Didn't this discussion of colour start with the post of a sample 19mm image? How does the sharpness of the 19mm image compare to others 21-35mm or 19mm lenses?
I'll try to find some 19 vs 21-35 images later this week--been quite ill today....I should have some, although I deleted about 5 GB of images from most of my testing....My 21-35 is pretty sharp, even wide open, but I'm sure it's not as good as the 19 wide open.
>>> ...It is probably the signature of the DMR. Everything else in the picture is the right colour from the toy wagon to the row boat and play centre.
> ...I have looked at the sky and looked at the pictures and the colour is correct. Are you on a Mac by any chance? It may be a browser issue.
This really interests me. I have mentioned before that what I perceive as non-linear colour shifts in posted results have for me been the number one obstacle to buying Leica digital.
Although this might be a Mac/PC thing, or other setup inconsistency, I post below Rob's images followed by adjusted copies which look more natural TO ME. I would be interested to see if people find the skies too indigo.
Interestingly, I had to vary application of hue shift across the sky in the boat picture as the horizon went too indigo when the higher sky was indigo enough.
brainiac wrote:
...which of course would suggest that some compression of the sky colour range has occurred either in capture or post-processing.
Both of the pictures are in my Back yard. I live in Nova Scotia.
The change in colour from Horizon to sky is common when you can see a horizon that is actually on the horizon or sea level. At the horizon, there is more of the atmosphere for the light to go through and that shifts the colour, not the camera doing it.
Someday when it is sunny I will shoot a colour card in the sun with the sky in the background to see how accurate the sky colours really are. If there is no shift to cyan on the colour card then the cyan on the sky is the colour of the sky.
I am surprised how many people obsess about this sky colour thing now that we have digital imaging. I don't ever remember putting a slide up on the screen and somebody complaining about the colour of the sky. In the case of my current shots, excluding early conversions with Flexcolor, the sky colour is just that Kodachrome signature of the DMR that I like.
Edited by robsteve on Jan 08, 2007 at 10:27 AM GMT
On the boat image I prefer the original look.
Richard's adjustments in the sky affected also the greens and reds, which I like better in the original image.
If you use Hue/Sat adjustment on the cyans with hue +8 and saturation -40 the sky improves (for my taste) while the other colors remain almost intact.
Anyway Rob,
the original jpeg is quite good, although I recall the Zeiss 21 on 1Ds slightly better in the corners (full frame also).
But I know from experience, as I sold the Zeiss to get the Leica 19, that the latter improves at f/4 and by f/5.6 is absolutely on par with the Zeiss, with better colors and less distortion.
Overall, even if the Distagon was available in Leica-R, I'd prefer the Leica to the Zeiss.
> ...The change in colour from Horizon to sky is common when you can see a horizon that is actually on the horizon or sea level. At the horizon, there is more of the atmosphere for the light to go through and that shifts the colour, not the camera doing it.
I totally agree that the sky is more cyan nearer the horizon, even here in the UK on the annual sunny day ;-). My point was that when I came to adjust the boat shot, there wasn't enough variation between higher sky and near horizon sky. I applied a shift toward blue/purple, but found that when the higher sky looked accurate, the sky nearer the horizon was no longer cyan enough to look normal. This is consistent with the idea that a camera or processing shift has shifted blues towards cyan but not cyan towards green. That is why I say it looks compressive, i.e. lossy. Colour differentiation between blue and cyan is being lost, not just in skies, and in this case I had to put it in with a brush. That is the kind of colour shift which, unlike a simple cast or rotation of the whole colour wheel can cause problems if you want accurate colour and can be quite difficult to fix. One of the things I didn't like too much about Reala was that it sometimes made a cyan horizon look a bit blue. Sometimes you can fix it if you have an eye, you know what you are doing, and you have time.
Does it matter? To me yes. I prefer to start with natural colour, and bend it when I need to. As I've said many times, you could change your film, but your sensor is a fixture. Pondria often suggests that this sort of thing is just a calibration problem. He may be right, but as usual, I'm dull enough that I need to be shewn some things before I believe them.
Marco, it wasn't the cyan->blue shift that changed the greens. I actually adjusted the greens deliberately. I often find greens come up a bit too yellow for my taste on the Leicas. I think I have an idea now of the colour 'footprint' of the DMR, or at least many of the DMR shots that are posted. The sky is the easiest place to spot this as it is a colour we know pretty well. Vegetation colour varies a lot, so it's harder to be sure about the olive or yellow grass. Many DMR shots appear to me to have undergone the following transformations:
If you look at the lawn example you may notice that I removed blueness from the reds too. Of course I have no way of knowing what is accurate, but I have often seen slightly bluey reds from the DMR so I just guessed it's doing it here too.
Edited by brainiac on Jan 08, 2007 at 11:55 PM GMT
While I prefer less cyan in the sky, which however I didn't notice in my few DMR shots - maybe Canada has more cyan than Europe - I do love DMR color rendition.
DMR colors are rich but not oversaturated.
Much depends on the raw conversion, but the fingerprint is there and in my workflow I immediately noticed the nicer colors and contrast ("texture" I'd say) compared to my 1Ds and 400D.
Again though, only a matter of taste.
I have to say that as I recall, the colour of the sky in the maritimes is somewhat unusual. If anyone here has been to Skagen in Denmark, there the colours are also different than in other places. Painters specifically go there to capture that light. There is something about the air and the light which is different. In southern California there is a pastel look to everything. I think before people complain about the sky in Rob's images, they should visit Nova Scotia. Most parts of Canada where people actually live are also much further south than England. Toronto is almost the same latitude as Rome. The southern position might have something to do with it, although I am not sure how far south Nova Scotia is compared to Europe.
Edited by carstenw on Jan 09, 2007 at 01:12 AM GMT
Agreed Marco: to each his own. But colour _accuracy_ isn't a matter of taste, it's measurable by devices like colorimeters and spectrometers. If you lay a leaf next to a print of itself it really is possible to measure how unfaithfully its colours were reproduced.
brainiac wrote:
The 19mm looks great as far as I can see in these low rez images. Hubsand's test showed that it had slightly less distortion than the Zeiss too.
Lucky you Rob, it looks like a beautiful place to live.
If you look up at the original post, there is a link to download the full resolution jpeg.
Now that you have seen the full image from the 19mm, how does its f2.8 performance compare to your CZ21mm at f2.8?
Considering it's at f2.8 it's not bad.
I am curious about your method of processing though. Sometimes, to combat colour noise I switch to L-a-b and run a very blunt dust&scratches on the two colour channels but not the luminosity channel. This has the effect of only blurring colour and therefore preserving a lot of detail while reducing apparent noise. I then paint the effect in where I need it. I'm sure most people here know about this technique. If you overdo the colour blurring you can end up with colours bleeding into each other, i.e. colour information is presented at a much lower frequency than luminosity. I am seeing lots of this in your image:
Of course, we aren't looking at 24x36mm (full frame) corners and the 19 isn't being punished as hard as it would be on a higher megapixel body so it's hard to say much in comparison. All I can say is that in terms of resolution alone, I would be disappointed by this result if it had come from my 5D/Zeiss21.
But it's a nice result, and the DMR has other virtues.
Update: sorry - this needs correcting as I misremembered. The test I saw was L21-35 v Z21. Hubsand does not appear to have tested the L19.
[error removed]...his tests are always very revealing. If anybody here hasn't come across hubsands excellent lens test site, it's a must: http://www.16-9.net/
Edited by brainiac on Jan 09, 2007 at 02:30 PM GMT
- I do love DMR color rendition.