Trying to get color accuracy is often an exercise in squeezing the balloon: you get hold of one end, and the other gets away from you. I am more interested, however, in understanding how different cameras in my kit render color so I can know what I must do to color-match (as best I can) when I use two of them in the same shoot. To this end, I decided to see for myself about the claims that D7000 color rendition is off when compared to other cameras in the Nikon line-up. We should always expect some differences, but will they be large enough to matter?

The first set of photos were shot under different lighting conditions, using the same lens (85 f1.4D) and Picture Controls, using Auto White Balance.

1. D300, overcast window lighting	1. D7000, overcast window lighting
2. D300, mixed overcast window lighting and flash	2. D7000, mixed overcast window lighting and flash
3. D300, mixed overcast window lighting and flash	3. D7000, mixed overcast window lighting and flash

The results are fairly close. It looks like the Auto white balance (Auto-WB) algorithm is doing a fairly good job of accounting for whatever color-casts the imaging chain (including the Anti-alias filter, Bayer grid and sensor) may be introducing. Would we get closer if we set our own white balance setting? To see this, let’s take the 2nd and 3rd samples and apply the Flash WB setting.

4. D300, mixed overcast window and flash lighting	4. D7000, mixed overcast window and flash lighting
5. D300, flash lighting	5. D7000, flash lighting

Surprisingly (to me at least) these seem to show a greater difference between the D300 and D7000 than the Auto-WB samples. Perhaps, knows something we don’t. For the final set of samples, we use a reference in post-processing, either true white or gray, to generate our own white balance.

6. D300, mixed overcast window and flash lighting	6. D7000, mixed overcast window and flash lighting
7. D300, flash lighting	7. D7000, flash lighting

I still need to ponder these results a bit better, but to my eye, it seems like at least for these lighting conditions, I can trust Auto WB to get me pretty close to a match between my D300 and D7000. No doubt other lighting conditions (Incandescent, Fluorescent, and a mix thereof) might prove more difficult. I’ve also learned that my strategy to set a given temperature or default WB setting may not be best when I want to color match. As for the gray card results, at first glance they appear no better in terms of color matching than the Auto-WB results. Regarding which of these approaches yielded the most accurate color, I’d have to say, again, Auto-WB did surprisingly well, and the gray/white reference point method also yielded reasonably good results. I am more concerned with matching camera(s) output, though, than in producing accurate color, as I have learned first hand how getting “real” color is a slippery goal.

Lots to think about here, but the good news is that the colors aren’t wildly different, and that for most practical situations, color matching is fairly good.

Continuing where we left off, we now take a look at how to capture art pieces with accurate color. To illustrate the problem, here are two attempts to capture a particular piece, one using flash, and the other ambient light. The flash exposure was taken with the flash pointed up, to prevent reflection from the canvas, and the flash unit was “flagged” to prevent direct forward spill of light, again, to avoid reflection in the paint. The ambient light sample required a long exposure (1/6 of a second), so use of a tripod proved essential.

Using flash

Using ambient light

Both of these images exhibit a yellow, overly warm color cast. The issue arises from the white balance used to capture these images. Some cameras do a better job than others auto-detecting the correct white balance, but this often proves challenging when light sources, say daylight and overhead incandescent lighting, mix. While it is possible to adjust white balance on the spot, a method I prefer for this type of work involves a reference shot using a gray card (like this one), as shown here.

Using a gray card

The gray card becomes a reference that, when used in post-processing, yields white-balanced color. You’ll have to trust me, but these next two corrected images match the original much better than the captures we showed above.

Flash, color-corrected

Ambient, color-corrected

And now you can appreciate why the “final product” we showed yesterday looked so “right.”

With cropping and resizing

The question of whether to shoot in JPG or RAW, and what if any are the advantages of using the RAW format comes up often. The following is not a definitive proof of the benefits of one approach vs. the other, but a sampling of how “getting it right” in camera with JPG shooting works when compared to getting right or almost right in-camera with RAW, then following up with some post-processing tweaks. We will approach this by answering 3 basic questions: (1) How does an out-of-camera (OOC) JPG compare to a JPG resulting from straight-up, none-edited conversion from RAW in post processing; (2) How do minor exposure adjustments to a fairly well-exposed OOC JPG compare to a JPG resulting from the same adjustments on the RAW file; and (3) How do white balance (WB) adjustments on the OOC JPG compare to a JPG resulting from the same adjustments on the RAW file?

Equipment, setup and software used
Images were taken with a Nikon D80 and AF 35-70 f/2.8 lens, using the “Normal” picture mode and WB=Direct sunlight. A tripod was also used to ensure exact framing. Straight conversions to JPG and 16-bit TIF were performed in ViewNX without any adjustments. All exposure and WB adjustments were performed in Paintshop Pro X2, ensuring that the same adjustment was applied to both the OOC JPG and its corresponding TIF version. JPG files resulting from adjusted original files were saved using the same and high JPG quality settings in Paintshop Pro X2.

