Upload & Sell: Off
| p.1 #3 · I've started down this light path, and now.... Whata mess |
Let's start with how the process should work from the beginning:
In the camera, to obtain a neutral color process control baseline, you want to set Custom WB off the gray card as step one. That creates an R=G=B = neutral "by the number" baseline for process control for your highly adaptive, not to be trusted eyeballs and brain. By setting custom WB you make your camera, not what you see on your monitor the trusted "gold standard" for when color is "technically" correct color-wise.
Step two is to ALWAYS record images with a full range of tone. That means exposing to the right for detail in a white brides dress (or white towel proxy) and a groom's black suit (or black towel proxy) based on what your camera sensor produces judged by eye both on monitor and print. The 18% gray card has no part in this process, because modern cameras do not render it in the exact middle of the white > black tonal range (i.e. center of histogram 128,128,128).
If you can do both at capture the rest is easy. The file will looks perfect on the monitor. Anything you do to it will mess it up, not improve it. The file when printed will have a full range of tone and neutral WB using the full gamut of the printer. Will it match the monitor? No. Should you expect to? No. If you do and start diddling with the file on screen to make the print match you will mess it up more than improve it. The beauty of profile based color management is that it works automatically if: 1) the file has a full tone range and neutral WB, and 2) you don't screw with it thinking you can make better.
The Gray Card? If your nose and toes itch do you scratch your belly? Then how do you think obsessing on the middle of the tonal scale will help you render the more important end points correctly. That might have been the case with negative film but it doesn't work with digital.
Why the Kodak Card is 18% is an interesting story. It was used as a graphic arts control long before Ansel Adams adopted it as the perceptual/technical process control point for his zone system. Apparently even Kodak doesn't know exactly when it was first marketed but some olld timers remember it being around in the 1920s. Back in Adams' day light meters like the Weston were calibrated to 18% reflectance, the same as the card. Thus if the card was held up and metered and the readings use the image of the card on the B/W print would also be 18% reflectance. Adams Zone system, in the days before spot meters, was done by metering off the Gray card to get the exposure for Zone V - 18% then holding the card up in the same light and the scene being photographed and perceptually comparing the card to the reflectances of the scene — you'd find "Zone V" in the scene by visual comparison or measuring different tone areas, and from that middle control point interpret its range and how to expose/develop it. 1° spot meters made the Zone System easier by allowing direct reading of the Zone 1 and Zone 9 areas with detail to compute the scene range and needed development time.
Eventually photo scientists developing the ANSI standard realized that 12-13% was actually a more accurate reflectance value for what appears to be perceptually the center of a gray scale from black > white and they adopted that as the new meter calibration point. That means today if you meter a white card and expose it per the reading on a hand meter or TTL meter in camera the resulting tone on the print will be 12% not 18%. If you "expose to the right" and accurately reproduce the white towel in a sunlit shot the spike from an 18% card will not be in the center of the histogram. But if you use a WhiBal card it will be because it makes its cards with 12% reflectance not 18%.
So why is the Kodak card still 18%? According to anecdotal accounts I've read Kodak had planned to change its gray card to 12% to conform with the new standard, but Adams fearing it would confuse his Zone System disciples and invalidate it's Zone 18% = Zone V = Perceptual middle of scene range premise. Adams reportedly went to Rochester and lobbied Kodak execs. relentlessly until they agreed to keep the card 18% and Adams Zone System franchise intact. I've lost the instructions for my R-27 Kodak Card set but recall the section on exposure stating that a reading off the card had to be adjusted by 1/2 stop to obtain correct exposure. That's to shift the 12% the meter reading would render it to its actual 18% tone. Kodak revised the R-27 card in November 2010 in collaboration with X-Rite (See press release): http://motion.kodak.com/motion/uploadedFiles/PCN110610_Q.pdf It is still 18% and still will not fall in the middle of the digital sensor histogram when files are exposed correctly.
So the long and short of it is you are aiming at the wrong target. Your goal should be fitting scene to sensor range on both ends, not pegging the middle to the 18% value.
The exposure dilemma is that outside of a studio using artificial light it is usually physically impossible to fit a full scene range to sensor. This ambient only shot was taken as part of a flash test facing west at 11AM. Exposure was set per the clipping warning to keep all parts of the white towel just below detail (i.e. exposed to the right, but with the clipping warning which is more accurate than the histogram).
Because all the important elements of the scene are "key lit" by the sun the photo, overall looks pretty "normal" which is to say as perceived by eye in person. The gray card isn't creating a spike in the middle of the histogram. Had I instead exposed per centering the histogram vs. highlight detail the highlights would have been blown. The 18% card just isn't an appropriate tool for setting digital exposure.
