cgardner
Offline
Dedicated FM
Upload & Sell: Off
|
 p.1 #3 · Shooting in direct sunlight. | |
There are two factors to consider: how the direction of the light models the objects it hits, and how the contrast range of the lighting in the scene fits the range of your camera sensor...
3D Modeling:
Flat lighting gets its name because it makes what it hits look flat. Things look 3D in a photo because our brains match patterns of light / dark CONTRAST in the image to stored memories of real objects seen in person in similar light. Its why you gave that little darling brightly colored geometric shapes to play with as an infant; to train her still forming brain to make those associations. There's also a specialized part of the brain called the fusiform gyrus which is involved with recognition of faces and familiar objects. Its why you can quickly recognize a family member almost immediately in a crowd. Conversely it explains why you can't find your car keys sitting right under your nose if they are turned at an odd angle and don't look like your stored mental image of what your keys look like. Your eyes see the keys, but your brain doesn't recognize the pattern.
The most realistic 3D model of shape occurs when the dominant light is behind the object. So in terms of photograph you'll get more realism and depth if you put the camera where is is shooting into the shadows. That's not to say you always need to shoot into the shadows, it simply explains the cause and effect of how the brain creates the illusion of 3D. If you want something to look flat, such as if copying a book, you'd want to use flat lighting. But if you want something to look 3D the best strategy is to shoot into the shadow side.
Exposure and Dynamic Range:
Shooting into the shadow side created a technical dilemma. Your camera sensor can record about a 7 stop range of light with detail. You can find out what the range of your camera / lens by mapping the response of your sensor with a simple test with a gray card.
Put the camera on a tripod and frame the card in flat sunlight so it fills the entire viewfinder without any glare on the card. Set custom WB off the card so the RGB values will be the same in the file images (more or less). Put the camera in M mode, use your fastest lens, open it wide then find the shutter speed which overexposes the card to the point where it just barely clips in the playback warning. Then increase shutter speed by 1/3 stop so the card is just below clipping (i.e. eye dropper value of about 245). Then make a series of exposure by closing down the aperture 1-stop with each successive shot. As you cut exposure you'll see the spike the card creates on the histogram because there's only a narrow range of tones on the card march to the left on the horizontal axis. The red lines in the illustration above correspond to where the spikes were on my 20D histogram when I did the test. The little tone patches are cut/pasted from the file so show what tone the card was reproduced as. Its an 18% card, but it can reproduced as anything from white to black depending on how it is exposed.
If you examine ordinary images you take with the eye-dropper tool you'll find that the point where you can discern the difference between 255 specular highlights and detail is about 245, which is why I use it as a starting benchmark for the test above. On the shadow end its difficult to perceive any significant detail in values under 30 in part because the digital capture process creates artificial noise in the shadows. Between 0 and 30 much of what you see is amplified noise. The higher the ISO the more its amplified and noticed. So the range in f/stops between where the card was reproduced at RGB values of 245 and 30 define the dynamic range WITH PERCEPTIBLE DETAIL the camera sensor can record. The test showed that my 20D has a range of about 6-2/3 stops with my 24-70mm f/2.8 lens.. The lens is a variable because each lens will have different contrast characteristics due to lens barrel reflections and other variables.
A flat lit scene will have a range of about 10 stops on a clear sunny day. So that test of the sensor range tells me that if I exposed my file for detail in a brightly lit t-shirt I will lose about 3-1/3 stops of detail in the shadows. Exactly how much detail that is would require measuring the scene with a spot meter. Suffice to say things like shaded eye sockets will be very dark and lack detail.
Even if you shoot on an overcast day the brow will shade the light from above and make the eyes darker than the cheeks. The only solution for that situation where the front of the face is oriented to the direction of the sun is to raise the chin of the subject into the light so it reaches the eyes. Fill flash in that situation won't help because it will hit both the shadows and the highlights. When flash overlaps the ambient highlights and you add flash it doesn't reduce the contrast between highlight and shadows (the crux of the problem) As you add more flash you'd need to increase shutter speed to cut the ambient, which will just make everything darker, highlights and shadows.
This gets us back to creating the illusion of 3D. If you turn the subject so the light is at there back their overall form will have a greater illusion of 3D than if flat lit, but their face will be in light which is about 3 stops darker than the highlights on the shoulder of the white t-shirt where the sun hits directly. If you expose to keep detail in the shirt the face winds up looking very dark in the photo because your camera range is so short. Three stops from the highlights will fall somewhere around the middle of the histogram.
The solution is to add fill to lift the shadows. The reason adding fill will work with backlit subject is because the fill you add will not overlap and lift the ambient light if the subject is positioned to the ambient so no direct sun hits the face or head. As you add flash fill the shadow side gets lighter but the sun-lit side remains unchanged. Some areas where the rim light is visible to the camera may get blown, but that will seem natural perceptually. By virtue of being able to independently control the light level on the front of the subject separately what happens is that on the front of the subject the contrast range of the scene is reduced to match the range of the sensor. That allows the sensor to record detail EVERYWHERE in the foreground. The background beyond the reach of the flash will still have a 3+ stop loss of shadow detail, but in a situation like a portrait a darker background is actually a good thing because it makes the normally lit foreground contrast more.