Comparison of OOC JPG against post-converted RAW-to-JPG version
To answer the first question, here are two sample image pairs. You decide what, if any, are the differences between the OOC JPG and the RAW-to-JPG converted (using ViewNX, in this case) versions. Please note that the OOC JPGs are going to be rotated as I did not want to make any modifications to the camera output (and saving JPG is a lossy process).

OOC JPG	ViewNX JPG

Comparison of exposure-compensated OOC JPG against post-converted RAW-to-TIF version
Now we start by toning down the highlights and adjusting midtones with the Highlight/Midtone/Shadow (HMS) tool, then adding a little contrast with Unsharp Mask (USM) as I prefer doing to some of my images to add pop; and I do this both with the OOC image and the RAW (after 16-bit TIF conversion in ViewNX).

Edited OOC JPG	Edited TIF

Since the exposures we’ve seen were a tad clipped in the highlights, I took a second set of sample pairs, this time with -0.7EV of compensation. Here are the OOC JPGs, prior any adjusments. As you can see, the highlights are in better shape, but the shadows and midtones are under-exposed and dull.

OOC JPG 1	OOC JPG 2

Comparison of WB and exposure-compensated OOC JPG against post-converted RAW-to-TIF version
And here they are after using the HSM tool to bring out the shadows and lighten the midtones a bit, and some USM contrast added in. I would point you in particular to the color/shade transitions in the blue tile for the first sample pair and to the color/shade transitions in the red petals of the second sample pair. See a difference, and if so, is it significant enough for you?

Edited OOC JPG	Edited TIF

Finally, we look at the third question and adjust WB to better resemble the colors in the scene. We do this with the images taken with -0.7EV for the first sample pair, so we also have to add the exposure adjustments discussed above. Again, look for color transitions, and if you are so inclined, go digging through the shadows in pixel-peep mode and see what you find.

OOC JPG	ViewNX TIF

This is a controversial subject and one where expressing any hard conclusions is bound to be unproductive in the face of what is ultimate a matter of preference for many digital photographers. I will say that given these very mild (low contrast) and unchallenging (regarding exposure) images, even the subtle differences I can see are enough for me to continue shooting and enjoying the benefits of the RAW format. I will try and come with additional samples, and if anything more significant than these results comes of it, I will follow-up with a future blog entry.

If like me, you don’t have that Ansel Adams knack for visualizing an image in gray tones, perhaps the following tips will help you figure out how your photographs can work as Black and White (B&W) images. First, if you want to receive immediate feedback as to what your image will look like, shoot monochrome. This may be a paradigm shift for you because the experts have told you to capture in color, then convert to B&W in post-processing. The myth, left over from the days before you had a competent DSLR, is that capturing directly into B&W is an irreversible process. Not so if you capture RAW images.

When you do this, avoid if you can the blah grayscale option. Instead apply color filtering to your B&W capture. Here’s what I do with either of my DSLRs (Nikon D80 and D90). I go into the Picture Mode (D80) or Picture Control (D90) menu and select the monochrome option and enable a color filter. Which color you use is up to you, but I find red, orange and yellow to give me the best results for most situations, and I use green occasionally. When you select one of these color filters, this is somewhat equivalent to attaching a physical color filter of the same color without having to worry about having the right diameter for your particular lens. Nifty, but keep in mind that it’s not perfect. Think of it as an approximation to give you a rough idea of how a B&W capture will work out.

Nikon D80 Monochrome filters

Nikon D90 Monochrome filters

Make sure you set the camera’s capture format to RAW, and later, if you repent from having chosen B&W, or if you want to try a different filtering technique, you can switch back to color. With Nikon RAW files (NEF), ViewNX will let you swap Picture Controls to revert back to color. Whether you do this or capture your original image in color, this brings us to our next option for generating B&W images with pop to them.

Bring your color image into your editor of choice, make any adjustments you want to make to color, white balance, sharpness, or whatever else you want to modify, and then bring up your application’s color filtering tool — again avoid the grayscale conversion. In Paintshop Pro X2 (PSPX2), this interface presents you with a color wheel, as shown below, that lets you set any color filtering you want. You can start to see the additional flexibility in this approach: you can have a plethora of shades of yellow or green or cyan (try finding that filter at your camera store) or anything in between, allowing you to tweak the output exactly as you want it.

To demonstrate how this works, let’s review a couple of images from yesterday’s discussion.

B&W, as shot in-camera	Color, reverted from RAW in ViewNX with White balance correction
B&W conversion with color filtering in PSPX2	Result from B&W conversion (including contrast and sharpening)
B&W, as shot in-camera	Color, reverted from RAW in ViewNX with White balance correction
B&W conversion with color filtering in PSPX2	Result from B&W conversion (including contrast and sharpening)

As you can see, applying color filtering for B&W conversion in post-processing gives the greatest flexibility and enables you to improve monochrome output quality. My recommended approach is to shoot RAW with a Monochrome filter in the field to preview whether a shot works as B&W on the spot, i.e., as an aid to visualization, then go home and tweak away in post-processing to optimize and maximize the quality of the final image.

