or 100mm. Jim mentioned downsampling a 33MB 1Ds file to 9MB to put it on the same footing as the D30 and then upsampling both files to search for differences. I'm not sure if such an experiment is going to be at all useful. One would expect any differences, if any, to be very subtle. But if you did see any differences, with a magnifying glass, how would you know they were not differences in the lenses used?
On the issue of converting digicam pixels to line pairs/mm, I've assumed that a camera pixel is roughly equivalent to a scanning dot and that it takes a little more than 2 dots to depict the narrowest line pair.
On the issue of converting digicam pixels to line pairs/mm, I've assumed that a camera pixel is roughly equivalent to a scanning dot and that it takes a little more than 2 dots to depict the narrowest line pair.
Sergio, Michael too is wondering the very same thing.
In Part 5 of the 1Ds review, Michael writes, "the 1Ds is a much more sophisticated camera than the Nikon-based 14n." Is there a fuller discussion of this comparison somewhere? I wanted to know more. Thanks.
QUOTE (Ray @ Oct. 09 2002,20:44)
The 1Ds at 11.2MP does not have a significantly greater pixel density than the D30, and has less than the D60.
Ray,
According to dpreview, the pixel "pitch" (in µm) of the 1Ds is 8.8 x 8.8 which is roughly halfway between the D30 (9.9 x 9.9) and the D60 (7.4 x 7.4).
Hence, the 1Ds pixels are more densely packed (smaller) than those of the D30, by about the same amount as the 1Ds pixels are less densely packed (larger) than those of the D60.
You can see the comparison table here.
Enjoy!
-- Jim
Hi,
I have been following this thread.
Why do we assume that the resolving power of the lenses is limited to XYZ microns?
How have we measured this?
Could it not be that our measuring devices are not up to date?
That we have had no necessity to improve our measuring devices?
Just putting down what came to my mind about this theme.
I have been following this thread.
Why do we assume that the resolving power of the lenses is limited to XYZ microns?
How have we measured this?
Could it not be that our measuring devices are not up to date?
That we have had no necessity to improve our measuring devices?
Just putting down what came to my mind about this theme.
When Michael goes to the Photo Expo, or any of you for that matter, I hope he'll(you'll) go to the Kodak booth and see what he(or you) can learn about the 14n and any sample prints or display posters they might have. And also ask about the AA filter issue.
QUOTE (Doug_Dolde @ Oct. 29 2002,01:11)
In fact his projected Normalized Image Quality published on this website shows the Canon 1Ds at only 1.03 which seems a bit low based on Michael's real world tests.
Yes! Koren's IQ figure for the 1Ds does seem a bit conservative. I'm reminded of some good advice I heard recently. We should strive to be emotionally optimistic but intellectually pessimistic.
Reading Norman Koren's tutorial on scanners recently, I came across the following statement.
"Given the difficulty that optical systems have in reconstructing the original analog signal, the rule of thumb that the number of line pairs per mm a scanner can resolve is less than half its pixel (or dots) per mm resolution is essentially correct. If the Nyquist theorem could be applied precisely (using sinc(x*dscan) reconstruction), it would be very close to half. But in reality you need more than two pixels to resolve one line pair of actual image detail (for example, a window screen). I've seen the number four used as a rough estimate, but that seems overly pessimistic in light of the results below. Three seems to be a pretty good number, especially when sharpening is applied."
The above paragraph caught my attention because I had previously arrived at a position where it seemed to me there could be a 'worst case' scenario where 4 pixels were required to depict one line pair. Samirkharusi pointed out that only 2 pixels are required. It now seems that 4 pixels is the pessimistic extreme and 2 pixels is the optimistic extreme. 3 pixels is just about right.
Now I might be making an erroneous assumption if I equate scanner resolution with digital sensor resolution, but on the basis that the same principles apply and that it takes 3 camera pixels (on average) to depict one line pair, we get the following results. Maximum resolution for the 1Ds (at MTF 50% ?) is 38 lp/mm and for the D60, 46 lp/mm.