This next shot was taken a few minutes later facing east with the targets in cross-back light from the sun:
It demonstrates two things: 1) how short the sensor range is vs. the scene, and 2) how content and composition of scenes affect perceptual of exposure. It was exposed exactly the same as the first shot, per the clipping warning on the white towel to keep it under clipping. The sensor can only render about 6 stops with detail. I determined that with a simple bracketing test, shooting my gray card from the point it was overexposed and rendered at 245 (Zone 9) just below clipping with a lens wide open, then stopping down the lens until it was rendered as close to black as I could make it and retain detail on screen and print (16 - Zone 1).
The red lines are where the card spike was on the histogram, the tonal patches how that exposure rendered the card, and the numbers the eyedropper readings on the 0 - 255 histogram scale representing sensor range. The sensor on my 20D has a Zone 9 - Zone 1 detail range of just 6 stops. A sunny scene has a range of about 10. That explains the results in the backlit shot above. The front of the white towel, in shade is rendered around Zone V middle gray because skylight is 3 stops darker than sunlight on a clear day (Sunny 16 / Shady 5.6) and the camera has a 6 stop range so shaded white will be reproduced in the middle of it. The Zone V 18% card is rendered down around Zone 1 barely discernible from the black patch of the QP card on top of it.
That's the sad fact of life with digital. It's great in the studio where key and fill can be overlapped to fit scene range to sensor...
But it a huge fail in rendering an outdoor scene correctly....
The best we can do outdoors is to work around the technical limit of the too short camera range to fool the brain of the viewer into thinking the scene is better exposed than it actually is. One way to do that is illustrated in the west/east test shots at the same time of day: put the important content in "flat" lighting. I put flat in quotes because outdoor lighting isn't ever flat like flash on camera except at sunrise and sunset. All other hours of the day and night (by moonlight) light hit things at a downward angle creating shadow clues regarding 3D shape. When you put the sun at your back outdoors you might call the light "flat" but in terms of actual cause and effect it is modeling everything in much the same way a centered butterfly pattern models a face, downward at about a 45° angle with the shadows cast behind the trees in a landscape framing the brighter front in the same way the shadows on the side of the face in a butterfly pattern frame the highlight front of the face.
Our brains are easily tricked. That's what makes this huge con of rendering 3D stuff with 2D gradients work. Like any successful con artist or magician you need to know what the audience expects and how they will typically react, then use that knowledge to fool them ...
What to you think this is?
Going back to what I said at the beginning, on a basic technical level part of what makes the "con" work is being able to record a full tonal range, at least in the stuff the viewer cares about in the photo. That's why my first "flat" lit test shot of the gray card and towel seems "normal" but the back lit one doesn't. Had I instead opted to expose the second shot to make the card in the shaded the correct 18% gray (by spot metering and adjusting reading by 1/2 stop) the result would be more correct perceptually, but fail technically because it would blow the sunny highlights by 2 stops. But since the highlights are small and inconsequential most viewers accustomed to seeing outdoor shots exposed like that for 50 years — since the advent of color film — that would seem "normal" and par for the course.
Back in the days when all photographs were B&W the par for the course was a full tonal range all the time. That was the "secret sauce" that made Adams and his systematic approach famous, making it simple for any kultz with spot meter, darkroom timer and thermometer to make full range prints of any range scene every bit as good as his technically. The film didn't have a fixed range, but the print paper did, about 10 stops of negative density (range of 3.0) Adams's system adjusts development of each shot, per film development, to fit various scene ranges from a 7 stop overcast day to 12 stop cross lit beach scene to the paper range.
Color negative film has a range of about 6 stops, identical to today's sensors. So faced with a backlit subject a photographer could no longer hope to record the scene accurately with seen by eye detail everywhere so they started exposing "perceptually" not "technically" by exposing off the shaded gray card or shaded palm to get the faces "normal" knowing it would blow the sky from blue to white and turn the backlit hair into a nuclear halo. That became the new "par" for the course when forced to use just ambient light.
Flash had been used for years indoors with B&W so it didn't take a huge leap of intellect to realize it was a partial solution to the outdoor contrast dilemma. A single flash allows balancing foreground and background ...
But the Catch-22 is the camera still can't render the sunny background correctly. When exposed for sunny highlights everything ambient lit which is darker than white will be rendered 2-3 stops darker than "normal". Grey objects in shadows will be rendered black, and black objects will be rendered with nothing but residual noise...
