The earliest meters for lighting ratio measurement consisted of a card with two holes and a set of neutral density (ND) filters on a strip. The photographer would hold the card up in front of his eyes aligning one hole on the highlighted cheek of the subject and the other one on the shadow side. The hole for the highlights contained the strip of ND filters. The photographer would pull on the strip until the ND made the brighter highlight side match the shadow side visually. He would then read the ND value and ratio. A ND of .30 density units cut the amount of light in half resulting in a ratio of 2:1. A 2:1 ratio means the highlight side of the face, when viewed from the camera, looks twice as bright (one f/stop) than the shadows.

Today lighting ratios in a studio setting are made with an incident meter. An incident meter reads the light hitting the subject, not what is reflected and recorded by the camera. That accounts for the biggest difference between conventional lighting ratio and the wireless TTL (Through The Lens metering) system Canon devised for its EX flashes (580ex, 550ex, 430ex, 420ex and ST-E2).

An incident meter is placed near the face of the subject and aimed at each light separately. The strength of the light, not how much is reflected from the face is measured. Thus if key and fill are both the same strength the incident readings will be indentical. But long before electonic incident metering was invented the convention of expressing ratios in terms of the relative amounts of reflected light had been established. So the standard practice for incident measurements became measuring the incident light, but expressing the rato in terms of how the overlapped key and fill would reflect off a face when fill was behind the camera and the key light was 45 degrees from the nose of the subject; which is where a key light producing a "short" lighting pattern on the front mask of the face will always be in relation to the bridge of the nose.

So a lighting ratio obtained with a hand held incident meter isn't the strength of the lights, but an inferred assumption of how they will reflect. Again the underlying assumption here is that fill is on the camera axis illuminating the entire front of the face and the key overlaps it like this:

Those ratios will always produce the same results. In time your eye recognizes what a metered 2:1 or 3:1 looks like from the modeling lights and in the photo and on the screen and prints.

So now you get a Canon system, put A on your camera and B on a stand, dial in 3:1 and fire away.... Whoa! Something doesn't look right here. The shot is flat like it was taken with a single flash. Where is the modeling? Well in the Canon system A winds up being fill and B the key so a 3:1 H:S (i.e. key / fill) ratio translates to A:B = 1:3.

OK, so now we know Canon ratios are the reverse of 100 years of conventional use of ratios in still photography. We set the master ratio to A:B = 1:3 Whoa! Something doesn't look right here. The highlights are properly exposed, but the shadows are much lighter than I see at 3:1 with my studio lights (i.e. key 1-stop brighter (incident) than fill. So then I experiment, trying all the Canon ratios until I find an A;B which look on screen and print like my Alien Bees with a 3:1 ration obtained with incident metering. It winds up being 1:4

Why? Canon does its ratio setting using reflected light, not incident. It's ratios set up on the assumption that key and fill DO NOT overlap as in he studio lighting ratio convention. That's obvious just from the fact there is a 1:1 ratio. Think about it, there is no way to get a 1:1 ratio if two studio or any manual flashes overlap.

Now look at the illustration from the top of page 33 of the 580ex manual which I have reproduced below:

Note how the "A" and "B" flashes are each on opposite sides of the face. That's how the engineers at Canon apparently think a face should be lit. But that's horrible lighting for a face and not even very good for paper towels! Note the dark band down the middle in all of the test shots below taken with three 580ex in the configuration show above:

The Canon system computes ratios by first firing "A" group only, then "B" group only, then both together and then comparing the reflected brightness.

A:B
1:0
0:1
===
1:1 So if both sides reflect the same amount of pre-flash Canon calls it "1:1"

In that configuration either an A:B = 2:1 or A:B = 1:2 will produce shadows and highlights which are similar to a conventional overlapping pattern of two indentically powered lights:

You may have noticed in my copy of the illustration of A;B rations from the 580ex manual that Canon also apparently thinks we should buy a third 580ex or ST-E2 controller to trigger the off camera A and B flashes. The more logical and economical arrangement with the Master (Group A) on the camera or braket which I recommend in my multiple flash tutorial (shown below) is not shown in the 580ex manual or others that I am aware of.

When two Canon flashes are used as "A" Master (Fill) on a camera bracket with the "B" Slave (Key) off camera we enter territory the Canon system ratios can't comprehend. Remember those engineers who designed it and people who wrote the manual apparently assume A and B will be creating highlight and shadow separately without overlap. You may be thinking, "But A and B might not always be used in an way that they overlap!" That's true and there is no problem with that creatively, but we can't compute lighting ratios if they don't, at least not in the conventional way they have been used for 150 years.

Canon "New Think" is only necessary for TTL. When used in wireless manual power mode the Canon flashes are like any other manual flash, but with one significant difference. In wireless M mode the master still sends visible coded control pre-flashes to the slaves which will prevent the use of a hand held meter or other types of non-EX series optically triggered slaves.

With a single flash all we need to worry about is FEC (Flash Exposure Compensation). We take a shot at FEC = 0, "chimp" the LCD and histogram. If the light reflecting from the scene is fooling the TTL metering of camera we add more or less flash via FEC. When using two flashes in an A;B TTL configuration we still must use FEC to get the highlights exposed correctly, and then control the amount of shadow detail using the A:B ratio we dial-in.

It doesn't matter what the A;B ratio numbers wind up being compared to conventional H:S ratio numbers. We just need to know from experience what 1:1, 1:2, 1:4, and 1:8 actually look like on a face and which are best suited for babies, young kids, teen age girls or boys, young women, mature women, executive portraits, character studies of knarly old guys, etc. We can get that experince in about 30 minutes by shooting a face or other test scene with all the ratios from 8:1 to 1:8 and comparinging them side-by-side. That is the first thing I did when I got my pair of 580ex and my test is shown below:

You will note that the A:B ratios 8:1 to 2:1 look similar. That is because I used the Group A Master as fill on my Stroboframe camera flip bracket (connected to the hotshoe with the TTL extension cord 2). In neutral fill configuation like that the "A" fill will overpower the "B" key light until both are indentical in power where the hit the subject. At A:B = 1:1 the highlight/shadow difference is noticable and as the ratio increases to 1:2, 1:4, and 1:8 the TTL metering, once I had zeroed in on the correct amount of overall FEC, did a very good job of keeping the highlights correctly exposed as the shadows become darker. I would encourage all Canon flash system users to perform a similar test.