Registered: Nov 18, 2002
Total Posts: 9376
Country: United States
The lighting ratio convention for portraits 2:1, 3:1, 4:1 etc. expresses the relative brightness of the highlighted side of a face or similar toned object and the shadow side, as seen by the camera in undefined units with the shadows aways a constant = 1. A ratio of 2:1 means twice as much light is reflecting off the highlights as in the shadows.
The convention pre-dates electronic metering. The earliest and very accurate ratio meters consisted of a card with two holes, one with a strip or wheel of gradated neutral density filters. .30 ND cuts light in half like a lens f/stop of if you wanted key one stop above fill you'd put the .30 ND filter over the highlight side hole, the clear hole over the shadow side then adjust power of the lights relative to each other until the two matched. The result was highlights reflecting 2x more light than the shadows, a 2:1 ratio. For a 3:1 ratio a .45 density would be placed over the highlight side, and so on. That would seem to explain why in the convention notation the shadow side is always 1 and the highlight number gets greater as the shadow get darker.
Ratios get even more confusing when determined via an incident meter. If you were to measure the lights set at 2:1 via the card ND "null" meter with an incident meter you'd find they were exactly the same incident strength. That wasn't any surprise back in the days of tungsten illumination because key and fill were typically identical light fixtures and to get a 2:1 ratio you'd need to place key and fill at the same distance.
F:1:1 even centered fill 8ft from face
K:1:0 identical overlapping key light 8ft from face
2:1 H:S ratio (reflected)
To get a 3:1 ratio the key light would need to be placed where it was "one stop" closer to the subject than the fill. Per the inverse square law that is 5.6 ft if fill is at 8ft.
F:1:1 even centered fill 8ft from face
K:2:0 identical overlapping key light at 5.6 ft.
3:1 H:S ratio (reflected)
The reflected H:S ratio is always 1+Key Power:Fill power because the convention assumes there is identical fill on both sides of face. There's no "rule" fill must be placed that way, but if it is moved off axis and lighting the right side of the face more than the left it simply doesn't follow the convention and the incident meter readings will not accurately match the reflected results. For example if equal key and fill are placed at opposite 45° angles as neophytes often do this is the result:
F:1:0 fill 45° to left
K:0:1 identical key 45°.
1:1 H:S ratio (reflected)
The result is flat light with a dark streak two the middle with criss-crossed shadows from the nose and very dark smile lines and mouth because no light reached those places. This illustration from one of my flash tutorials show a range of crossed lighting ratios:
Predicable results were obtained with primitive equipment by applying the laws of physics systematically. I first learned to set lighting ratios by distance in high school back in 1970 from a Kodak "How to Shoot Portraits" book with a pair of shop reflectors and 100W bulbs. Kodak assumed that is the type of lighting most hobbyist would use and provided instructions for setting lighting ratios by distance. I followed Kodak's tuition and got these results in my first portrait session (I saved all my proof sheets):
Seemed like pretty sound advice to me at the time, and two years later when I went to work for Monte Zucker I found he used exactly the same distance/ratio method for dual flash candid shots, always placing the key light "one stop" closer to the subject than the fill on the camera bracket for a 3:1 ratio.
Why a 3:1 ratio and not 2:1 or 4:1?
Apart from directional modeling created with lighting there are two fundamental qualities a color photo must ideally have for the content to seem realisitic to the viewer: the same tonal range (detail from shadow > highlight) experienced in person, and similar color rendering of known objects.
The 3:1 lighting ratio addresses the first one. If you were to put aside your meter and set key and fill entirely by eye on this subject you would want to wind up with a result similar to this with a full range of tone...
Lacking a meter you would likely adjust fill until you saw shadow detail recorded in the playback and key until the highlights in the shirt were below clipping. You could start with key and add fill, or start with fill and then add key afterwords, or start with both on and fiddle until the full range of detail is recorded faithfully by the camera just based on the darkest parts of the coat and the brightest parts of the shirt.
I would suggest you try it all three ways and compare.
If you start with the key light and set it for correctly exposed highlights first, before adding fill from over the camera you will find that once you get the fill to where it is revealing the shadows you'll then need to reduce the key light because the fill overlaps the key light and will blow the previously correctly exposed highlights. The same thing occurs when you have both lights on and adjust them simultaneously; it becomes a bit of a cat-and-mouse game to get both end of the tonal scale with detail. But if you start with the fill first and set it you'll find all you will need to do to arrive at perfect overall exposure is just raise the key light until its just below clipping.
During this process youl pay no attention to the face or the ratio of light it. In fact you don't even need a face around to set the lights that way, just a light stand where the subject will be and the a white shirt and black jacket, or proxies for the them such as white and black wash rags. But when you call the subject in and put them in the same placed you'll get the results seen above.
Fitting Scene to Sensor Produces "Normal" Baseline
The photographic process isn't some happy accident. It is a carefully engineered process designed to meet a perceptual goal: to match the range of a photo to the response of the human eye. Because of the way it is engineered when you do manage to record a full range of detail in white and black clothing simultaneously everything in the middle automatically winds up looking very "normal" identical to how you would perceive it by eye in similar light. The broader goal technically of the lighting ratio, beyond the numerical ratio on the face, is fitting the range of the scene to the sensor .