Now, if you think I don't know what I'm talking about, do me a favour and let me know, won't you?
While I understand folks feelings about the large price of the 1Ds, a quick calculation shows me that film, processing and drum scans are costing me about US$2,500/year, not counting the driving involved in all of those. Even mitigating this a bit for batteries and memory sticks, the 1Ds might just pay for itself for me.
Michael,
You mentioned that some people have criticised you for your method of doing reviews and tests, for using postprocessed images, or not being scientific enough, etc. Well too bad for them! I wanted to let you know that I very much appreciate your style of reviewing and testing. We already have two excellent sites (dpreview and imaging-resources) which try to go into exhausive objective detail about new technology. What I appreciate and learn from your website is YOUR take on a camera (or technique) based on YOUR subjective opinion. It's refreshing and helpful to see how things relate to your professional experience. Please, keep doing what you are doing!
You mentioned that some people have criticised you for your method of doing reviews and tests, for using postprocessed images, or not being scientific enough, etc. Well too bad for them! I wanted to let you know that I very much appreciate your style of reviewing and testing. We already have two excellent sites (dpreview and imaging-resources) which try to go into exhausive objective detail about new technology. What I appreciate and learn from your website is YOUR take on a camera (or technique) based on YOUR subjective opinion. It's refreshing and helpful to see how things relate to your professional experience. Please, keep doing what you are doing!
Jim Warthman writes:
Canon has proudly proclaimed that the 1Ds will at least equal—and in some cases perhaps even surpass—professional slide film. It stands to reason, then, that if Canon believed the camera were capable of the broad latitude provided by print film, it would have stated as much. It didn't. And as far as I know neither has any other digital camera manufacturer. Yet.
All of the speculation aside, however, it's an easy thing to test with one roll of slide film, one roll of print film, and one digital camera. My own tests show that CCD digicams and DSLRs provide less latitude than slide film, with almost no tolerance for overexposure. I have not had the opportunity to test a CMOS camera but plan on doing so soon. Nevertheless, I doubt that that test will conflict with Canon’s announcement proclaiming the 1Ds equal to professional slide film—not print film. And at the end of the day, that’s an improvement.
Will Tompkins
QUOTE
I think it does everyone a disservice to propogate the subjective opinions that DSLRs have a smaller dynamic range than (negative) film.
Canon has proudly proclaimed that the 1Ds will at least equal—and in some cases perhaps even surpass—professional slide film. It stands to reason, then, that if Canon believed the camera were capable of the broad latitude provided by print film, it would have stated as much. It didn't. And as far as I know neither has any other digital camera manufacturer. Yet.
All of the speculation aside, however, it's an easy thing to test with one roll of slide film, one roll of print film, and one digital camera. My own tests show that CCD digicams and DSLRs provide less latitude than slide film, with almost no tolerance for overexposure. I have not had the opportunity to test a CMOS camera but plan on doing so soon. Nevertheless, I doubt that that test will conflict with Canon’s announcement proclaiming the 1Ds equal to professional slide film—not print film. And at the end of the day, that’s an improvement.
Will Tompkins
Hi Jim,
In the page www.digitalcamera.jp
http://www.digitalcamera.jp/report/EOS-1Ds-020924/index.htm
http://www.digitalcamera.jp/report/SD9-020927/index.htm
On this 2 pages you find the same motif taken with the 1Ds ~11' MioPixel as well as with the SD9 ~3' MioPixel.
I downloaded the 2 files ISO 100 from the page and made a print test 12x18 inches and you really see a difference. I resized both to have 300dpi for printing on an Epson Stylus Pro 7500 with 1440dpi on the printer.
The difference in sharpness is clearly visible.
My wife due to the fact that she does not do any photography is always called to check the prints.
Try this 2 files. It is very interesting and eye opening.
In the page www.digitalcamera.jp
http://www.digitalcamera.jp/report/EOS-1Ds-020924/index.htm
http://www.digitalcamera.jp/report/SD9-020927/index.htm
On this 2 pages you find the same motif taken with the 1Ds ~11' MioPixel as well as with the SD9 ~3' MioPixel.
