Yes, well, it's difficult to stay away when you know that not clarifying something now will spawn ten times the discussion next time. As in this case.
Ok, I'll make one more attempt at explaining and demonstrating and I'll do it step by step. I'll start by addressing the issues you raised in your previous post.
Yes, you are right L in Lab is "Lightness" not "Luminosity" - my bad. However the point stands - neither Lightness nor Luminosity is relevant for analyzing colors. RGB space in general is thouroughly useless for color analysis as color information is mixed in with lightness information. A & B channels in Lab space is better - especially in this case when we are not interested in absolute color levels but in color separation.
Now, as for color clipping - the images posted were resized and sharpened in RGB mode. It's not unusual for some clipping to occur but it's nothing that changes the image dramatically. Nevertheless, since you think it may make a difference, I'll work with the full sized image from now on.
There are some differences in contrast and a slight difference in exposure, but that's to be expected. The 50 MP is a more contrasty lens and f/5.6 is just a measure of the aperture, not the actual light transmission. So when using the same exposure settings on the camera there are bound to be differences in exposure when you use two different lenses. This isn't of any particular interest in this case as we're discussing color separation.
As you can see, absolutely no clipping. We can see some differences, like for instance that the R channel is wider on the Zeiss and the relative positions of the means of the bell curves are shifted. That's simply a difference in color rendering, but it still says little about color separation.
Now before I go on, I'll say that at least to me the difference between the two is quite obvious by simple ocular inspection of the crop. However, I'll do the full analyis here in case you don't agree.
As the crop contains predominantly red hues, let's take a look at the a channel as it's responsible for the reds. As Lab space is huge (much larger than RGB) looking at the channel in grayscale would produce a very low contrast image. So here I adjusted the black and white points to the maximum/minimum in the image so that we get a more readable picture:
This shows the color separation thing quite clearly. There are regions that are almost completely white (=very large a) in the left crop while variations are visible in the right crop. This means that the second crop will have a much better separation of different hues of red. In the left one it will be just one shade of red in a region while in the right you'll se graduated differences.
If that is not convincing enough, here's a full analysis. I'll be using a Discrete Wavelet Transform (DWT) which is essentially a series of high-pass filters - one at each level that removes high frequency information (i.e. detail) and saves it. What remains after the filtering and wavelet reconstruction (called the "approximation level") is the image with the high frequency details removed.
We do this to get rid of the texture which makes the task of decoding color differences much more difficult. Should we use a simple blur, we'd destroy the colors themselves as we'd blend them together. Using a DWT allows us to just look at the color shades sans textures while not sabotaging the colors.
I must say that I was very surprised by the results myself. I had for one not expected it to be so dramatic plus I thought the Canon would have a similar but narrower curve compared to the Zeiss. Not sure why it blends them together into specific clusters, but it's very likely a combination of camera sensor and optics that is the cause.
Anyway, regardless of the cause it unambiguously shows the difference in color separation. The Zeiss differentiates between similar shades of red while the Canon tends to just blend them together into one.
This comparison was between two 50 mm lenses, and you also found similar color differences between the copy you had of the Canon 100 2.8 L IS macro and your ZE 100 MP.
thedigitalbean has made attemts to reproduce differences in color Here but it is very difficult to spot any differences at all in those images. Do you have any suggestion to what the reason could be?
There are certainly some differences in his shots, and in the comparison shots I've done but noting of the order you see in the 50 comparison. The reason is very simple - the 100L is a very good lens, the 50/1.4 is not
Colors can be a bit tricky though as they very much depend on the type and intensity of light. For instance the Leica 35/1.4 Summilux ASPH and Leica 28/2 Summicron ASPH can produce entirely different colors in some situation with the Lux producing an almost desaturated image while the Cron creates colors that seem radioactive. In other situations however the difference is barely noticeable.
