That's a bummer Carsten. Isn't there a way of adjusting the camera yourself? My experience is that few repair departments take the time to calibrate by trial and error, and that's the best way. I had similar problems with a Mamiya 6 and a Contax RTSIII many years ago. The Mamiya came back with no adjustment made, and Kyocera UK just refused to accept that the RTS was off, when it was worse than anything I had seen. I drove to Reading 3 times but they never got it right. If you can, do it yourself. Best service department you'll find.
> ...Richard, look at the detail and contrast in the wood.
The thing you are talking about is called 'acutance'. You can add as much or as little of it as you like when you convert your raw file, or you can set it in 'picture styles' on the 5D. You can also add it using a Photoshop filter called 'Unsharp mask'. Acutance is not the same thing as resolution. Here are crops from the 5D image, the M8 image, and the 5D image with unsharp masking. I hope this illustrates that acutance is not necessarily anything to do with a camera's actual resolution:
It is worth noting that the M8 will have more DoF in these test images, as it uses a wider lens for the same angle of view. It is also worth noting that it is focussed slightly behind where the Canon is focussed. Finally, it is worth noting that the 5D shot is taken with a Canon zoom, and one would expect such a lens to provide much less acutance and colour contrast than a German prime.
brainiac wrote:
...
Pondria, do you still think, when you look at these images, that the camera has no effect on colour? I think this is a great illustration of how subtle but important our judgement of colour is, and how cameras do have clearly distinguished colour footprints.
.
Sorry, I missed your post. I was busy with "real work"
Yes, I do, although my position is not as extreme as you put. I say, "Effect of Camera is far less than RAW converter setting".
Sorry to Tom, the test has little meaning in terms of comparing color characteristic of cameras. With ACR, just play with calibration settings. Every combination makes your camera a different camera.
It is kind of like this. With old CRT TV set, you could change R,G,B, Hue, Sat settings. Those knobs change the colors dramatically. You walk into a large Appliance store and look at TVs on display and say, "Wow, SONY does red colors better than Samsung". You are looking at either the factory setting or last change by someone
Richard, overcome the illusion ! There is no such thing as "Better Red". There is either correct Red or incorrect Red. You calibrate your RAW converter to get correct Red, correct Blue and correct Yellow. Once you do that. Every camera will produce exactly the same colors except boundary colors around the gamut limit.
Pondria - we've been over this before. It is my view that some cameras appear to compress and/or shift certain hues at certain luminosities. In those colours tone and contrast can be irretrievably lost at the capture stage. The data is gone and no amount of reinterpreting the numbers can guess back what was there. I think the effect is visible in the surface of the M8's red pepper. It looks kind of two dimensional, almost posterised, compared to the other cameras. Important tonal information is being lost there. I accept that you don't agree that this problem can't be fixed with calibration.
However, what interests me more is what we actually see in images posted by people using the cameras. If it is very difficult to get the colours right then lots of work will need to be done. We routinely see unconvincing colours in posts from DMR and M8. I don't make the inference that the people posting the images are failing to apply some as yet unseen process; I make the inference that it probably takes too much hard work to produce satisfyingly accurate colour with those cameras.
By "better red" I would mean more accurate. Colour can be inaccurate by degree.
BTW - I would like to thank Tom again for posting a genuinely comparative test. It's incredibly helpful in assessing these cameras, which is difficult to do when some in the press have been untrustworthy. It can be a thankless task too. Thanks Tom for sharing information which would have taken me hours to generate myself. Your test shows me a lot of things that weren't visible in my own tests, and gives me much more insight into the image quality differences between these camera systems. This is exactly the kind of information for which I originally looked into this thread.
I think the colors are ok, though I wonder why the 5D and DMR dont show the same (pleasing) shadows on the yellow paprika like the m8. On the red fruit the shadows are similar on all samples
> ...the 5D and DMR dont show the same (pleasing) shadows on the yellow paprika like the m8.
I will not be surprised if most prefer the M8's rendition of the yellow pepper, but that is an area where I feel the M8 is poor. It seems to me to be the result of a non-linear tonal range. The effect to my eye is similar to the difference between T-max films and the more traditional BW films. The response to the luminosity scale seems steppy, i.e. there is a mild posterisation in the tonal range which is no longer smooth and continuous. As a result of this, seductive though the high contrast of the M8's rendition is, to my eyes, all the other cameras draw something that looks more like a real yellow pepper. Like I said before, the M8 yellow pepper seems to have mysteriously lost its translucency. It looks like a yellow eggplant (aubergine).
I have just noticed that there is almost no luminosity variation in the 5D's yellow pepper at all. A 100% desaturation makes it look really flat and bad. I don't know if different processing for black and white could cure that.
Richard, I have also observed this kind of flat response to reds with the 5D, although not all reds. This was originally one of my reasons for wanting a Leica, although I cannot say that the M8 has delivered on that promise yet. Let's see what happens with an IR filter.