I downloaded the 2 files ISO 100 from the page and made a print test 12x18 inches and you really see a difference. I resized both to have 300dpi for printing on an Epson Stylus Pro 7500 with 1440dpi on the printer.
The difference in sharpness is clearly visible.
My wife due to the fact that she does not do any photography is always called to check the prints.
Try this 2 files. It is very interesting and eye opening.
QUOTE (Ray @ Oct. 02 2002,04:06)
Jim mentioned downsampling a 33MB 1Ds file to 9MB to put it on the same footing as the D30 and then upsampling both files to search for differences. I'm not sure if such an experiment is going to be at all useful. One would expect any differences, if any, to be very subtle. But if you did see any differences, with a magnifying glass, how would you know they were not differences in the lenses used?
Ray,
In the steps I described, the same image would be printed twice - once using full resolution of the 1Ds, and once after downsampling (to emulate a D30) then interpolated to make the second print the same size as the first print. The differences cannot result from the effects of different lenses because it starts with a single image!
Like you, I expect the differences to be subtle. Recently I printed two copies of a 12 x 18 print on my Epson 1270. One was printed by setting the Epson driver to 720 PPI, the other using 1440 PPI. Both prints were quite good. But placing them next to each other, and viewing from a few feet distance, the one printed at 1440 was somehow deeper and richer. Upon close examination (without a loupe), the differences were hard to pinpoint.
Perhaps printing an 8 x 10 from the D30 and from the 1Ds will produce a similar difference - we'll see!
Enjoy!
-- Jim
Michael
Thanks for the early and thorough info on the 1Ds.
In you first report,on the subject of moire,you mentioned that since
a large proportion of your work is wildlife and scenic in nature you
didn't expect moire to be an issue for you.
As a former 1d owner (I had one for 2 weeks before getting an
inside tip that the 1Ds was to be announced and fortunately deal
with an understanding supplier) I ran into moire quite a few times in
my short stint with the camera-specifically in birds feathers,some
animal fur,sand dunes,and the occasional brick wall.
Before you have to return the camera to Canon and if the opportunity
presents itself could you shoot some bird images.
I'd be curious to see if the higher pixel count will help alleviate this
problem.
Thanks
Mark Tomalty
Thanks for the early and thorough info on the 1Ds.
In you first report,on the subject of moire,you mentioned that since
a large proportion of your work is wildlife and scenic in nature you
didn't expect moire to be an issue for you.
As a former 1d owner (I had one for 2 weeks before getting an
inside tip that the 1Ds was to be announced and fortunately deal
with an understanding supplier) I ran into moire quite a few times in
my short stint with the camera-specifically in birds feathers,some
animal fur,sand dunes,and the occasional brick wall.
Before you have to return the camera to Canon and if the opportunity
presents itself could you shoot some bird images.
I'd be curious to see if the higher pixel count will help alleviate this
problem.
Thanks
Mark Tomalty
Thank Michael for his work by buying his Video Mag.
Michael, I reupped in anticipation of you getting Canon to get you the 1Ds in record time. After you sold me on the D-30 in a timely way, I figure I have saved $4,000 in film. So another extension of my subscription for $99 seems a more than fair exchange.
Ressing Up. I could get some 13x19 prints off some subjects with the D30. I am assuming that posters 20x30 are now in reach. But with what printer?
By the way, regarding the question someone had of going to 20meg chips seems useless for 35 size since this site has discussed that at 11 we are maxing out the resolving power of Canon lenses, true? From here on out the increase in dynamic range to negative film levels would be more useful.
Additionally, Michael's suggestion that we talk to the camera to make it change control is interesting, I'm already talking to my camera, but it isn't responding.
Michael, I reupped in anticipation of you getting Canon to get you the 1Ds in record time. After you sold me on the D-30 in a timely way, I figure I have saved $4,000 in film. So another extension of my subscription for $99 seems a more than fair exchange.
Ressing Up. I could get some 13x19 prints off some subjects with the D30. I am assuming that posters 20x30 are now in reach. But with what printer?