A 3:1 ratio worked very well for color negative film/prints because it just happened to be the lighting ratio which when set for the face would also wind up recording the shadows and the highlights accurately. The 3:1 ratio works the same magic with digital sensors because they have about the same overall H:S dynamic range (DR).
Finding Your Native Sensor Ratio
DR is not the same for all cameras because sensor DR varies with the sensor site capacity to store photons. Sensor sites are like buckets. Some in the highlights get filled very fast, some in the shadows very slow. In high contrast lighting like a backlit outdoor scene the highlight sites will usually fill and end the exposure before any light is recorded in the darkest shadows. The actual metering is done off the viewfinder before the exposure (why it sucks so badly most of the time) but it predicts when the brightest areas will clip (max. out) the sensor.
That's why the native "fit scene to sensor" ratio need for a FF pro body will vary from that of a 1.6 crop body. On the same size sensor as more and more MPs have been added the camera makers have found other clever ways to increase photon capturing ability and keep DR relatively constant but higher MP sensors will usually have slightly shorter DRs. A 3:1 ratio on my camera may not look the same as yours. It doesn't even look the same on my two cameras: the DR on my 8MP 20D is longer than my 50D.
Indoors with studio lighting the DR isn't a limiting factor because fitting a full range scene to any DR with artificial lighting is simply a matter of adding enough fill until the shadows are recorded, then overlapping the key light to just below clipping. If a sensor only had a range of only 3 stops you could still record a full range of tone, you'd simply need a lot more fill. The reason B&W film can handle direct sunlight contrast without a problem without the need or fill is that it's range can be adjusted to 12 stops or more. A digital camera's effective range is about half that which is why an outdoor scene exposed for highlights winds up looking like this:
But it can wind up looking more normal, like this by adding flash with the same overlapping key over fill strategy. The only difference is the flash is added over stronger ambient light outdoors than indoors...
Outdoors as indoors in the studio my goal is to make it look real, not hit some numerical ratio target but experience tells me that same 3:1 key: fill ratio that works indoors to fit scene to sensor also works outdoors.
But only on the flash lit foreground. The camera sensor can't handle the contrast in the ambient background, which is why digital cameras are not great for shooting landscapes unless HDR is used.
What about when "normal" isn't the look you want?
Try taking a subject sometime wearing a dark suit and white shirt and keep the pattern and highlight exposure the same change the key/fill ratio from a "normal" looking 3:1 (key = 2x fill) to 2:1 (key=fill) and 5:1 (key = 4x fill) . Then show them to someone who knows nothing about lighting technique and ask them how the lighting makes them react emotionally to the subject.
Even beyond the goal of fitting scene to sensor to make the photo seem real the broader goal of the exercise of taking and sharing one to evoke some emotional response in the mind of the viewer. If the photo can't do that what is the point of taking it?
Human perception works on several levels to discern things. Overall shapes, contrast with the background, will tell the brain what might be, with additional clues about shape on the front provided by frontal lighting. The object is recognized when the brain finds a match to some previously seen object or face...
The basic perception process has been studied by psychologists and is predictable because our wiring is similar, but subject to cultural variation. For examine Westerners will scan to the eyes first on a face but Asians will instead fixate on the nose because direct eye contact is considered rude. Studies have found Asians will fixate on and remember background context in photos more than Westerners who pay more attention to the foreground.
What is more individual is the emotion response to the content based on familiarity and experience. If you love cats and body modification you'll process a photo of a woman with tats and studs carrying a cat similarly to someone who hates both, but your emotional reactions will be different. But emotional reactions to things like shadow tone, lighting patterns that shade rather than highlight eyes, body language in posed and expressions are predictable.
Goals should drive strategies...
Lighting ratios work best when they match the context of the mood implied by the expression of the subject and the environment they are seen in. In a series of photo with the same expression and pattern but progressively darker lighting ratios the perception of the shape of the face will not change much. What will change is the viewer's perception of the subject's emotional state and the environment.
Beginners all seem to want to learn the "Rembrandt" lighting pattern because some famous dead painter used it in a self-portrait. But they fail to notice he was a gnarly old man. As a result their first portrait of the smiling wife with the same pattern and 4:1 lighting ratio makes her look like a gnarly old man.
"Rembrandt" is a good lighting strategy when you want to make a subject look old, reclusive and grumpy, but all things considered the wife will probably like her portrait more and complain about the money spent on the lighting gear less if you photograph her in a full face pose with a butterfly pattern and 2:1 ratio or an oblique short-lit view with the same light toned, wrinkle hiding low contrast lighting ratio.
Because only one ratio fits scene to sensor adding more fill to change the ratio to 2:1 on the face then adjusting the overall exposure to keep the highlights the same will make the shadows abnormally light and washed out...
