Color Accuracy of Latest DSLRs?

[e_b7]

How's the color accuracy of the latest DSLRs versus, say, a Canon G3,

and also, versus film?

 

I find with the G3 that colors are somewhat off at times, and

fiddling with Photoshop can make colors look different, but not more

accurate. If I am concerned about accurate colors (usually), I shoot

film.

 

Any comments would be much appreciated. Thanks in advance.

[mike_elek]

People have gotten used to the look of saturated color. There's always been debate over the color accuracy of film, and most magazines did "shootouts" with various photo labs. Some people really prefer the saturated look of Velvia, while others like a more neutral film like Ektachrome or Portra.

 

It's been said in the past that Canon digital cameras recorded reds as "hot." I can't recall what others say about other digital cameras -- if any of them favor one color.

[Ed_Ingold]

I really like the flesh tones from my D2h. The auto white balance is effective over much larger range than previous Nikon cameras. Is it accurate? That depends on the ambient light and the subject. You are kidding yourself if you think film is "accurate." Everything distorts, you just pick a distortion that pleases you (or your customer).

 

I can create a very accurate profile in using a Color Checker chart and the InCamera plugin for Photoshop. This is not always enough. For one thing, not every subject looks good corrected to daylight. For another, the D2h is rather sensitive to infrared light, which imparts a green or purple cast to images, particularly to facial shadows and dark cloth. An hot mirror filter is helpful under low, incandescent light which is rich in infrared.

[doris_chan]

"If I am concerned about accurate colors (usually), I shoot film."

 

Care to share with us the secret of which film gives "accurate colors"?

[e_b7]

Kodachrome 25

[andrew robertson]

K25 renders purple as a blue. It's definitely NOT accurate.

[peter_phan]

<i>If I am concerned about accurate colors (usually), I shoot film. </i>

<p>

The way to truly ensure accurate colors, under any lighting condition and any light source, is by shooting digital and shooting a color calibration target like a <a href="http://www.gretagmacbeth.com/index/products/products_color-standards/products_colorchecker-charts/products_colorchecker.htm">Gretag Macbeth Color Checker</a>, or even just taking a shot of a <a href="http://www.argraph.com/cat-sect/New%20Products/QP%20Card.htm">QP Card</a>. It isn't enough to just "fiddle with the colors in Photoshop". You need a reference source on which to base your adjustments. The makers of QP Card also make software the works in conjunction with their color calibration targets to quickly and effectly ensure accurate colors <a href="http://www.qpcard.com/?p=easy">here</a>. So if you're really interested in shooting accurate colors, you should be shooting digital. You just need the right color reference tools.

[robin_sibson1]

Most colour films have a problem with flower colours that we see as intensely blue, such as blue gentians. I believe this is because these flowers also reflect very strongly in a part of the spectrum just outside the visual range (?near IR) and this effectively 'fogs' the film so that the flower appears pinkish. Maybe there is a filter that helps, don't know, but I do know that on flash photographs the problem is much reduced, presumably because the output of the flash does not contain much energy in the problem part of the spectrum, whereas daylight does.

 

This is not a problem on my DSLR (20D) nor, I assume, on other digital sensors. What I do notice is that intensely orange-red flowers are a problem; the red channel saturates at an exposure level well below what is required for the picture as a whole, and the colour looks too blueish. Early saturation of the red channel can arise just as badly with yellows as seen from the three-colour histogram, but the visual effect is more nearly acceptable. Intensely blue flowers can produce early saturation of the blue channel, but it is easier to control (note that this is a different problem from the 'pink gentian' on film). These problems would be more controllable if the sensor had greater dynamic range.

 

Otherwise, I would say that the opportunity to manage each colour channel separately, and the powerful tools available to do so, mean that more accurate colour matching can be achieved digitally that by conventional colur film processes. That's not to say it is easy.

[e_b7]

Thanks for all your responses. It sounds like I need more image processing tools. I shot a purple image recently with digital, and the purple was fine, but yellows were a bit green, reds a bit pink, and blue was a little aqua. Skin tones were good.

[conrad_hoffman]

Color reproduction using three channels is inherently flawed. Film or digital, there are colors you just can't get right. If you want to try a depressing experiment, set up a glass prism in sunlight and try to photograph the resulting spectrum. Compare the result to what you see by eye. You'll quickly understand the exposure limitations, and the problems that occur in the deep reds and violets. My wife knits using some very red, orange, purple, and violet yarns, very similar to the spectrum excercise. Getting accurate photos of the results is nearly impossible, and I do get slightly better results with film. The digital result seems to break the colors down into discrete values, a less subtle effect. OTOH, I don't have a high end dSLR. BTW, you think photographers are fussy, try artists!