By the way, regarding the question someone had of going to 20meg chips seems useless for 35 size since this site has discussed that at 11 we are maxing out the resolving power of Canon lenses, true? From here on out the increase in dynamic range to negative film levels would be more useful.
Additionally, Michael's suggestion that we talk to the camera to make it change control is interesting, I'm already talking to my camera, but it isn't responding.
JacMac asks:
"By the way, regarding the question someone had of going to 20meg chips seems useless for 35 size since this site has discussed that at 11 we are maxing out the resolving power of Canon lenses, true?"
Reading the www.imaging-resource.com 4/3rds project report I came across the following:
"...According to Olympus, most current lenses for 35mm SLRs resolve down to about 10 microns, and top-end lenses to around 6 microns."
From dpreview's 1Ds report we can see that the 1Ds' pixel size will be 8.8 microns. Looks like the answer is "getting darned close".
Since the 1Ds fits 11.4 million pixels in a 35.8 x 23.8 mm imager and the 14n will fit 13.89 in the same area, that suggests the 14n's pixel pitch will be 7.2 microns. Working backward (and presuming my math is correct), 16.72 million pixels, each at 6 microns, would meet the ceiling imposed by lens resolving power.
This is not to suggest that even a 16.72 megapixel camera would provide the ultimate resolution for 35mm format. I think Foveon has taught us that. I want a full-frame, 17 megapixel X3 dSLR for XMas.
"By the way, regarding the question someone had of going to 20meg chips seems useless for 35 size since this site has discussed that at 11 we are maxing out the resolving power of Canon lenses, true?"
Reading the www.imaging-resource.com 4/3rds project report I came across the following:
"...According to Olympus, most current lenses for 35mm SLRs resolve down to about 10 microns, and top-end lenses to around 6 microns."
From dpreview's 1Ds report we can see that the 1Ds' pixel size will be 8.8 microns. Looks like the answer is "getting darned close".
Since the 1Ds fits 11.4 million pixels in a 35.8 x 23.8 mm imager and the 14n will fit 13.89 in the same area, that suggests the 14n's pixel pitch will be 7.2 microns. Working backward (and presuming my math is correct), 16.72 million pixels, each at 6 microns, would meet the ceiling imposed by lens resolving power.
This is not to suggest that even a 16.72 megapixel camera would provide the ultimate resolution for 35mm format. I think Foveon has taught us that. I want a full-frame, 17 megapixel X3 dSLR for XMas.
A couple of points which hopefully add a little on aliasing/tonal smoothness, and dynamic range. My biggest concern with the review and conclusions is how the #### I am ever going to afford a 1DS...
I agree with Michael's 4,000ppi 'diminishing returns' assessment as far as image detail is concerned, but it isn't enough to reach conclusions regarding tonal smoothness.
Film grain is irregular both in size, distribution and topology, and all of these attributes interact with the fixed geometry of pixel size and distribution. What many people believe is grain in their scanned images is often aliasing, which produces an exaggerated grain-like structure comprising false colour/false luminance pixels.
Avoiding this in low-pixel-count scanners is only possible by using a low-pass optical system (soft lens or antialiasing filter) which hurts image detail as well.
Grain aliasing can be so extreme as to render a scan unusable, especially with grainy, fast B&W silver-based films, but generally it is acceptable with scanners >=4,000ppi.
When it happens, the result is far more 'texture' than the film image exhibits when viewed or printed by analogue techniques.
Having seen comparative drum scans of an ISO100 tranny done at 4,000, 8,000, and 12,000 ppi, there isn't much gain in resolution of image detail above 4,000ppi, which accords with what Michael says in his review. By 8,000 ppi the scan has virtually all of it. Yet 12,000ppi shows a significant increase in grain detail over 8,000ppi, with better rendition of individual grain shapes. A lot of people say this doesn't matter, grain detail is not image information we want...
