joekraft wrote:
Yeah, I'm not arguing anything either way. I just think it is an important clarification, because there are solutions now, that are very easy to implement. So I'm just wondering if these differences can be overcome, just like we calibrate monitors.
This would be the question. If you like the way colors are reproduced by the A900, can you calibrate the output of the 5Dll or the D700 in yout raw converter to match it?
edwardkaraa wrote:
[One recent test on lenstip shows that there is very little difference in sharpness between the AA filterless M9 and the D3X, which means the AA filter on the D3X is almost non existant.
I don't have a dog in this fight, but I'm unclear on the logic here. The D3X is 24.5 mp while the M9 is 18 mp. If the AA was almost non-existent, then the D3X should be sharper than the M9 at the pixel level. Even with the filter, the I wouldn't be surprised if the D3X probably out resolves the M9 (discounting sharpness at the pixel level)...
Even if a sensor without an AA filter resolves like a sensor with an AA filter and x1.33 the resolution, that means that an 18 AA-less sensor would resolve like a... 24 mp sensor with an AA filter. And the 24 mp sensor wouldn't have all the moire problems that the AA-less sensor has.
I didn't pull the factor of 1.33 out of thin air. I've seen it kicked around before, and maxmax uses the example of a 10 mp sensor with AA only resolving 7 mp.
I'm not saying that the D3X doesn't have a weak AA filter, that's quite possible.
Lotusm50 wrote:
This would be the question. If you like the way colors are reproduced by the A900, can you calibrate the output of the 5Dll or the D700 in yout raw converter to match it?
My guess in no. The color rendering is interpolated from the information reaching the sensor in the RGB channels. Weaker RGB filters means less separation between the channels and less accurate info available for interpolation.
One of the differences I noticed between my Canon DSLR and the A900 is the histogram. The RGB peaks look quite separated, and remind me of the histograms I used to get from my Coolscan film scanner.
kidtexas wrote:
I don't have a dog in this fight, but I'm unclear on the logic here. The D3X is 24.5 mp while the M9 is 18 mp. If the AA was almost non-existent, then the D3X should be sharper than the M9 at the pixel level. Even with the filter, the I wouldn't be surprised if the D3X probably out resolves the M9 (discounting sharpness at the pixel level)...
Even if a sensor without an AA filter resolves like a sensor with an AA filter and x1.33 the resolution, that means that an 18 AA-less sensor would resolve like a... 24 mp sensor with an AA filter. And the 24 mp sensor wouldn't have all the moire problems that the AA-less sensor has.
I didn't pull the factor of 1.33 out of thin air. I've seen it kicked around before, and maxmax uses the example of a 10 mp sensor with AA only resolving 7 mp.
I'm not saying that the D3X doesn't have a weak AA filter, that's quite possible. ...Show more →
I think results are clear. Anyway, an 18mp sensor will look sharper than a 24mp sensor of similar AA strength, even if the 24mp resolves more detail ( or not, depending on the lens resolving power).
edwardkaraa wrote:
My guess in no. The color rendering is interpolated from the information reaching the sensor in the RGB channels. Weaker RGB filters means less separation between the channels and less accurate info available for interpolation.
One of the differences I noticed between my Canon DSLR and the A900 is the histogram. The RGB peaks look quite separated, and remind me of the histograms I used to get from my Coolscan film scanner.
It might be worth trying just to see...
... and to resolve this.
edwardkaraa wrote:
My guess in no. The color rendering is interpolated from the information reaching the sensor in the RGB channels. Weaker RGB filters means less separation between the channels and less accurate info available for interpolation.
One of the differences I noticed between my Canon DSLR and the A900 is the histogram. The RGB peaks look quite separated, and remind me of the histograms I used to get from my Coolscan film scanner.
To add on the same subject, weak RGB filters will result in color contamination in pure colors for instance.
One of my main complaints about my previous Canon DSLR is for example while shooting a subject with a pure red shirt. The A900 gives me very low GB values in the red shirt, while the GB values in a Canon file are substantially higher. Eventhough I am able to manipulate the file to get a purer red, this will inevitably mess with the rest of the scene, skin tones, green foliage... etc. Hope my poor english is understandable
edwardkaraa wrote:
To add on the same subject, weak RGB filters will result in color contamination in pure colors for instance.
One of my main complaints about my previous Canon DSLR is for example while shooting a subject with a pure red shirt. The A900 gives me very low GB values in the red shirt, while the GB values in a Canon file are substantially higher. Eventhough I am able to manipulate the file to get a purer red, this will inevitably mess with the rest of the scene, skin tones, green foliage... etc. Hope my poor english is understandable
Color calibration in raw conversion maybe able to handle this.
Lotusm50 wrote:
Color calibration in raw conversion maybe able to handle this.
Partly. See the underlined text.
From Wikipedia:
"The color filters filter the light by wavelength range, such that the separate filtered intensities include information about the color of light. For example, the Bayer filter (shown to the right) gives information about the intensity of light in red, green, and blue (RGB) wavelength regions. The raw image data captured by the image sensor is then converted to a full-color image (with intensities of all three primary colors represented at each pixel) by a demosaicing algorithm which is tailored for each type of color filter. The spectral transmittance of the CFA elements along with the demosaicing algorithm jointly determine the color rendition."
edwardkaraa wrote:
Partly. See the underlined text.
