Shootout: Is the conclusion clear to all?

[LEPING ZHA]

Alex: You have just pointed out the obvious missing bar truth in the D1s crop that I was jumping in for.  Personally it is not how "clean" the digital captures are, but that they all tend to show the "digital flavor" when enlarged beyond a certain limit.  The same is not true from the film scans.  I am a Ph.D. physicist and I tend to think the origin of the difference is on the diffusion layer in front of the sensors that makes the color capture work, and the square grid position of the pixels (the film grains are random).  Also to my eyes the film scans are far more natural, just like, wow, sorry to say, the music out of a good turntable.  Yes soon we will all get into this LP vs. CD and tube vs. IC bandwagan, but as many knows through the disputes moden high end audio has learned a lot about their basic fool-proof assumptions (such as the less distortion, the better the sound, etc.)  Noise free sometimes do bring in uncomfortable feelings to the human psycho.

Personally I can not afford the D1s, so at this point I am staying with my Pentax 67II (which has, to my opinion, the best lenses in any MF system especially for high contrast subjects, thanks to the SMC coating) and Ebony (4x5).  Out of my 6x7 chromes I routinely do Lightjet prints of 32x40 from drum scans, with details that have to be examed with magnifying glass.  Yes, even with glass they look totally natural, which to me is the key.  I've seen countless digital prints, including many the others were admiring, grossly oversharpen.  Bill Atkinson, my mentor, always speak against anything beyond 0.3 radius and 350 amount USM.  Thoses numbers out from the Nik Sharpening tools are just way way too large.  I have many musician friends and most of them could not stand CD sound.  It is too far to say that digital audio had produced a generation with no taste and appreciation to the true natural beauties of music.  But certainly I hope there will be no degeneration in taste of arts.   ::

[Dave Millier]

Leping

I agree with you about the widespread tendency to over sharpen - it can lead to a "sharp" but somewhat "mechanical" look to images. I've been experimenting with lower levels of sharpness in a bid to discover whether it's possible to learn to prefer the subtle smooth look to the brilliant sharp look. Not really sure yet though!

I'n not sure I agree with the hi fi analogy though. I have a specialist but budget priced Cd player of some vintage and although I wouldn't say it blew my Pink Triangle Turntable away, the turntable really doesn't demostrate anything over the CD player (that cost 1/4 the cost).Unless that is, you consider hiss, clicks and pops and a rather over rich bass as an advantage...

[John Beale]

The grain comparison shot confused me at first, but I think I understand it now.

My understanding (assumption) of what's going on with the MF scans vs. 1Ds is based on my guess about the MTF curve of the whole system. Resolution is not just one number, it is a curve (modulation transfer function) showing how image contrast goes down as the detail becomes smaller, until finally (limiting resolution) the contrast goes to zero.

My assumption based on these tests is that the MTF curve from the 1Ds is higher- greater contrast- for those details which are visible in prints up to 11x17". Now, if you print bigger, you see that the contrast on smaller details drops to 0 for the 1Ds file, while there is some detail left in the film scan. But since those details are smaller than the inkjet printer resolution can reproduce in a 11x17 print, it is of absolutely no advantage to you for final images of that size or smaller. So digital is the winner in that arena as Michael stated.

You might argue, why not use some sharpening process to raise up the MTF curve on the film scan? Of course you can, but that brings up the grain and you would have sharper but objectionally noisy prints; the digital images clearly have less noise.

[Alex Tsow]

I'd be interested to see the full jpegs of the Pentax (Imacon scanned) and the Canon 1ds.  With the crops comparing the noise of the ressed up Canon to the full sample Pentax, I see more detail in the Pentax: there seem to be extra window bars on the right hand building for the Pentax (detail that eluded the Canon).  Also (for that snippet) the noise is easily reduced to 1ds levels by a smart blur in photoshop.  I'd have to see the whole image to create sensible global settings though.  Smart blurring is a critical noise reduction technique that I'd bet is rolled into RAW to TIFF conversions.  With what little this test shows, I have to conclude that MF is still pulling more detail.


Alex Tsow

[Doug_Dolde]

Leping,

Just wanted to say that I've seen many of your Pentax 67 images on Photonet.  You are certainly a master with that camera.  Your images are spectacular and very impressive !

[Roger_Cavanagh]

With the crops comparing the noise of the ressed up Canon to the full sample Pentax, I see more detail in the Pentax: there seem to be extra window bars on the right hand building for the Pentax (detail that eluded the Canon).  

It doesn't look to me like these close-ups were of the same section of window. I downloaded both images and measured the size of windows: the left-hand windows in the Canon image are 83 pixels high, in the Pentax they are almost the same at 81. The right-hand windows in the Canon picture are each 60 pixels high, whereas the Pentax window with bars is 86 pixels high.

Web versions are images are not the best for quality comparisons (as Michael intimates), but looking at the Royal York hotel, the detailing on the ironwork on the roof and balcony is much more clearly defined in the 1Ds image than the Pentax Imacon scan. The drum scan seems about the same the the 1Ds, but at an extra 300 bucks, it wouldn't take many scans to pay for the 1Ds.

Regards,

[Ray]

I've just been re-reading Miles Hecker's article on 'Digital Camera Image Quality' where he produces a 'normalised IQ table' which includes a bit of PS enhancement for the Fuji Provia 100F images (use of smart blur). When such images have had grain reduced in this way, the table shows the following ratings:-

35mm Provia 100F (PS enhanced) - -1
D60 - - - - - - - - - - - - -  - - - - -0.78
1Ds - - - - - - - - - - - - - -  - - - -1.03
645 Provia 100F (PS enhanced) - - -1.75

The above results seem significantly out of step with Michael's 'shootout' results. I get the impression from Miles Hecker's article, that scanned film lends itself to certain improvements in PS which would show only a very marginal improvement when applied to the digicam image. What's going on here? Is it now recognised that Hecker's results are far too conservative and significantly underrate the potential of the 1Ds?

Another issue that concerns me; can I be assured that edge sharpness of the two lenses used in the tests is the same as centre sharpness, at F8? The blown up sections Michael's selected are from the extreme edge of the frame.