However, if we don't capture precise grain information, what we get through the conversion to pixels is aliasing, and a characteristically false and 'noisy' rendition of the film image. An Imacon scan at 3,200ppi necessarily adds an unknown extra dimension of aliasing noise.
How much, and how destructive depends on the anti-aliasing filter and CCD properties, even the lightsource, but it is important to recognise that any comparison with scanned film is not a comparison with film itself.
Pedantically, I think it would be worth doing a comparison involving a 12,000ppi drum scan, to get a more absolute measure of the differences between the film images and the EOS1DS. Not for the sake of resolution per sebut to judge relative tonal smoothness more accurately.
Having said all that, I am sure digital will still win, but the existing methodology probably makes film look rather worse than it is in this respect (and JPEG'ing of an image full of aliasing products compounds the issue, to film's disadvantage).
Michael's observation of moire and the artefacting of the red-shed boarding are more of the same. Aliasing is inherent with pixel-based systems, though the Foveon sensor will avoid colour aliasing and only alias luminance.
For me, the only obvious remaining advantage of film (apart from avoiding divorce and bankruptcy) is dynamic range.
My main interest and use of scanning is so I can use colour negative and B&W chromagenic films, exactly because they have a mad dynamic range : what you can see with adaptive eyes, you can photograph, pretty much.
I have tested and found 12+ stops with these materials. 10 stops of this is completely usable, there is severe compression in the rest, but it is still there. This is absolutely liberating compared to tranny, and gives me the control I only had in the wet darkroom with B&W, and more.
Alongside this, the D1S is, as reported, distinctly poor at 6 stops.
What's more, it is almost certainly an engineered inhibition. CCD sensors can now achieve 14 stops range, and do so completely linearly. I don't know any reason why CMOS should be worse, so it sounds like Canon have knobbled the range.
I hope this will prove to be something that applies only to JPEGS, and the RAW files will contain far more range.
There is a very good reason why Canon may have done so : wide range images look absolutely awful on-screen, impossibly flat, low contrast. You have to work with them, as raw material, and you wouldn't want to have to cope with the consequences of JPEG, it has to be clean, raw, high-bit data.
So the JPEG format is probably assumed to be for rapid onward transmission as a production file, with levels set, and only 8bit/ch available. To do anything else would frighten the desk editor who receives the file, and involve them in making judgements which are best left to photographers...
Fingers crossed, anyhow.
Regards
Tony Sleep (sometime filmscanner reviewer - suddenly retired
Tony Sleep Photography tonysleep@halftone.co.uk ::
QUOTE
Of course there will now be a chorus of those who say, "Ya, but a drum scan would have really shown a bigger difference in favour of film." Humm. Maybe. But here are my thoughts on this recurring topic. I have had drum scans made from my 35mm and medium format film on several occasions. Yes, an 8000 ppi scan is impressive, and can make bigger prints. But, I'm also convinced that while they give me more pixels, I don't get a whole lot more real data. There simply isn't that much more information on film than about 4,000 PPI. Above that we get bigger files, but not much more information. Maybe, 20% more than the 3200 PPI scans that my Imacon Flextight Photo scanner is capable of, but not 2 or 3 times as some inexperienced people presume from the numbers.
I agree with Michael's 4,000ppi 'diminishing returns' assessment as far as image detail is concerned, but it isn't enough to reach conclusions regarding tonal smoothness.
Film grain is irregular both in size, distribution and topology, and all of these attributes interact with the fixed geometry of pixel size and distribution. What many people believe is grain in their scanned images is often aliasing, which produces an exaggerated grain-like structure comprising false colour/false luminance pixels.
Avoiding this in low-pixel-count scanners is only possible by using a low-pass optical system (soft lens or antialiasing filter) which hurts image detail as well.
Grain aliasing can be so extreme as to render a scan unusable, especially with grainy, fast B&W silver-based films, but generally it is acceptable with scanners >=4,000ppi.
When it happens, the result is far more 'texture' than the film image exhibits when viewed or printed by analogue techniques.