Although I do have a habit of doing lens tests, I also know the exercise to be futile in most cases. Lenses have complex non linear behavior and what you get out of them will depend on a large number of parameters. The only real way of knowing a lens is to take hundreds of photos with it. After a while you see in what kind of situations it does well and when there is a better alternative. MTF charts can give you some ideas but they are no guarantee. I have for instance been constantly annoyed with the Leica 75/2 APO-Summicron ASPH. In theory that lens should be near perfect. It should blow the 100 MP out of the water in every respect, yet I've never really liked it and I'm seldom excited by anything it produces. The 100 MP on the other hand, still gives me a kick to use and I love the results - although it is objectively inferior in terms of optical performance.
I should add also that looking at colors in sRGB space is a sad story. For full appreciation of the colors a camera/lens combo can capture one should use ProPhoto RGB on a wide gamut monitor. As 99.9% of the world uses sRGB it is unfortunately impractical to do so, but it's a shame because much more is lost by using a too small color space than is by choosing one lens instead of another.
denoir wrote: I thought the Canon would have a similar but narrower curve compared to the Zeiss. Not sure why it blends them together into specific clusters …If we increase the number of bins we can observe the same phenomenon with the Zeiss albeit with much finer distinctions between the cluster ... it unambiguously shows the difference in color separation. The Zeiss differentiates between similar shades of red while the Canon tends to just blend them together into one
I've made one entry in this thread, on page 8. I mentioned that I use two low priced C/Y Zeiss lenses. I never claimed that Zeiss has stunningly better resolution, or 3D ability.
But what I noticed from day one, was its ability to shoot the exact same scene as my Canon lenses, such as a green leaf, or a concrete wall with many rust stains. The Canon picture would show the leaf in one shade of green, or the rust stains in one shade of red-brown. The Zeiss would show ten different shades of green or red-brown. That ability is the single quality why I choose to use Zeiss over Canon.
Denoir has shown quantitatively, what I spotted that first day.
Don Clary wrote:
But what I noticed from day one, was its ability to shoot the exact same scene as my Canon lenses, such as a green leaf, or a concrete wall with many rust stains. The Canon picture would show the leaf in one shade of green, or the rust stains in one shade of red-brown. The Zeiss would show ten different shades of green or red-brown.
Don,
Do you still have access to those images? Again, those are practically the same shots taken with the same camera under the same lighting conditions and processed using the exact same PP workflow, correct? Can you post at least two of them side by side for us, please? Thanks!
charlesk wrote:
It is routine for threads to drift over time. There's only so much that can be said about the actual lens test that started the thread. People talk about that, and they say their peace, and then they go to related topics and the thread goes as it will. That's how it works.
This thread was never about the Zeiss 21mm at any rate, so how is that interest of any relevance? Regardless, if that's what you want to talk about, why not do so? It would certainly be more constructive than complaining about off-topic bickering by posting more off-topic bickering yourself.
Or you could start a new thread, you know. Nobody is forcing you to read this one if it doesn't suit your interests. But you not liking the direction it has gone is not a valid justification for trying to get it shut down for everyone else....Show more →
I started the thread so I know what is was about. You are right that it was not about the Zeiss 21 lens. But it was not about the Canon 100 IS either
Hi Lars,
I remembered back then somewhere around 2004-2006, you couldn't stand it when folks praised Zeiss lenses (of course, Contax lenses made by Zeiss back then). That was before the alternative forum was created; actually, that's the main reason that that board was created. Just like now, there were two opposing sides; one that loved the Zeiss rendition and the other that couldn't tell the difference (I am generalizing right now). The situation didn't change much if at all from the situation back then, huh ?
Anyway, I am curious as what made you change your mind? After you have bought Zeiss lenses and tried them out yourself? I didn't mean anything negative with this post but I was curious about what triggered your liking Zeiss lenses now.
