jzucker
Offline
• • • •
Upload & Sell: On
|
 p.1 #4 · Best Grey Card / Panel for luminance and color accuracy? | |
the problem with evaluating pictures on a screen is that it's subjective IMO. The only accurate method has to involve exposing a known grey color and then testing with the eyedropper in photoshop. If the card isn't a true 128,128,128 grey then you're screwed I guess. From an engineering perspective, the problem with the method you describe is that the white detail method is subjective and dependent on the output device calibration and how much light source it's exposed to as well as your own perception of detail.
cgardner wrote:
In the distant past Kodak 18% cards weren't color neutral but in recent years they have been made under contract by Tiffen, and currently in conjunction with x-rite and are Munsel Gray neutral. So while there are more expensive and more durable gray balance targets on the market the Kodak card will work for shooting to a technically neutral R=G=B starting baseline for color balance. A you probably know that starting baseline isn't always the most perceptually pleasing color or what you want in the final image to convey a warmer or cooler vibe, just a consistent baseline for judging the color when first opening the file on the computer to evaluate it.
As for calibrating your meter? An 18% gray card isn't the best tool for that task with digital for a couple reasons. You might be assuming the 18% card when correctly exposed will be dead center in the histogram when correctly exposed, but it won't be because digital cameras and meters are now calibrated for a 12-13% reflectance mid-point. Secondly even if you have a 12% reflectance card and center the histogram that may not mean your highlights are correctly exposed. By way of analogy metering to the middle is like scratching your belly if you nose itches: if you are concerned about highlight exposure you need to use a highlight target to calibrate the meter, not a middle gray one.
Gray cards worked for negative film because the negative had a longer linear section of it's curve than the print. You could err on the side of overexposure of +1-2 stops above what it took to put shadow detail on the negative and the lab when making the print compensated by with longer exposure of the print.
But with transparencies and digital, if meter readings which correctly expose white clothing, teeth, skin highlights, etc. is your goal the more logical target to use for calibrating the meter is a piece of white textured cloth like a terry towel.
Try this:
Drape a white towel or shirt over a light stand and illuminated with a single flash above the camera at about a 45° downward (butterfly) angle to create some modeling on it. Take a reading with the incident meter dome out, dome pointed at camera not light for the exposure reading. Let's say the reading is 5.6 on the meter. Bracket exposures from f/4 to f/8 in 1/3 stop increments. Open the 7 files on your computer and look at how the highlights in the towel are reproduced. Odds are the one with the best highlight exposure will not be the f/5.6 file with the best highlight detail in the towel or shirt.
Take the three best exposures (what looks perfect in the RAW, 1/3 stop less, 1/3 stop more) and using your normal PP workflow make 600 x 800 pixel JPGs — what you might typically use on a web page and evaluate the highlights again. The point of this second evaluation is so you allow enough exposure "headroom" at capture per the compensated meter reading so you don't wind up clipping the whites or blowing the red channel in the skin highlights when converting the 16-bit Photoshop files to 8-bit sRGB JPG.
The difference between the meter reading (e.g. 5.6) and the JPG which looks best would be the factor you'd enter into the meter. My Canon bodies (20D and 50D) require about 1/3 stop adjustment, meaning the file shot at f/6.3 looked better exposed in the highlights. That's because ISO100 set on the camera is actually around 120 per the way the meter is calibrated. The same lighting with a .3 compensation factor entered in the meter results in the meter display changing from 5.6 to 6.3 and the meter reading producing perfectly exposed highlights.
Where does the spike of the gray card fall? I don't really care because I control the tone of the midtones and shadows with the lighting ratio and it will vary with the ratio I use.
|