The key factors when using fill flash are: 1) understanding that in backlight situations its really the key light raising it above the heads of the subjects for natural modeling; 2) understanding the role fill from the sky plays in a backlight scenario; 3) balancing the fill perceptually to the context of the scene.
Natural light comes from overhead. It creates a downward pattern of highlights on top and shadows below the brain recognizes as "normal" 3D modeling of the face. When the light creating the modeling gets to eye level the lighting looks flat because there is no highlight/shadow contrast created. When the light creating the modeling drops below the eyes the highlight/shadow pattern becomes the reverse of natural light and looks weird and unnatural in a way which your brain will sub-consciously react to, but you might not consciously be able to understand if you are new to lighting (i.e. it doesn't look normal but you don't have a clue why). The reason is that the brain matches the contrast pattern of the image to its memory of what faces look like in natural downward lighting and the patterns don't match. These are the patterns which do match:
When you use flash with stronger ambient back-light the flash isn't fill, its the key light for the foreground. The fill on the front of the face comes from God's own softbox, the northern sky north of the equator and the southern sky south of the equator. Remember that soft wrap around light is there, its just a bit to low in intensity for your lame ass camera sensor to use it effectively. The face will look too dark in just the ambient light, but when you use the fill flash -- raised over the head to create downward modeling on the facial features - it looks fine perceptually as the shadow values for the face in a full-face shot...
That shot was take with a single flash on a bracket with my 9 x 12 DIY foam diffuser. I picked a light background - light bouncing off a river - to make the clothing non-distracting, used the raised flash to create the "mask" pattern of natural highlights (the bracket makes it pretty much a given and a no-brainer) and the the sky in the direction she was facing lifted the shadows on the front of the face, which complement the highlights the flash added on top of them. It works because the flash complements the natural lighting instead of trying to fight and overpower it.
The other shot in the blurred mask example was taken with two direct, undiffused flashes to illustrate how much soft fill the sky will add if you use natural light wisely...
Getting just the front of the face highlighted in an oblique pose (i.e., short lighting) is a variation of "shoot into the shadow side" which I think reveals a face in the most natural and flattering way possible because the angle to the camera reveals the shape of the cheekbone with a nice compound curve from eye to chin, and the the lighting from the far side makes the highlighted front of the brighter, tricking the brain into thinking the face is slimmer and more symmetrical that a flat lit full or oblique view would look. Getting that pattern is really easy. Put the off camera flash 45 degrees from the center of the nose, whichever way its pointing.
Softness is also a perceptual illusion. Photons are neither hard or soft. Its how many direction the photons come from relative to the object which make the shadow edges distinct or fuzzy, but how dark or light the core "umbra" area of the shadow more than anything triggers the "soft" vs "hard" perception of the object the light hits.
So to make the lighting look softer in the test shot with direct flash I added fill from the camera to the light placed 45 degrees from the nose on the right side. Again, both lights were direct flash, but all both were actually doing is lifting the wrap-around sky fill that was already there.
Without any flash the face would be -3 stops below the light on the hair and shoulder in the sun. Perceptually I didn't want the flash to exactly match the sun lit parts. That's a recipe for flat lighting. So what I wanted was:
1) Sunlit highlights exposed just below clipping with ambient exposure
2) Highlights created with off camera flash about 1 stop below the sun highlights, similar to what would be seen by eye which adapts when looking at the face to make it seem brighter, but not as bright as the background ambient.
3) Fill on the shadows of the face with where about 1 stop below the highlights.
So from the baseline of the ambient only face being 3 stops below the sunlit side:
1) Add 1 stop flash fill to the natural fill, which raises the shadows on the face 2 stops below ambient highlights
2) Overlap the -2 stop (from ambient fill) with + 1 stop of light from the flash on the right to make to create the H:S contrast on the face.
So in that backlight + 2 direct flash lighting scenario the fill flash was only really lifting the shadows by 1 stop over natural and the off camera flash was only lifting the combined natural+ flash fill by one more stop. In that scenario the light bouncing off the sky did most of the heavy lifting. The flash just kicked it up a notch to the point where the camera could record it as it would be seen by our adaptive eye sight.
Here's the shot without the test card (which tends to skew perception of the face lighting) not blacked out:
So the points here are:
1) Understand the how light angle creates the illusion of 3D
2) Understand the range your camera can record with detail
3) Develop strategies to deal with the limitations of the camera's range such as shooting into the shadow side and using flash to lift the natural fill from the sky to reduce the contrast of the foreground to match the range the camera can record with detail.
I keep a flash on a bracket with a diffuser when shooting outdoors.
It is not the most convenient or comfortable thing to do when hiking to 11,500 ft. in the Rockies, but I feel the control of the contrast and lighting in outdoor situations makes it worth the effort.
The problem shooting with flash in direct sun is the x-synch limit of 1/250th on most cameras which puts you at f/11 when shooting at ISO 100. The solution to that dilemma is to use a system flash like the 580ex which has high-speed mode. The pulsing mode cuts flash range, but that is compensated for by shooting at wide apertures, which reduces the the amount of flash power need.
See this test I did with high speed flash. It explains the difference between technically correct exposure (i.e. don't blow the highlights) and what looks correct perceptually how average metering and human perception render it.
http://super.nova.org/DPR/Canon/HighSpeedFP.pdf
Chuck
|