AGeoJO wrote:
Hi Lars,
I remembered back then somewhere around 2004-2006, you couldn't stand it when folks praised Zeiss lenses (of course, Contax lenses made by Zeiss back then). That was before the alternative forum was created; actually, that's the main reason that that board was created. Just like now, there were two opposing sides; one that loved the Zeiss rendition and the other that couldn't tell the difference (I am generalizing right now). The situation didn't change much if at all from the situation back then, huh ?
Anyway, I am curious as what made you change your mind? After you have bought Zeiss lenses and tried them out yourself? I didn't mean anything negative with this post but I was curious about what triggered your liking Zeiss lenses now.
Yes because this is the Canon mount forum. And then they never posted about lenses with Canon mount.
This post is a ZE lens with Canon mount. just like a Sigma or Tamron with Canon mount. I have never done any post here with Zeiss contax lenses.
And I have not changed my mind. The reason I use Zeiss lenses today is that I can buy them with the Canon mount. I'm not interested in using adapters/stop down metering/ no exif-electronics working and so on......
But I'm interested in using the Zeiss ZE lenses. Because most of them have really excellent image quality
Lars Johnsson wrote:
Yes because this is the Canon mount forum. And then they never posted about lenses with Canon mount.
This post is a ZE lens with Canon mount. just like a Sigma or Tamron with Canon mount. I have never done any post here with Zeiss contax lenses.
And I have not changed my mind. The reason I use Zeiss lenses today is that I can buy them with the Canon mount. I'm not interested in using adapters/stop down metering/ no exif-electronics working and so on......
Back then the Alternative Board didn't exist and those folks used Canon bodies. So, you liked Zeiss lenses even back then and you expressed your displeasure because they just posted them on the Canon forum, correct?
Yes the problem that many years back was with a few people here that have left FM today. Sometimes they had more posts on the first page with alt-lenses than here where posts with Canon mount lenses.
And yes I liked the lenses (but I don't bought them) after a while there where just to many posts that not belonged in this forum. And a lot of posts with both Leica lenses and Leica cameras. Maybe more than the Zeiss
Don Clary wrote:
AGeoJO,
I deleted them, but since I still own both the Canon 28 and the Zeiss 28, I can shoot them again.
Hi Don,
I thought you had them readily available but no problem and there is no need to re-shoot them. I do have a few Zeiss lenses as well.
Don Clary wrote:
The Canon picture would show the leaf in one shade of green, or the rust stains in one shade of red-brown. The Zeiss would show ten different shades of green or red-brown.
This was based strictly on memory and not that you revisited and compared them again then. I didn't mean anything negative, but I was just curious.
Denoir, that analysis you posted on the color was very interesting.
I have compared (for my own purposes) the color differences between the ZE 21 vs. 24 TS-E, the ZE 35 f/2 vs. the 35L, the 50 MP vs. 50 f/1.2 in addition to the 100 MP vs. 100L. The difference in color IMO was least pronounced when comparing the 100s is probably the most interesting observation I can make at this point.
Denoir's tests are curious and if I can find the time in the coming weeks I'll see if I do something similar. My current thinking is to try and plot a chromaticity diagram for each of the lenses and compare that. The trick is in finding a test chart with all the colors. I'll see if one of the Macbeth charts I have access to will do.
No questions.
After reading 27 pages, I just want to say it was worth it (finally).
Thanks for the informative comparison. I really appreciate your time and effort.
Thanks, excellent comparison!
As for the presentation of the color chart with overlay, The Canon patches all look a little brighter. I downloaded it and adjusted all Canon patches with Brightness -5 in CS5, and the visual difference all but disappeared. You might want to add an overlay with brightness alignment.
On a seperate note, managed to find similarly framed cathedral shots with an 85L and 70-200L IS f/4. Must be a favourite test subject . The 85L looked very similar in rendering to the ZE. The zoom looked a lot flatter.
Note the cross, on the zoom shot it looks on the same plane as the wall behind it, on the ZE it stands out a bit. I had 4 identical shots with the zoom, the same sharpening was applied. This is far from being a precise test, take it FWIW
?