I’ve written a short article on mushy misconceptions.
Also, there is a link for a Canon 5DMKII RAW file and an exported DPP JPEG file .
Feel free to download the images and make judgements for yourself. Heck, process and post your own results and conclusions!
I selected an image I felt was typical of what a landscape shooter, like myself, would photograph. Yes it it's not perfect, but IMHO typical of good quality.
I applied sharpening that is appropriate for printing on my Epson 9880, so the jpeg file may be a hair sharp for screens.
I find Canon's colors, like Nikons, to exemplary and quite "non-mushy".
I use to find Nikons reds a bit too yellow as well as the greens. But, I found adjusting my workflow and RAW processor, and of recent, the camera calibration file in ACR, the results to be A-ok. The culprit was me. :0

Here is the picture I'm referencing (just a visual reference):
http://mattanderson.zenfolio.com/p15207892/h3d71784d#h3d71784d
Here is the Link the Article and RAW/JPEG file .ZIP download:
http://blog.widen.com/blog/industrial-retouch-and-color

I look forward to seeing/hearing your own opinions (and I'm not a fan boy, I shoot Nikon, Canon, Hasselblad, and as of recent Sinar)

Kind Regards,
Matt Anderson
http://www.commercialfineart.com/

Download a trial of DxO 5 and you can render your NEFs as 'every camera with raw available' and than you can see the differences.

Never heard of mushy colors. What are they?

I'm not sure I really understand - so you're saying that both Nikon and Canon have good colors with the appropriate post-processing?

Mushy or not, your website has some great landscape photos

Many forums I frequent have certain "individuals" stating a certain brand has mushy looking colors. Basically the chroma definition isn't well defined, is what I think their stating. I thought I'd give them a venue to, um, prove it. Or, at least, dig in, and make a fair assessment. So I gave those interested a RAW file to have a go with...

The misconception is often really about what "camera colour" really is... :-) And it's nothing like what you get out of any raw-converter. And most cameras can get "good, roughly accurate" colours out of most scenes with the right camera profile installed.

But there's quite a lot of truth about "Canon having worse colour", and this is not some fanboy propaganda, this is a digital colour signal processing truth. Canon, and Olympus (and in a few examples Pentax too) have got very expanded colour filters on the CFA on the sensor. This is to increase light throughput, and thereby lower picture noise. Unforunately it also lowers colour hue placing precision quite a lot. Both the Canons and the Oly's have quite a lot of points on the hue-curve where they cant fulfill the Luther-Ives condition (two or more colours in front of the camera can under several different lighting conditions result in the same values IN the camera, the raw-file).

The leader in colour precision right now is Sony, with the A850/900 models. Unfortunately for Sony, their lower light-efficiency (that in part stems from having better quality, steeper colour filters!) has given them a bad reputation regarding the high ISO performance, without giving them credit for having the best colour precision and presentation this side of medium format backs.

The problem with the 5D2 is mostly in the greens>reds, where the red colour filter "swings back" in efficiency when it should just fade away - and this into a green that is all to wide to be able to discriminate between the sensitive middle greens.

Consider the graph below. There is NO way that a raw-converter, or ANY kind of NASA/NSA black technology, can ever know if values in the marked area (mid green to yellowish) are either bright mid green or dark yellow green. It's all guesswork. Same goes for quite a lot of the yellow-orange hues, where the 5D2 is to "flat" in the filtering to be able to properly discriminate between yellow-orange and red-orange. This type of filtering is common to all Canons in production today, and give the famous "Canon" skintones.

Interesting! Do you have the corresponding curves for the Sonys you mention and any of the newer Nikons?

What is the source of the information - your own measurements?

theSuede wrote:
The misconception is often really about what "camera colour" really is... :-) And it's nothing like what you get out of any raw-converter. And most cameras can get "good, roughly accurate" colours out of most scenes with the right camera profile installed.

But there's quite a lot of truth about "Canon having worse colour", and this is not some fanboy propaganda, this is a digital colour signal processing truth. Canon, and Olympus (and in a few examples Pentax too) have got very expanded colour filters on the CFA on the sensor. This is to increase light throughput, and thereby lower picture noise. Unforunately it also lowers colour hue placing precision quite a lot. Both the Canons and the Oly's have quite a lot of points on the hue-curve where they cant fulfill the Luther-Ives condition (two or more colours in front of the camera can under several different lighting conditions result in the same values IN the camera, the raw-file).

The leader in colour precision right now is Sony, with the A850/900 models. Unfortunately for Sony, their lower light-efficiency (that in part stems from having better quality, steeper colour filters!) has given them a bad reputation regarding the high ISO performance, without giving them credit for having the best colour precision and presentation this side of medium format backs.

The problem with the 5D2 is mostly in the greens>reds, where the red colour filter "swings back" in efficiency when it should just fade away - and this into a green that is all to wide to be able to discriminate between the sensitive middle greens.

Consider the graph below. There is NO way that a raw-converter, or ANY kind of NASA/NSA black technology, can ever know if values in the marked area (mid green to yellowish) are either bright mid green or dark yellow green. It's all guesswork. Same goes for quite a lot of the yellow-orange hues, where the 5D2 is to "flat" in the filtering to be able to properly discriminate between yellow-orange and red-orange. This type of filtering is common to all Canons in production today, and give the famous "Canon" skintones.