[scott_eaton]

<I>These problems would be more controllable if the sensor had greater dynamic range.</i><P>True, although most of the problems can be solved by shooting in RAW/Adobe RGB mode and/or taking carefull white balances, something that typically can't be done with the toy digicam many of you are using as a reference here to glorify film. Anybody claiming they are getting more accurate color with film than a dSLR has spent too much time squatting over their precious slides on a light table and has cut off oxygen to their brain.<P>I do a good deal of flower macro work and insist on accurate color reproduction, and can only recall a handfull of plants that cause problems. In general I discovered when I moved primarily to digital that there was no contest when it came to color accuracy and it was much easier to nail particular specimen on digital than film, although it's easier to 'amplify' subtle colors under low contrast conditions with Provia/Velvia than dSLR. I do admit that light violet/purple flora are tough for any media, digital or film. <P>The problem is that digital sensors (on decent cameras) are for all intents and purposes linear within their intensity range, but our perception and preference of color isn't linear. It's funny, but those films renowed for color accuracy like Kodak Professional E-6 films do so at the expense of trashing more saturated colors, while portrait print films like NPH do so at the expense of compressing the entire spectrum. Kodachrome and Velvia are like random number generators.<I><P>If you want to try a depressing experiment, set up a glass prism in sunlight and try to photograph the resulting spectrum</i><P>

 

 

Build a capture device that can record near infra-red all the way to deep Violet/UV, and then take that same sensor out in the field. You'll be sadly dissapointed at it's bland rendition of visible, primary colors.

[robin_sibson1]

Scott, I think we are agreeing, but since you have quoted me, here is chapter and verse; these are just a few examples. Problem reds: Delonix regia (Flamboyant Tree, that I have many times photographed in the tropics), Crocosmia masoniorum (blazing away in my garden at the moment, alongside about 1500 other species that I grow), several of the central american species of salvia, of which S. miniata and S. fulgens are two good examples that I have in flower just now. Problem yellows: the British native Tulipa sylvestris. Problem blues: Gentiana verna (and many of the other bright blue species), Salvia patens. These are as far as I can tell only a problem on the 20D because they burn out a channel at the point where the rest of the picture is still a bit under-exposed; they do not seem to be outside the gamut. I saw a recent posting where someone was having the same problem with very bright red sportswear. We are all familiar with burnt-out white highlights, but the ease with which it can happen in a single channel does take a bit of getting used to.

[bill_tuthill]

Take a look at Dpreview.com testing. Most DSLRs reproduce Macbeth

chart colors very accurately, given proper whitepoint setting.

 

The best things I can say about color accuracy of the Canon G3 is

that greens are fairly good, and some Olympus digicams are worse.

Print films like Reala and NPH can be scanned or printed to almost

perfectly match the Macbeth chart, something not possible with any

Kodachrome. Off-chart indigo/violet hues sometimes suffer in

oddball lighting, as Scott implied.

[doris_chan]

"Kodachrome 25"

 

That's a very amusing answer EB. Other than the fact that Kodachrome of any flavor is far

from "accurate" colorwise, where are you buying your Kodachrome 25 and when was it

reintroduced?

[e_b7]

What's the best (or a good) post-processing method? Is it including a color chart in images and adjusting in PhotoShop, or using the QP software, or something else? Thanks all.

[andrew robertson]

My 10D and R-D1 are both far more accurate than any film I have ever used. As far as

processing, shoot a grey card in a scene with the same lighting as your subject. Some folks

complain about DSLRs inability to provide correct color balance in mixed lighting, but they

overlook the fact that this is impossible. The best thing to do in such a situation is to place

the white balance so the most important tones are accurate, and let the others fall where

they may.

 

Practice it. A modern DSLR provides more accurate color than a G3, by the way.

[e_b7]

I've had trouble with fill-flash (mixed lighting) and white balance. The Canon G3 provides some flexibility with white balance, but not much (advantage to the DSLR on that). I as able to correct the problem to some extent with Photoshop and a hue adjustment.

 

Thanks again!

[lex_jenkins]

To follow up on Scott's observations, I don't see what dynamic range has to do with skin tone, to which point this thread has digressed a bit. Virtually all skin tones of which I'm aware fall within a narrow dynamic range that virtually any camera or film can capture.

 

Now, getting the subtleties and nuances the way we like 'em is another matter. I'd rather have skin tones that appeal to my eye instead of tones that are "accurate."

 

I like the skin tones produced by my Olympus P&S digicam because they're muted, never reddish, etc. The overall color saturation can sometimes be dull but it's great for people pix, especially with direct flash. Suits my needs for this particular camera.

 

Same with my D2H. Moderate saturation. It's usually easy to get pleasing skin tones right out of the camera just by tweaking the white balance and tone compensation settings.

 

Are those skin tones "accurate?" I dunno. I like 'em. And they're well within the camera's fairly limited "dynamic range" (which is largely irrelevant here).

 

Overall color accuracy is another matter. Every dSLR maker struggles to maintain accurate color across the spectrum. So do film manufacturers. Neither has a lock on any claims.

[stephane_albi]

If you really want very good color accuracy, shoot using wider color space (adobe rgb). Set

your white balance and shoot a color target before making your shot. To process your

pictures, profile your camera using the shot of the target (using a profiling software) and

then assign your home-made profile to your next shots. I don't have any idea of how the

color accuracy compares to film but I would say that I am really satisfied with it and I am

really not very to satisfy usually :p

 

Very pleased with my monaco color calibrating system. They have a special product for

photographers to profile screen and digital camera.

[shane_hu]

The images (other than the raw format) produced by the camera has already been processed trying to map to the color response similar to that the human vision expects. The sensor most likely has different spectral response than the human eye's. The human vision system also adapts to the environment in a way much more complex than the so called auto white balance. For accurate color reproduction, it is a very difficult task under "un-controlled" lighting conditions.

 

After the mapping, the damage has already been done. No amounts of Photoshop processing can bring it back. Unfortunately, most camera do not reproduce color "accuarately". They want to make the garbage dump look beautiful so that people will buy it.