As those of you who know a thing or two about what goes on ‘round here know, last week’s dekePod was devoted to the topic of faking an HDR portrait. As usual, the technique flew by in the blink of a bug’s eye. A really scary, creepy bug’s eye. Which is the idea, of course. Few know this, but dekePod won this year’s Nobel Prize in Subliminal Anti-Training. (They give out that particular award in a tiny, dimly lit room off the janitor’s closet, so it’s not widely covered.)
But anti-training doesn’t always work. In fact, one study suggests it kills roughly 1 out every of 16 lab rats. (Thankfully, we haven’t heard of any human fatalities—yet.) Which is why I offer this traditional step-by-step description, as it applies to my youngest son, Sammy, who quite obviously really enjoys his ice cream.
Happily, while I consider this a relatively advanced technique, it’s not an elaborate one. It involves just three effects—Shadows/Highlights, Curves, and High Pass—applied to a smart object in the Lab mode. In fact, the actual technique, from beginning to end, takes up just 2 minutes of a 6 and a half-minute video. The rest is setup, context, and goodbye. (I probably shouldn’t be telling you this, but very little of any given dekePod is cluttered with conventional instruction. That would rather defeat the purpose. Let the lure be long and the katch be kwick.) Meaning that once you understand how the technique works, you can race through it in a matter of minutes.
Step one: Open an image. Choose Image > Mode > Lab Color. That converts the RGB image to Lab, where a mere 8 bits per channel goes an awfully long way.
Step two: Normally, it’s nice to keep things nondestructive. For this technique, it’s absolutely essential, In the video, I go to the Layers palette flyout menu and choose Convert to Smart Object (below). But you could just as easily choose Filter > Convert for Smart Filters. Either delivers a smart object, which protects the image from all earthly harm.
Step three: HDR is largely about bringing the invisible into the visible range. Which means calming down the highlights and waking up the shadows. In other words, tugging the extreme colors into the mainstream of the midtones. To do this, choose Image > Adjustments and then choose the only command available (below), Shadow/Highlight. In Lab, Shadow/Highlight can’t harm the colors (as it does in RGB) because it affects just the luminance information.
Step four: My dark areas of the image are hideously filled in, so I raise the Shadows value radically, to 90 percent. The light areas need less work, so I scoot the Highlights value up less dramatically, to 30 percent, as shown below.
Step five: The result is actually remarkably ugly. To gain better control over the process, I turn on the Show More Options check box (above) to reveal several technical looking options that make an enormous difference in the command’s performance. Here are the values I set, and their results:
In the Shadows area, I lowered the Tonal Width to 30, which limits the shadows to the darkest 3 percent of the colors, and left the Radius at 30 pixels, which keeps the halos from growing so large that they fill in the eyes.
In the Highlights area, I lowered the Tonal Width value to 40 percent, so the lightest 40 percent of the colors are affected, and raised the Radius to 90 pixels, which diffuses the effect and all but eliminates halos in the bright areas.
In the Adjustments area, I took the ineffective Color Correction value down to and raised the Midtone Contrast value to +35, which prevents the image from flattening out.
Step six: After making the changes you deem appropriate (because you’re working with a smart object, you can always change your mind later), click the OK button. In my case, the new image contains newly wakened details, including much brighter irises in my son’s eyes.
Step seven: This image has sufficient luminance problems that the only viable solution is Curves. So I add a Curves adjustment layer. In Lab, you can edit just one channel at a time, and the Curves dialog box starts off set to the Lightness channel. I lock down the quartertones (just north of the shadows) at Input and Output values of 31 each. Then I set a point at Input: 11, Output: 6 to plunge the shadows and another at Input: 75, Output: 92 to brighten the highlights (see below).
Step eight: To raise the saturation values, I set symmetrical points at Input: -128, Output -67 and Input: 127, Output: 67 in both the a and b channels. The image needs more pink, so I nudge the light point in the a channel to Input: 127, Output: 64. Then I infuse the image with yellow by adding a point to the middle of the b curve at Input: -13, Output: -3 (see below).
Step nine: When you are satisfied by your adjustments, click OK. If you feel like you might have gone too far (in my case, Sammy verges on jaundiced), then back off the Opacity value in the Layers palette. For my part, I took the Opacity value down to 60 percent.
Step ten: Now to sharpen the image. The best way to sharpen a low-frequency portrait shot is with the High Pass filter. So click on the smart object layer to make it active. Choose Filter > Other > High Pass. And set the Radius value to something that’ll translate well to print, such as 3.0 pixels, and click OK. That leaves the image looking nearly completely gray. So double-click on the tiny settings icon to the right of the words High Pass in the Layers palette and change the blend mode to Linear Light (below). Click OK to accept the most intense sharpening effect you can achieve with High Pass.
That’s it. You are done. The final fake HDR portrait appears below. Check out the blades of hair, the healing scab, the threads of fabric, the fuzz and the wrinkles (on a 6-year-old!) and the bumps, the definition around the jawline and ears. This is a portrait with real tangible, textural, tactile depth.
If you look closely, you’ll see a light halo rising up from Sammy’s shirt. That’s a function of combining a high Amount value (90 percent) with a low Radius value (30 pixels)in the Shadows area back in Step Five. Undesirable. However, had we done otherwise, we wouldn’t have had a chance of brightening the eyes. I could mask this away, of course. But here’s my point: There are tradeoffs. Even so, it’s better to go “fake HDR” than try to shoot an authentic multi-exposure HDR portrait, which would require locking down the model in some kind of Clockwork Orange-style body brace.
It’s a trick that’s also a treat. And what could be better this time of year?
For a detailed examination of the Lab mode including this very technique—examined in luxurious, real-time detail, replete with sample image and everything—check out my video series Photoshop CS3 Mastering Lab Color, available exclusively from the excellent folks at lynda.com.