Here's something else weird: the M8 file seems to have had some sort of criss-cross texture applied. Maybe Tom can explain it. Perhaps it is something to do with post-processing:
The first one is what you would get from ACR by using the Default setting. The second one is after the calibration. What's surpring to me is that the 2nd one not only retains the fine mid tones but also looks simply more detailed resolution-wise. One can make a lot of assessments on color characteristics of 1DsII from the first shot.
By the way, I also know that 5D needs similar Red correction in ACR calibration.
OK - point taken. Conversion/calibration probably makes a more noticeable difference to perceived colour than the inherent differences between these cameras. It would be interesting to see if calibration by your methods makes these cameras indistinguishable on colour. That would surprise me, but I'm ready. I too would like to know more about your calibration method. Nice picture BTW.
The top one looks to have a lot higher contrst setting than the bottom. Are you sure they were processed with the same settings (other than your calibration) ? If the top was was 'as good as it gets' without calibration, I would be VERY suprised!
brainiac wrote:
The thing you are talking about is called 'acutance'. You can add as much or as little of it as you like when you convert your raw file, or you can set it in 'picture styles' on the 5D. You can also add it using a Photoshop filter called 'Unsharp mask'. Acutance is not the same thing as resolution. Here are crops from the 5D image, the M8 image, and the 5D image with unsharp masking. I hope this illustrates that acutance is not necessarily anything to do with a camera's actual resolution:
Acutance is easy to adjust for a uniform surface, as seen here, but not the same for a detailed picture with a variety detail, textures, and contrast. Adjusting a picture like this is about as easy as it gets for the unsharp mask.
Richard and Dave,
The gist of this problem ( with the default settings ) is that they apply some contrast and curve to make the picture punchy and visually pleasing.
And I use the method based on Eric Chang's[click...] but mine is more streamlined and simple. I know you can use script. But I do with Excel files. It took me less than 20min to calibrate. I have offered multiple times that I can calibrate for anyone just by sending me the MacBeth photo.
The calibration is not perfect. You can get only close.
Richard,
After the calibration, different cameras will produce almost exact colors for less saturated colors. The colors with high saturation ( Near the Gamut boundaries ) will be different.
> ...Acutance is easy to adjust for a uniform surface, as seen here, but not the same for a detailed picture with a variety detail, textures, and contrast. Adjusting a picture like this is about as easy as it gets for the unsharp mask.
That's not my experience with the 5D. The AA filter removes acutance evenly all over the frame. That's the 'smear' or softness that people mention. A 150/1/0 unsharp mask, when required, doesn't seem to cause any problems, even with camera JPEGs. It compensates for the AA filter evenly all over the frame. There isn't much moiré to woiré about.
That's why I don't understand complaints about the 5D's AA filter. It doesn't seem to hurt much whether you use raw or jpeg. What it does seem to do is avoid moiré. There may be a slight loss of high level detail, but it still seems to outresolve the M8 in Tom's test and mine.
Edited by brainiac on Jan 05, 2007 at 07:24 PM GMT
i'm throwing in the towel on comparisons. there have been many valuable points made but way too much
"my mind is made up don't confuse me with facts"
I see a critical issue with the aa filters. people avoid the AA filter issue like the plague, or confuse/dilute it with Nyquist frequency baloney suggesting it is only evident at x lpm. the aa filter is only effective at high frequencies so it doesn't really matter?
duh what is the spatial frequency of the separation of an absolute black adjoining and absolute white.
also George pointed out my test of the 5D and Kodak was invalid and the clients made decisioins based on visual cues from me?
George you weren't there so how could you determine the testing procedure.
the test?
basically a pile of pics and the clients seperated them into 2 piles. they did this while i was working and i did not evaluate thier choices until afterwards so i did not impact the test.
they simply chose the kodaks by a staggering percentage except for portraits where there was virtually no statistical difference.
anyway it is interesting to read this stuff.
Got my M8 recently and this will be the first weekend i'll have some play time so looking forward to it.
i have consistently said that if i were still a working pro i'd likely be shooting a 5D. it is a better overall solution than my kodak and leicaly my leica :-)
Richard, I did calibrate mine and if you haven't you should, i think you'll be even happier with the camera, it sorted some nasty red issues for me making the files much easier to work with.
bill vann wrote:
Tom, thanks for the comparison test i enjoyed it.
i'm throwing in the towel on comparisons. there have been many valuable points made but way too much
"my mind is made up don't confuse me with facts"
I see a critical issue with the aa filters. people avoid the AA filter issue like the plague, or confuse/dilute it with Nyquist frequency baloney suggesting it is only evident at x lpm. the aa filter is only effective at high frequencies so it doesn't really matter?
duh what is the spatial frequency of the separation of an absolute black adjoining and absolute white.
also George pointed out my test of the 5D and Kodak was invalid and the clients made decisioins based on visual cues from me?
George you weren't there so how could you determine the testing procedure.
the test?
basically a pile of pics and the clients seperated them into 2 piles. they did this while i was working and i did not evaluate thier choices until afterwards so i did not impact the test.
they simply chose the kodaks by a staggering percentage except for portraits where there was virtually no statistical difference.
anyway it is interesting to read this stuff.
Got my M8 recently and this will be the first weekend i'll have some play time so looking forward to it.