Having seen comparative drum scans of an ISO100 tranny done at 4,000, 8,000, and 12,000 ppi, there isn't much gain in resolution of image detail above 4,000ppi, which accords with what Michael says in his review. By 8,000 ppi the scan has virtually all of it. Yet 12,000ppi shows a significant increase in grain detail over 8,000ppi, with better rendition of individual grain shapes. A lot of people say this doesn't matter, grain detail is not image information we want...
However, if we don't capture precise grain information, what we get through the conversion to pixels is aliasing, and a characteristically false and 'noisy' rendition of the film image. An Imacon scan at 3,200ppi necessarily adds an unknown extra dimension of aliasing noise.
How much, and how destructive depends on the anti-aliasing filter and CCD properties, even the lightsource, but it is important to recognise that any comparison with scanned film is not a comparison with film itself.
Pedantically, I think it would be worth doing a comparison involving a 12,000ppi drum scan, to get a more absolute measure of the differences between the film images and the EOS1DS. Not for the sake of resolution per sebut to judge relative tonal smoothness more accurately.
Having said all that, I am sure digital will still win, but the existing methodology probably makes film look rather worse than it is in this respect (and JPEG'ing of an image full of aliasing products compounds the issue, to film's disadvantage).
Michael's observation of moire and the artefacting of the red-shed boarding are more of the same. Aliasing is inherent with pixel-based systems, though the Foveon sensor will avoid colour aliasing and only alias luminance.
For me, the only obvious remaining advantage of film (apart from avoiding divorce and bankruptcy) is dynamic range.
My main interest and use of scanning is so I can use colour negative and B&W chromagenic films, exactly because they have a mad dynamic range : what you can see with adaptive eyes, you can photograph, pretty much.
I have tested and found 12+ stops with these materials. 10 stops of this is completely usable, there is severe compression in the rest, but it is still there. This is absolutely liberating compared to tranny, and gives me the control I only had in the wet darkroom with B&W, and more.
Alongside this, the D1S is, as reported, distinctly poor at 6 stops.
What's more, it is almost certainly an engineered inhibition. CCD sensors can now achieve 14 stops range, and do so completely linearly. I don't know any reason why CMOS should be worse, so it sounds like Canon have knobbled the range.
I hope this will prove to be something that applies only to JPEGS, and the RAW files will contain far more range.
There is a very good reason why Canon may have done so : wide range images look absolutely awful on-screen, impossibly flat, low contrast. You have to work with them, as raw material, and you wouldn't want to have to cope with the consequences of JPEG, it has to be clean, raw, high-bit data.
So the JPEG format is probably assumed to be for rapid onward transmission as a production file, with levels set, and only 8bit/ch available. To do anything else would frighten the desk editor who receives the file, and involve them in making judgements which are best left to photographers...
Fingers crossed, anyhow.
Regards
Tony Sleep (sometime filmscanner reviewer - suddenly retired
Tony Sleep Photography tonysleep@halftone.co.uk :
:
Perhaps the following is useful and easy to remember:
1. Nyquist theory says that you need two pixels to resolve a line pair. Yes, I actually verified this with a D30. Not 4 pixels.
2. Despite Foveon's claims, the Bayer array does NOT destroy the resolving power of a sensor all that much. In my tests with a D30, you lose 15% on the horizontal and vertical axes and almost nothing on the diagonal. Yes, I also verified this.