The face may look "softer" but without a full tonal range the photo will lose it overall "normal" look because the shadow clues aren't the same as seen by eye. But changing ratio this way does have the benefit of reducing shadow noise by virtue of more signal in the darker areas. I will sometimes overfill at capture to minimize noise, knowing I can adjust the shadows back to normal when editing the RAW file or more selectively with adjustment layers in CS5 like burning in a print on an enlarger.
Another way to maintain the suit>shirt detail and make the face lighter is to first fit the range to sensor with key and fill, then add a reflector near the face as in this set-up shot...
Making shadows darker than "normal" is a bit more problematical. Any reduction in overall fill on the suit will result in a loss of detail. Moving the fill lower to put it further away from the face will result in an upward fill direction. Natural fill from the sky comes from a downward angle, which is why low under-chin fill if too low will look odd: because it creates darker areas on the top of the cheeks, lips and chin; the opposite of natural sky fill.
A better solution is to keep the fill in the same place — centered around chin level or even raise it a bit more and aim it down away from the face and/or switch to a smaller fill source to better control its direction.
Photographers who automatically associate soft lighting and big modifiers and never try very small ones may never discover their benefits in controlling where the light doesn't go. Mood is controlled with shadow tone and that is controlled with fill placement. I don't change the overall ratio from the "normal" baseline to create darker moodier lighting, I just "feather" the fill away from the face.
I will usually start all my portrait sessions with a butterfly pattern with both lights back at 8-9ft near the camera, set to record a full range of tone. I set the lights using a target as seen in the set-up shot, usually before they arrive.
The overall look is similar to the contrast of open shade outdoors with no harsh distracting shadows. Keeping the lights 8ft. away minimizes the rate of fall off front > back on the face and body and foreground > background making it easy to light full length shots evenly. It is a very flattering and forgiving pattern because the subject is free to move around without changing the the lighting pattern on the face in any unflattering way.
Both those kids (my neighbors) have nearly perfectly symmetrical faces. Not everyone does....
I analyze faces before shooting and usually spot asymmetries, but I'll still shoot some full face butterfly lit shots anyway to help come up with the best strategy to try to minimize them with an oblique angle. This is a "lazy" oblique view where I changed backgrounds but didn't switch from the centered lighting pattern...
.. but seeing it wasn't working well (note the shadow hanging out behind the nose) we took a break and I changed to a 45° from the nose "short" lighting strategy.
One of the reasons I use a 3D target like a towel for set-up is so the subject can go change or chill out in the next room while I reset the light for the next strategy. I haven't stuck a meter under anyone's nose for several years now. The 3D target allows me to see the interplay of rim light, key and fill and ensure the full tonal range will be captured. On average it takes only about 2 minutes to move the sliders on the remote control for my lights next to the camera and take a few test shots to dial the ratio and exposure. Then I call the subject back in and start shooting...
Those photos were for his confirmation so I wanted them to have a boy becomes young man theme this darker more serious profile shot. Above is the result out of camera from the RAW file, all I wound up doing is cropping...
The overall ratio stays true to my Zone System compulsion for always recording a full tonal range in any photo I take. The look of the lighting was changed by using a very small key light and moving the fill from near the camera to the front of his face so it would fall off front>back from the tip of his nose just as in my normal fill strategy for the oblique view: key and fill are at the same angles relative to the face. It's a trick learned from window lighting where you don't move the lights you move around the lighting. The key light is the hair light from the other shots which is a 16 x 22 SB with a 12" circle mask and 40° grid. Fill was a med SB with a 20" circle mask, both used close to make the footprints small and the fall off from the sources more rapid than in the other shots.
I didn't meter any of the shots, or worry about what the ratio was. I expose as I did with the Zone System to place print values (Zones) on specific places in the image, using the target as a proxy...
First I'll set the brightest highlights in the shot, which for that set-up was the rim light. I raise to to the point of clipping then backdown 1/3 stop using the clipping warning not the histogram.
Next I turn on fill and adjust until I see shadow detail in the playback I gauge it by the black patch vs frame of the MacBeth target or used a black towel.
When rim light is used in back the key light needs to be one zone darker than if rim light isn't used to the two combined to look "normal". Setting Key light is very simple. Raise it to clipping then back off 2/3 stops until it is 1/3 stop below the rim lit detail.
If using a white background I will raise it to clipping, even the lighting, then lower it until I can see the rim light contrast. It winds up the same as the key light foreground, about 2/3 stop below 255 clipping...
It works for me because I never use 255,255,255 white backgrounds. That is a poor strategy perceptually for web design because it makes all the white tones in photographs look duller and darker by comparison...
Notice how the same values look on a 128,128,128 background. Overall the dots look pretty similar even with side-by-side comparison, but note how the 255 catchlight dot contrast more on the darker 240 "Zone 8" tone vs the 250 "Zone 9" white.
Maybe this will give you some food for thought, maybe it will just cause you to gnash your teeth and clutch your incident meter closer to your heart. Either way you will not find out if it will work for you unless you try it. You might find you like it