I don’t think one can accurately make a blanket statement about a brand.
I do agree though, it seems every where I have read, Canon has made gone route with a CFA that is “less dense”. But, I contend that for print and web, there is more than enough information to create a robust file with decent color accuracy and chroma detail. In a lab, I’m sure we could set up a LUT array and compare statistically what percentage X is better than Y, but, IMHO, I believe both X and Y to be at a level that is sufficiently high for current media trends. (with savvy processing) I also agree, neither X and Y is at a phase one, hasselblad, or sinar level. But their getting closer.

I also would be intrigued to see side by sides from Canon, Nikon, and Sony. Lets see it.

by the way, this isn’t emil (ejmartin) is it ?

You can find side to side color accuracy comparison at image resource. You'll find NONE of the manufacturers produce 100% accurate color.

Not entirely sure how to read their test results charts but the % error seems to say the D300 is a bit worse for color then the 40D or the 50D or E-3. So what does that say? Absolutely nothing! The differences between any of them are so small.

I think the OP's right there isn't much difference. And when he said mushy colors I had an image of "color" singing stupid love songs and crying

gotak wrote:
... And when he said mushy colors I had an image of "color" singing stupid love songs and crying

Very nicely put! Now I understand.

I'm new here having just gone over (or should I say back from predigital) to the 'dark' side. I was told I would be less happy with Nikon's colors - quite frankly, once processed I can't tell the difference. My focus tracking success is higher though

It is quite hard to tell the difference between brands, if processed correctly - and if you don't have identical shots to compare... But it is also quite easy to differentiate the brands with such a simple test as trying to capture a few "colour-blindness test charts". The ones supplied by Ishihara are quite easy (for a camera!) as they are meant to test for almost total congenital colour vision deficiencies, more accurate vision anomaly charts as those used when testing people that are supposed to work with colour (as in pre-press, printing, ink&pigment suppliers and so on) for colour hue resolution capacity can easily differentiate different cameras.

Considering the amount colour-blind people actually working as photographers, that "colour from capture" isn't as important as CONSISTENT colour through your workflow shouldn't come as a surprise. The finished product (print/screen) will never accurately mirror reality anyway - so why bother

Well, there are some VERY good reasons.
-Accurate "capture" profiles will make your PP workflow much easier.
-Having hues placed at the correct distances from eachoter may be just as important as the individual hues being of the exactly "right" XYZ-corresponding value.

Argument no1 above is a "blanket statement" that you may believe or not. Almost ALL photographers that I know (quite a few!) say that they have lessened their time spent in PP after correcting their capture directly with the correct camera profiles.
Argument no2 is actually quite self-explanatory if you just give it a few seconds of thought. Exactly how many shots have you spent a more than few minutes on, just because "SOMETHING" isn't right? There's a very good chance that that "something" is nothing more than a slight chroma imbalance, like in the Canons that the yellows stay "yellow" until a fair bit into the region where they should have turned orange - to then turn bright red immediately. Same with the blue-purple transition.

There are in fact cameras that would have problems in a test a simple as this:
http://www.xrite.com/custom_page.aspx?PageID=77 - given the corresponding prints done with perfectly balanced daylight lit pigment inks
-that I can consistently score full marks on.

But that any modern camera today (except maybe the "compact" class cameras) can - given the right working premises, i.e. a correct input profile - give you colour accuray, or "quality", good enough for almost anything outside controlled comparisons or lab tests is likely true. Different brands DO however take different amounts of correction to get there.

theSuede wrote:
The misconception is often really about what "camera colour" really is... :-) And it's nothing like what you get out of any raw-converter. And most cameras can get "good, roughly accurate" colours out of most scenes with the right camera profile installed.

But there's quite a lot of truth about "Canon having worse colour", and this is not some fanboy propaganda, this is a digital colour signal processing truth. Canon, and Olympus (and in a few examples Pentax too) have got very expanded colour filters on the CFA on the sensor. This is to increase light throughput, and thereby lower picture noise. Unforunately it also lowers colour hue placing precision quite a lot. Both the Canons and the Oly's have quite a lot of points on the hue-curve where they cant fulfill the Luther-Ives condition (two or more colours in front of the camera can under several different lighting conditions result in the same values IN the camera, the raw-file).

The leader in colour precision right now is Sony, with the A850/900 models. Unfortunately for Sony, their lower light-efficiency (that in part stems from having better quality, steeper colour filters!) has given them a bad reputation regarding the high ISO performance, without giving them credit for having the best colour precision and presentation this side of medium format backs.

The problem with the 5D2 is mostly in the greens>reds, where the red colour filter "swings back" in efficiency when it should just fade away - and this into a green that is all to wide to be able to discriminate between the sensitive middle greens.

Consider the graph below. There is NO way that a raw-converter, or ANY kind of NASA/NSA black technology, can ever know if values in the marked area (mid green to yellowish) are either bright mid green or dark yellow green. It's all guesswork. Same goes for quite a lot of the yellow-orange hues, where the 5D2 is to "flat" in the filtering to be able to properly discriminate between yellow-orange and red-orange. This type of filtering is common to all Canons in production today, and give the famous "Canon" skintones.