3. The "diffraction limit" is not a hard cutoff. It depends on definitions used. If you start with black & white bars and accept very low contrast in the output, lenses can deliver absurdly (and uselessly) high lp/mm. This is the primary source of those 100+ lp/mm claims that you used to see in decades past. Photodo has a lengthy expose on this on their MTF explanations. You need a definition that delivers a usefully high contrast level. Astronomers tend to use FWHM (Full Width at Half Maximum of the Airy disc) since this is what they use to match pixel sizes to telescope configurations, seeing conditions, and the Nyquist criterion (two pixels per line-pair). With this definition the theoretical diffraction limit of a lens in lp/mm = 1000/f-no for a wavelength of 4900 Angstroms (yellow light). Ie a theoretically perfect lens delivers 125 lp/mm at f-8. It would be more lousy in deep red light (< 90 lp/mm) and a lot better in deep blue (> 160 lp/mm). Ie real world lenses are unlikely to do better than, say, 80 lp/mm in white light at f-8 using the FWHM criterion. This also means that it is not particularly useful to use pixels that are smaller than around 6.5 microns if your lenses are optimised for f8 (like most 35mm lenses). Using all the above, a D30 is certainly good enough for lenses that deliver 45 lp/mm, a D60 around 60 lp/mm and a 1Ds between the two. I tested 3 Canon primes and NONE deliver, at a usefully high contrast level, anywhere near 45 lp/mm outside f5.6 to f11. Mass produced lenses are indeed that bad. Again, from my own testing with a D30, I would extrapolate that with a good Canon prime at f8 you can expect 2300 lines per picture height from a 1Ds (Lines per Picture Height is used in the digicam testcharts, just to confuse matters!). Depending on the printing resolution you desire (240dpi (9.6" high), 180dpi (12.8"), or whatever) you can determine what your own cutoff print size will be when compared to larger formats. Normal, unaided human vision considers 240dpi as pretty razor-sharp and should not be able to distinguish (resolution-wise) as to whether the source was a 1Ds sensor or an 8x12" negative. 180dpi satisfies many people too, especially when printer-related imperfections also come into play. Remember, a lot of people were and still are satisfied with D30 prints at 8x12". But this gets into subject matter/context arguments...
1. Nyquist theory says that you need two pixels to resolve a line pair. Yes, I actually verified this with a D30. Not 4 pixels.
2. Despite Foveon's claims, the Bayer array does NOT destroy the resolving power of a sensor all that much. In my tests with a D30, you lose 15% on the horizontal and vertical axes and almost nothing on the diagonal. Yes, I also verified this.
3. The "diffraction limit" is not a hard cutoff. It depends on definitions used. If you start with black & white bars and accept very low contrast in the output, lenses can deliver absurdly (and uselessly) high lp/mm. This is the primary source of those 100+ lp/mm claims that you used to see in decades past. Photodo has a lengthy expose on this on their MTF explanations. You need a definition that delivers a usefully high contrast level. Astronomers tend to use FWHM (Full Width at Half Maximum of the Airy disc) since this is what they use to match pixel sizes to telescope configurations, seeing conditions, and the Nyquist criterion (two pixels per line-pair). With this definition the theoretical diffraction limit of a lens in lp/mm = 1000/f-no for a wavelength of 4900 Angstroms (yellow light). Ie a theoretically perfect lens delivers 125 lp/mm at f-8. It would be more lousy in deep red light (< 90 lp/mm) and a lot better in deep blue (> 160 lp/mm). Ie real world lenses are unlikely to do better than, say, 80 lp/mm in white light at f-8 using the FWHM criterion. This also means that it is not particularly useful to use pixels that are smaller than around 6.5 microns if your lenses are optimised for f8 (like most 35mm lenses). Using all the above, a D30 is certainly good enough for lenses that deliver 45 lp/mm, a D60 around 60 lp/mm and a 1Ds between the two. I tested 3 Canon primes and NONE deliver, at a usefully high contrast level, anywhere near 45 lp/mm outside f5.6 to f11. Mass produced lenses are indeed that bad. Again, from my own testing with a D30, I would extrapolate that with a good Canon prime at f8 you can expect 2300 lines per picture height from a 1Ds (Lines per Picture Height is used in the digicam testcharts, just to confuse matters!). Depending on the printing resolution you desire (240dpi (9.6" high), 180dpi (12.8"), or whatever) you can determine what your own cutoff print size will be when compared to larger formats. Normal, unaided human vision considers 240dpi as pretty razor-sharp and should not be able to distinguish (resolution-wise) as to whether the source was a 1Ds sensor or an 8x12" negative. 180dpi satisfies many people too, especially when printer-related imperfections also come into play. Remember, a lot of people were and still are satisfied with D30 prints at 8x12". But this gets into subject matter/context arguments...
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.