One word of caution with this approach: it just so happens that the filters I like best, red and orange, tend to reveal more noise in blue skies. Green filters, on the other hand, produce the least noise. We could get technical here about Bayer sensors containing a majority of green pixels which when excluded with red or orange filtering leaves us with a noisy image. But instead I’ll just recommend that you experiment to see how this approach works for you.

The question of what makes a good Black and White (B&W) photograph has puzzled me more than any other photography-related issue. Frankly, for some time I operated under the principle that I never saw a B&W image that didn’t look better in color. Then came a trip to Paris this last spring. Following someone’s observation that “Paris was made for B&W photography,” I gave B&W a try, even setting my camera to capture RAW files with a monochrome color profile. I was amazed.

After reviewing my results and trying out B&W on images from before and after the Paris trip, I still don’t feel like I completely understand and appreciate B&W, but I have come to the following guidelines to decide whether I should even consider an image for a B&W treatment.

• First, if color is an intrinsic, required element of the image, for instance if the image is about an explosion of color, it is unlikely that it will work well in B&W. I say “unlikely” because there are always exceptions.
• Second, if the image is more about form or line, and/or if color is a distraction or a lesser component, consider B&W as a viable, sometimes preferable alternative.
• Third, B&W images work best when they have black and white in them, and preferably whites should not come from blown highlights and blacks should not come from under-exposed shadows. Look around and see whether you have dark and light colors and visualize whether these will turn into good blacks and whites.
• My fourth, more informal guideline is “don’t use B&W to salvage a photograph with a blown-out or otherwise uninteresting sky.” Maybe other people can use B&W to save a bad sky, but my experience is that a bad sky tends to stay a bad sky, no matter what I do.

To see these two guidelines at work, let’s use some images that put them into practice. This image was one where I actually tried to visualize what this would look like in B&W, but took the image in color. When I got home and tried various conversion methods, none yielded a good result. After scratching my head for a while I realized I was fighting a losing battle. Even though there are some cool forms in this shot and a curvy diagonal formed by the rocky hillside, color is an important element because it helps to separate the shades of green from the tans in the rock face. In the end I don’t think this shot is all that remarkable in color, so overall, maybe it’s just a plain miss.

Before I took this next shot in the Paris subway, I had a sense it would work in B&W. In fact, I took it with my Nikon D90 set to capture RAW with a Monochrome Picture control (which we’ll discuss further tomorrow). As far as I can tell, my hunch was correct. What makes this photo work in B&W better than color? Those strong vanishing lines really pop in B&W, and note that where the lines travel, the prevailing colors are gray tones, anyway. In addition, other colors are somewhat of a distraction and draw the eye away from our subject — that mysterious, beautiful lady walking away from us, and not those bright yellow signs to either side of her. Yes, it would be nice if we could keep her red coat in the image, but we’ll leave selective color for another day. We can also point out an emotional, less objective component: B&W better fits the mood of a busy, metropolitan subway.

I captured this final image in Monochrome mode as well, and later converted to color. Immediately, right in-camera, I knew I had something I found appealing. Now, looking back, I can understand why this image tends to work better in B&W. First, is color an important component of this image? Only to the extent that it lends some warmth to the scene, but the dim lighting and near monotone colors lend themselves to a B&W treatment. More so in this image, it is the emotional component, that noir feeling one gets in a dimly lit place like a bar or an intimate nighttime café that makes B&W work so well here.

As we wrap up this little discussion, I don’t know that I’ve helped you much in deciding for yourself whether an image will do better in B&W than in color. You might even disagree with some of my assessments above and, for instance, prefer all three of these examples in color, as I used to before I went on my trip to Paris. If so, I recommend you allow yourself to look at your own images afresh and see if you can find some that would benefit from a B&W approach. As you do remember that it’s your vision that matters, and that some images may serve equally strong and valid purposes in both B&W and color.

Come back tomorrow as we look at some techniques to capture and post process B&W images.

When it comes to getting the White Balance (WB) “right,” most suggest you apply adjustments in post-processing. I’ve done this and do it often enough to believe in it, but I also see a practical use for getting WB right on-site if you’re going for realism. Will you remember what the colors in a particular scene actually looked like days later when you post-process a set of images? For those cases where getting colors right at the time of shooting is crucial, the well-tested method of setting WB based on a white or gray card reading is the way to go. Similar solutions, such as White Balance lens caps are also available.

However, what if you’re shooting a band concert and the bouncer is not going to let you place a white card on the lead singer’s forehead?

Solution: If you have a D90 or like-camera with Liveview, switch it to Liveview and the WB to K, and adjust the WB temperature until the colors look like the scene before you.

Note that for this quickie method, you are assuming your camera’s LCD is color-accurate. In my experience, it’s close enough. In addition, keep in mind that Picture Controls affect color rendition. Either use the Neutral Picture Control (boring, contrastless), or use D2XMODE1… oh, wait, you can’t do that in the D90. But as luck would have it, someone’s put together an emulation of this mode for the D90 which you can download from a Custom Picture Controls page.