cgardner
Offline
• • • • •
Upload & Sell: Off
|
craig_oz_land wrote:
Chuck, according to Sekonic the histogram can be quite misleading for obtaining proper exposures as it works on reflected light and not incident and the camera tries to average it back to 18% grey.
No question about highlight clipping on the playback screen being of assistance.
What is your take on that?
Sekonic wants to sell more expensive spot meters... No surprise it would downplay the use of the histogram. Just use this common sense test: which is a more accurate indicator of how exposure is actually recorded by the camera, a meter reading or the last shot which was actually taken.
The key to using the histogram as a precise diagnostic tool is including a standard textured highlight test object. I use a white terry shop towel. I bought a bag of 60 for $10 at Costco. I define "perfect"digital exposure as the point where VISUALLY with just my eye balls the towel is reproduced accurately with its fine detail intact. In other words I use texture, not the histogram, as the benchmark, then corollate what the over-exposure warning and histogram spike from textured highlights look like back on the camera indicators.
I actually use the towel to calibrate my Sekonic L-358 meter to my camera. That is necessary for any meter because camera ISOs are not set to ANSI standards like the meter. Here's my latest test shot, annotated with the Photoshop eyedropper readings:
http://super.nova.org/TP/MeterCal6_3.jpg
Note - there was glare on the grayscale, but it wasn't a factor in the test.
If you read the links I provided earlier I provide a very simple way to benchmark a camera histogram by bracketing exposures on a gray card so the actual tonal value represented by each point on the horizontal scale is known. Here's the test from my 20D:
http://super.nova.org/TP/HistogramTest.jpg
So I can look at a scene, find the brightest highlights in it, then corollate the position of the spike that spot is creating on the histogram to the tone it will reproduce in the print. Conceptually its similar to the "pre-visualization" routine I started using about 35 years ago with the spot meter with Adams Zone System with B&W film. That is to say being able to corollate what the histogram is telling me to how the image will be recorded. The histogram is an immensely valuable tool, is one simply understands what it is saying.
The over-exposure warning "black-out" indicator is an even simpler, practical exposure guide. The basic law for digital is "Thou Shalt Not Blow the Highlights". Well, only an idiot could miss the fact that highlights are being blown if the playback is set to shot the OEW. If you raise exposure to the point areas in the photo were TEXTURED highlights are just starting black out, then drop the exposure back down 1/3 stop the exposure for the textured highlights will be perfect.
Since the OEW shows exactly where the highlights are being clipped it is possible to selectively expose, choosing to clip some areas to render others correctly. For example if shooting a backlit subject in sun I will have the subject hold the white towel, crumpled in a ball to create some shadows within it, then adjust the exposure based on the OEW on the towel. The background will be blown, but intentionally, with perfect exposure on the face in shadow.
As for evaluative metering, it actually works quite predictably. It evaluates all 35+ metering zones, finds the brightest one it thinks has texture, and sets exposure so as not to blow it out. When it errs, it will tend to do so in the direction of under-exposure, which can be remediated in post processing. My shooting workflow is quite simple:
1) Point and shoot a test shot at EC=0 and FEC = 0 (camera baseline)
2) Look at the playback, noting OEW and gap on right side of histogram
3) Adjust exposure based on feedback **
4) Capture perfectly exposed second shot.
** Of course the trick is knowing what the histogram and OEW are saying, a skill which can be acquired in about 30 min. using the exercises I've outlined in my tutorials.
As I mentioned previously, when any metering mode / evaluation workflow is used consistently it doesn't take long for noticeable trends to appear, such as nearly always needing about +1/3 FEC for indoor flash shots. Once I notice trends like that I adjust my initial baseline in step #1, starting at FEC = +1/3 rather than zero. The fact the camera needs 1/3 stop FEC dialed in doesn't make the metering "bad" it simply reflects the fact the metering is erring on the side of preventing the clueless from blowing the highlights. The assumption on Canon's part (I assume) is that once one notices that all the shots are consistently underexposed some EC or FEC will be used consistently to compensate. That's a much more realistic approach than writing off the camera metering as flawed and running to spend $500 for a spot meter.
I've got an entire tutorial devoted to trying to save people that $500 http://super.nova.org/DPR/Equipment/SpotMeter.html but if you already have one I also reveal how to calibrate it with a white towel so it can be pointed at the textured highlight you want to preserve, the camera set exactly as indicated on the readout, with resulting perfect exposure -- white with texture -- at the spot in the photo where the reading was taken.
Put the white towel out in the sunlight and take a reading from it with the spot meter. The indicated exposure will make it middle gray in the capture. Instead of doing the mental math of adding 2-2/3 - 3 stops to the indicated exposure to get the towel white, simply enter + 2.6 as the meter exposure compensation factor. That's just a suggested starting point. Adjust the meter's EC up/down until you can point the meter at the towel, shot at the f/stop / shutter indicated, and reproduce the towel perfectly VISUALLY IN PHOTOSHOP based on whatever subjective standard you have for what a perfectly exposed white towel shoot look like. What you will have done is two things at once: 1) moved the meter's default exposure point from 12-13% gray to about 95% white (w. texture), and; 2) compensated for any sensitivity difference between your meter's true ISO 100 and the camera's actual sensitivity of around 120-125.
Just try it! It will take maybe 10 min. of you life and save you 1000x more time futzing with the meter. Once you re-calibrate a hand held spot meter to the textured whites the workflow is:
1) Meter the spot where correctly exposed textured highlights are desired in the file
2) Set camera per the meter readings
3) Get perfectly exposed highlights
4) Optional for HDR: Bracket +1, +2, +3, and +4 stops**
** pick the one with the desired shadow detail during post-processing. I just use two layers, one exposed correctly for the highlights and a second with detail in the darkest shadows, placed over the normal one with a mask. I don't particularly like the "global" lighten-every-nook-and-cranny HDR look, so instead I selectively erase the mask just in the shadows where I want more detail. Even when I don't bother with a second exposure I can do this with a dupe / masked screen layer, or two files saved from the same RAW file with the second "pulled" to +2 to reveal more detail in the 3/4-Black tones.
Chuck
|