Photoshop allows manipulation of the tonal range Adams would envy and no doubt use. Here is the ambient only shot with adjustment in ACR to pull out more detail in the midtones and shadows...
Overall it s a closer match to how my eye/brain perceived the tonal range in person as it scanned from highlights into the shadow, but amplifying the signal in the shadows also amplifies the noise to the point it doesn't look normal in terms of texture. But there is a work around to that technical limitation. Instead of pulling up all the midtones and shadows "globally" all the way with sliders I will adjust the RAW file with fill and brightness to the point where the noise isn't objectionable, then finish the adjustment selectively with adjustment layers and masks.
In the shot below I started with the ambient original (reduced JPG) and lightened the foreground with a screen layer, and then darkened the sunny highlights with a multiply layer in the areas depicted by the mask icons on the layers...
Being aware that the brain of the viewer will tune out the unimportant background if it doesn't distract I'll opt to keep it darker and more uniform vs. more "normal" in tonality by filled with eye catching noise that makes the image seem technically flaws. It's really just a con. It's still technically flawed, that's par for the course due the range of the sensor, I just play the con in the way the "mark" (i.e the viewer) will accept as being more "normal".
The point here is that while control of the process is a technical one, the process is severely limited anytime the contrast of the lighting exceeds the range of the sensor and making the photographic content fool the viewer into thinking it is real becomes less a process of recording what is there accurately and more of a "hide the pea" shell game so they will not notice the technical shortcomings.
One way to do that is photograph scenics and sports where you can't use flash with the sun at your back so all the important content in the foreground is "key" lit by the sun and the detail lost in the shadows isn't even noticed.
Another clever "con" is to put something interesting in the foreground lit with a full tonal range with flash. For example if you were to shoot the Grand Canyon facing the sun it would be a not be a very interesting shot because most of the detail would be in the shadows in a single exposure shot. But if you put an attractive lady in the foreground lit "normally" with a pair of flashes it changes the focal point the the viewer will be less likely to notice or care the background isn't rendered very well. The ideal situation is a view of the background with the light hitting at a 45° side angle which models 3D shape most accurately with the flash lit model with light at her back exposed with two flashes. There will still be shadows beyond the range of the camera in the background but they will not be noticed.
Something to realize about flash use outdoors is that the goal technically shifts from matching the entire scene to the sensor to matching just foreground to it, and making the flash modeling appear natural. That is after all the point of dragging the subject, flash and stands outdoors isn't it?
When using one flash it will match the exposure foreground > background but at the same time will kill the modeling on the skylight was creating. For example if you were to pose a face to the skylight — which has a brighter "key" and wrap around "fill" components — then add "fill" flash from near the camera axis what will happen on a cause and effect level is the flat fill will cancel the modeling of the natural light and the net result will be a fake looking flat flash shot similar to flash on camera indoors. But if you raise the flash off axis it's no longer "fill", it becomes a "key" light in that it now creates highlights and shadows with its direction. The parts of a subject it does not hit remain filled by just the skylight....
The shot above illustrates several ways to overcome the sensor range outdoors with flash and have the net result look as natural as possible.
Step one was selecting the location based on ambient light direction. Placing her back to the sun resulted in her nose pointing at the brightest part of the reflected skylight creating a natural "butterfly" pattern on her face. When standing together on level ground the angle of the skylight was so high her brow shaded the eyes so I had her stand below me on the river bank and look up, to get the light in the eyes. The location next to the river was also selected to create a light background to avoid her white jacket from becoming a huge distraction.
Step two was getting the ambient exposed below clipping using the clipping warning. The ambient only shot looked like the target shots, underexposed face, middle gray jacket in front. I exposed the white jacket 1/3 stop below clipping with the clipping warning at capture (i.e., Zone 9).
Step three was adjusting flash power it was "perceptually" correct by putting the shaded white highlights on "Zone 8" 1/3 stop below the sunny ones placed in "Zone 9". That was simply a matter of raising FEC on the flash until the front of the jacket was clipping then backing down 2/3 stops.
The net effect wind up as "natural" as I would expect spur of the moment single flash vacation portrait to be. It wasn't a planned portrait session, just a trip to a park with the camera along. What makes the result look natural was the fact the flash raised on my bracket with a diffuser 16" above the lens standing 3' above her on the river bank allowed me to match the angle of the flash to the angle of the natural skylight modeling the flash. By moving the flash off axis made it a "key" light, not fill, and matching the angle of the natural key lighting and not canceling it as hot shoe flash would have was the key factor in the overall look here. The flash angle complemented the natural lighting instead of canceling it.
