Sorry about terse response. I now have a moment to discuss.
While these pictures show the extraordinarily rich colours that we very often see with Leica lenses and DMR, for me there are major problems with contrast, colour accuracy and tone.
For example, the blond lady's hair roots, the shadowy area by her right ear, and the shadow under her necklace on the left almost make this look like a print made with an empty black cartridge. The shadows seem to suffer from an unnatural luminosity scale which compresses contrast in the shadows in a way which is at odds with the more recognisable contrast in the mid-tones. It helps to make the image look more 2D. I don't know if it's down to processing, and of course I don't know how this image will look on output from any given printing device, but I do see it looking wrong on my monitor, and that's not a good start.
The same goes for the balanced portrait on this page. The light and shade just doesn't look right to me, let alone the khaki trees.
My experience of Leica lenses is that while superb in most respects, they often suffer from a non-linear response to luminosity, or whatever you want to call it, as shown in the shadow areas of these portraits. I think the DMR exacerbates this unreal contrast problem while adding coloration of its own. This makes my eyes suspicious and images seem less real. Dynamic range and rich colour are not ends in themselves. The luminosity response, colour, and skin tone in these pictures corroborate (to me!) my opinion that the DMR is a less good portrait camera than a Nikon D70 with the Nikkor 60 macro. I like my images to look real, and to me, these aren't excellent examples of that.
Thank you for your thoughts, Richard.
One should consider that these portraits are basically snapshots and out-of- the box unmanipulated at that, so they are obviously not showing their (or the DMR's) best.
To some extent I follow your reasoning, but when you say about the last image : The light and shade just doesn't look right to me,, I'm puzzled. To me this looks about right, skin tone as I remember it and the image has a certain "presence" that I often experience with the DMR.
It would certainly be educating if you would post some out-of-the-box, unmanipulated 5D or D70 shots to compare.
It's entirely possible that you and others might see properties with the colour response to Leica lenses and the DMR that I can't, after all I'm just a B&W photographer who dabbles in colour
When you mentioned the Nikon D70 and 60 mm macro (the Micro-Nikkor?) it occured to me that I have some shots of the same lady taken with the D2x and the 55 mm MF Micro-Nikkor. So, just for fun, here is a reconverted, unmanipulated (well, a slight crop), unsharpened Nikon image.
I should perhaps add that it is WB on the white wall in the background.
Edited by photoArne on Sep 10, 2007 at 06:08 PM GMT
Nice set of portraits Arne. I remember from Guy's thread that much of the discussion on colors that we saw in pictures posted here on FM that had been taken with the DMR revolved around the software used to process and convert the images, and if I recall correctly, there were a lot of issues with how these programs manipulated the DMR files.
>To some extent I follow your reasoning, but when you say about the last image : The light and shade just doesn't look right to me,, I'm puzzled. To me this looks about right, skin tone as I remember it and the image has a certain "presence" that I often experience with the DMR.
It is very good indeed, and I doubt your processing can be bettered. But I still see a swerve in the contrast scale in the shadows. If you look at the girl's hair at the bottom left of the image, or if you look through her hair at the shadows by her right cheek, retention of shadow information seems at odds with the punchier representation in her face. The shadow area in the crook of her neck and right shoulder almost looks like metamerism. What I'm driving at is there seems to be a slight distortion of the luminosity scale in order to retain shadow information. This makes the image look more like a painting than reality.
In other words, despite the better 'presence' in the DMR shot, I think I do prefer the Nikon shot for natural rendition, even though it could do with a little bit of work on contrast/colour balance. I think it is fundamental that a camera produces an undistorted representation of light and shade, and then that it produces accurate colour. After that comes saturation, sharpness, noise, handling et cetera. DMR pictures have frequently given me the uneasy feeling that this isn't what it really looked like. It is a subtle thing, and I know I may be alone in that view. Let's leave it there.
One thing is for sure - your pictures are always beautiful whatever camera you shoot with.
TeamSK jay wrote:
Seems to me that the color differences are more likely due to some other part of the data capture process than whether or not it is done with CCD or CMOS.
I've formed this guess from observation:
At base ISO, CCD per pixel quality is inherently superior (more accurate) than CMOS. But CMOS can have per pixel on chip amplifiers. So boosting ISO is much gentler on CMOS than CCD.
So the CCD quality starts high and drops like a rock. CMOS starts lower but has a gentler downward slope. CMOS can handle the extra 'charge' (or whatever it called) due to per pixel analog functionality, but only to certain limit (maybe ISO 1600 on Canon).
Software can make up a bit of the difference for CCD. Note the Phase P to P+ improvement with the same sensor. But in both CCD and CMOS, once the ISO is boosted by software only the results become more artificial.
DPReview or some similar site had a series of graphs about this. While the CMOS camera did show a much slower noise increase with iso, the CCD cameras didn't have much of an advantage at low iso.
brainiac wrote:
DPReview or some similar site had a series of graphs about this. While the CMOS camera did show a much slower noise increase with iso, the CCD cameras didn't have much of an advantage at low iso.
I don't believe it's a simple as noise. For example, if a photodiode had a consistent bias, would it register as noise?
With Nikon pushing further into software manipulation, quantitative testing will be increasing useless. Tests are simplified to get understandable answers. But those answers will not reflect what the camera does with real images.