From Wikipedia:
"The color filters filter the light by wavelength range, such that the separate filtered intensities include information about the color of light. For example, the Bayer filter (shown to the right) gives information about the intensity of light in red, green, and blue (RGB) wavelength regions. The raw image data captured by the image sensor is then converted to a full-color image (with intensities of all three primary colors represented at each pixel) by a demosaicing algorithm which is tailored for each type of color filter. The spectral transmittance of the CFA elements along with the demosaicing algorithm jointly determine the color rendition."
It will be interesting to see what the difference is in practice, and better understand what the relative contribution of these 2 aspects of the process is.
Edward, thanks. What was the wikipedia article you looked at, for my own further reading?
Just as an aside, alot of the candid shooting I do of my family is in a room that I'm not sure you could design to have worse lighting or color effects. I was really surprised to see how much a camera-specific color profile did for me, using one of the scripts out there, and later the dng profile editor.
I wish I had the means to do a side by side comparison between systems, but a few hundred dollars to rent a body and lens for a week, just to measure the color response, is a little too steep right now.
According to Borg, the A900 is closer to film response curves, rather eye response like most digital cameras. He states, "When the CFA is relatively weak and transparent in order to achieve low noise it results in colour reproduction non-linearity." Page fifteen shows a bit about the difference between eye and film response:
This apparently also becomes a problem for the 5Dii when using ETTR: "Here is what we have -a camera, spot metering from the brightest area of the scene as you suggested above, sensor reaches 7.5% saturation in green channel. (In jpg from the camera this is pushed to mid-grey of course.) Such design decision was taken part because of the low noise the camera has, part because the camera has problems reproducing colours in the higher range of raw values, above -1.5 EV from saturation."
m_appeal wrote:
From what I've seen Nikon 14-24 is sharp into the extreme corners even wide open. I have not seen samples with the D3X though....
Apparently, the Nikon 24-70 doesn't perform as well on the D3x as the Zeiss does on the A900, although I'm more speculating on that than anything... BUT I did read that 24-70 isn't perfect on the D3x.
By the way, here's an old but very interesting article that describes Zeiss design philosophy and may help to explain why a lens such as the Nikon 14-24 does not fit into their line.
Thank you very much for the link. It shows very clearly what I was trying to describe while talking about separate RGB histogram peaks.
douglasf13 wrote:
According to Borg, the A900 is closer to film response curves, rather eye response like most digital cameras. He states, "When the CFA is relatively weak and transparent in order to achieve low noise it results in colour reproduction non-linearity." Page fifteen shows a bit about the difference between eye and film response:
edwardkaraa wrote:
By the way, here's an old but very interesting article that describes Zeiss design philosophy and may help to explain why a lens such as the Nikon 14-24 does not fit into their line.
Reading Mr. Putz is always so, how should I put it, "enlightening" (maybe that's not the right word).
Here's an interesting quote from this (remember this is in the context of lens design): "The Contax RTS system however did not succeed in the market. It is of some historical interest to remark that the European camera makers are more successful with rangefinder designs, where the Japanese makers excel in single lens reflex cameras. The Contax G series was also based on these design principles. The G-series of lenses deserve high praise, but they again did not survive in the market."
Is one intended to draw the conclusion that the Contax RTS and the Contax G system did not survive becuase their lens designs (and/or philosophy) were flawed?? That implication is of course wrong. If that was not the intent of the statement, what could possibly the purpose of adding this particular snipe? I would point out that over their production lifetime, Contax RTS cameras and lens outsold Leica R, and the Contax G outsold Leica M. Still, it's irrelevant to the matter being discussed.
It's always a challenge to shift through Putz' obviously subjective and personal nonsense, to get to anything of value. And if you do get to it, you're never quite sure of its veracity.
I think he might have meant that as a rangefinder it wasn't so great? The lenses are meant to be fantastic, and he is usually spot on with his lens comments, so I doubt he was implying anything wrong with the lenses.
carstenw wrote:
I think he might have meant that as a rangefinder it wasn't so great? The lenses are meant to be fantastic, and he is usually spot on with his lens comments, so I doubt he was implying anything wrong with the lenses.
He was discussing lens design approaches, particularly with respect to rangefinder lenses. What does (essentially) Kyocera exiting the business of SLR cameras as well as rangefinder cameras, have to do with Zeiss' approach to lens design? Reading the essay, one draws the implicit conclusion that Zeiss' lens design approach is responsible for the failure of the RTS and G system and lenses to "survive" (which, by the way, we know not to be true). If that is not his intent, then he is sorely in need of an editor, because he doesn't know how to write to effectively convey his thoughts.
I mean specifically this sentence (and the one before and the one after), which seems to imply that the G was caught up in this situation:
"It is of some historical interest to remark that the European camera makers are more successful with rangefinder designs, where the Japanese makers excel in single lens reflex cameras."
Note the emphasis on cameras, not lenses. From what I have heard, I don't know what he could complain about w.r.t. the lenses.
In any case, he is definitely in sore need of an editor. He rambles on, gets off track, start discussing tangentially and increasingly less related things, and then makes weird conclusions out of the blue. I prefer his individual lens essays, short and to the point.