What about fill? The beauty of the butterfly strategy is you don't need any outdoors because the - 3 stop sky fill where the flash doesn't hit works very nicely to frame and model the sides of the face. There is a darker shadow under the chin due to the flash and skylight being shaded but it was there before the flash was added: the shadow is naturally placed, its just made a bit darker by virtue of shading the sky fill. Had I placed my single flash on a stand to the side as many might do the result would have been a similarly dark unfilled shadow on the side of the nose and face which would be less flattering. The butterfly strategy for a full face pose works so well in part because it eliminates shadows in distracting places. So part of the solution here was knowing what lighting strategy to use to avoid that problem.
A second fill flash is needed outdoors when the "key" flash is moved to the side because the - 3 stop sky fill (vs sunny side) is too dark to produce flattering results. The cause and effect is the same as indoors. With a single flash to the side in front the shadow not hit by the flash are dark. The only difference outdoors is that they wind up middle gray (Zone 4 - 5) instead of jet black (give or take spill fill). Again using two flashes on the foreground is just a "shell game" visual con...
Exposing the foreground with dual flash allowed me to fit its range to sensor perfectly, just as in the studio. Custom WB ensured it matched the color expected during the day. The angle of the key flash at 45° to the right and 45° down creates "normal" modeling on the towel. The target looks as "real" and "normal" in the photo as it will get because it matches what I perceived in person — in the foreground. The entire scene? Not a match because the camera can't render dark backgrounds correctly in sunlit when exposed "to the right". The solution?
The same one that makes the "shell" game con work.... Keep attention focused where you want the viewer to think the important content or action is which distracts them from noticing anything else. The way to eliminate the distraction of any under exposed background in a flash lit outdoor foreground is to crop to the point the background isn't really noticed, or in the case of the outdoor shot of my wife start by selecting a background lighter than you want it rendered in the final photo. A shaded white wall in a flash lit foreground shot will be rendered gray. If you know that in advance based on understanding the technical cause and effect you and incorporate it into your overall "holistic" exposure strategy for capturing the scene as realistically as possible, despite the physical limits of the process.
I understand this stuff to the extent I do because I started out using the Zone System where the norm was always capturing a full range of detail . Next I went to work for Zucker who had figured out how to do that in every shot taken at a wedding in color by using dual flash for every shot: fill on bracket to record the shadow detail overlapped by the off axis key light. The goal was the same — record full range accurately — but the shorter range of color prints made it necessary to use flash to accomplish it. At a time when most were using single flash on camera and either blowing highlights in the bride's dress or losing detail in the groom's suit his candid shots taken on the fly at a reception were better lit in terms of pattern and full range than most of is contemporaries studio work.
50 years later we have the same dynamic range as color neg/print but exposure is far more critical. With color neg you could be two stops overexposed and still get a "normal" print because the color neg. film had a longer straight line response than the print paper. So you could "aim for the middle" exposing for 18% gray and get a full range of tone on the print. The only way to screw up was underexposing and losing the shadow detail.
With digital we ideally want to expose for the highlights to put detail in them, but whenever scene range exceeds sensor we must:
A) change the range of the scene artificially using flash or reflectors.
B) use lighting strategies that minimize the perceptual effect of the loss of detail
C) manipulate what the camera captured in Photoshop to fool the viewer into thinking more detail was captured.
D) all of the above.
That's just par for the course. How best to play the shot?
1) Expose ambient "to the right" per the clipping warning.
2) Look at the histogram's left side
Are the bars running off on the left (scene range > sensor)? Too bad you are in the rough. Time to use the rescue club. Flash, reflectors, bracketing on a tripod.
Are the bars kissing both sides of the histogram (scene range = sensor)? Probably a cloudy / overcast day with bland lighting but at least the shot is in the fairway and you have good chance to beat par and record detail everywhere despite the 6-7 stop range of the sensor.
Spot meters? Gray cards? Like 4x5 film holders and roll film relics from a bygone era before instant feedback. They will make you feel like you are better controlling the process, but the end results will not be any better than if you use the camera clipping warning and histogram. After all in the "good old days" what did you do to confirm the meter reading? Take a Polariod to confirm it. Now all you need to do is look on the back of the camera and understand what is is trying to tell you:
THE SENSOR RANGE SUCKS SO YOU BETTER LEARN HOW TO USE TWO FLASHES AS FILL AND KEY
Edited on Nov 17, 2011 at 01:30 AM · View previous versions