This is a look showing common direct and bounce flash situations. Background is a very vague term, but a white wall is a predominately light colored scene that reflects well, maybe a white background. A distant background is something not very reflective, due to distance or maybe like a black background. Both fool the light meter. Distant backgrounds are an obvious reason the D lens distance info design watches overexposure. A distant background behind subjects of direct flash is frequently darker, simply because it is farther from the flash (inverse square law). That might be fine concerning the desired appearance of the background, but preflash metering sees the dark too, which then causes overexposure of direct flash on the near subject. Distance can be a worse case than the black paper background, less reflective, can be blacker. The D lens distance info is designed to help prevent this expected direct flash overexposure (for TTL BL mode). So below is another quick case to show this (specifically, a dark distant background). Still another contrived situation maybe, certainly not the best planned flash picture, but very real, and perhaps a more average situation, which shows what happens to the metering of dark backgrounds.
The images below are about TTL metering of Direct and Bounce flash, and are not about showing pretty pictures, and are not about "how to". It is about what happens, about "why" it happens in direct flash and bounce (and knowing why is usually as good or better than how to). All frames below are Nikon D300 and hot shoe SB-800, ISO 200, f/6.3, 1/200 second (manual), Matrix metering, no compensation. Ten foot ceiling six feet above. Nikon 24-70 mm lens at 70 mm. Actual distance 5 feet (1.5m), D-lens said 1.5m every time (so it is correct this time, not a factor for TTL BL). Blind point&shoot to see what actually happens. The dark or light backgrounds are intentionally not expected to meter a correct exposure (but flash compensation could always fix them). The only adjustment is that white balance was corrected.
1. GN mode flash mode, using distance and Guide Number. This GN mode simply computes Guide Number flash exposure for the five foot distance entered manually (and so this method is independent of the subject and its reflective colors, or the lack of reflectance from the background). It is computing inverse square law for five feet. Inverse square law says the wall at 16 feet is down three stops.
This is Basics, we should expect these things (maybe not the numbers, but the fact that this effect exists, and happens). The subject distance came out fairly well, but direct flash is not easy. Realize that the Inverse Square Law computes a 1/3 stop falloff between 5 and 5.5 feet, and 2/3 stop falloff between 5 and 6 feet. We're talking about inches.
(GN = 5 feet x f/6.3 = GN 31.5. 31.5/5'=f/6.3. 31.5/5.5'=f/5.7. 31.5/6'=f/5.2)
2. TTL BL mode, direct flash. Which does meter the scene, but yet in spite of the dark background, the exposure came out same as GN mode. Normally TTL BL mode reduces the flash level to balance it with the ambient, but there is insignificant ambient here, not a concern. The automatic metering is going to see the dark background behind roses, including the dark shadows from the preflash (simply part of the scene area now). System will meter it as dark, which will want to increase the flash to boost any dark scene brighter (towards middle tone), which will then overexpose the near subject (and TTL mode did that, see next). But in this case (direct flash), TTL BL metering was overridden by the D lens distance info double checking the computed GN exposure, which said "Oops, that can't be right." Overexposure was prevented, simply by substituting the more reasonable GN calculation instead of the metered exposure.
This D lens overexposure override works
1) only for TTL BL mode (normally default),
2) only for direct flash (flash head is level),
3) only on the hot shoe for communication, and
4) only with D lenses or later (since 1992).
3. TTL mode, direct flash. The only change was to select TTL instead of TTL BL. But TTL mode does not watch the D lens now, so we see the result that the metering does. It sees the dark background and made it substantially brighter, which puts the statue at the clipping point. I'd say at least 2/3 stop overexposed, so it needs -2/3 EV flash compensation. The TTL metering obviously was influenced by the dark background that it saw (just how life is). Since the goal of reflected meters is to make all scenes come out about mid-range, and this one started darker, so exposure was boosted, causing any near subject to be overexposed. We frequently see this with direct flash (dark backgrounds and overexposed subjects).
TTL BL in 2 above saw this same metering too, but then the D lens distance info could intervene as a safety check. That is what the D lens does. But not for TTL mode. The Nikon system is TTL BL by default, and TTL mode is an override. According to Nikon, the specifications for TTL Mode is to do its full thing, "regardless" of any ambient light.
Point is, the photographer is also aware of the distant/dark background, and knows he will need less direct flash exposure than normal, -EV flash compensation.
4. This is TTL BL Bounce. It simply aimed the flash head up. From #2, bounce lightened the whole room, so the background lightened up (without blowing the subject), no dark direct shadows, a lighter scene for the meter to see (needing substantially more flash power too). Now it needs about +1 EV flash compensation. The D lens is not a factor for bounce, since flash path distance is not known for bounce. No bounce card used.
The bounce lighting is more even, bounce shadows are very soft, contrast is not as extreme, a more natural look. The table behind the subject is lighted, the dark direct shadows behind subject are gone, and even the distant room is brighter. The meter meters preflash, so the metering sees that too, less "dark background" effect, so the foreground is not as overly bright. The ceiling acts like a very large umbrella up there (except it cannot be aimed), but the large surface makes shadows that are extremely soft and vague (mostly gone). Some are still visible on the "floor" and on the statue lady's neck. But the lighting is from above, is softer, and is a natural look we are used to. Bounce certainly seems worthwhile for hot shoe flash. A little flash compensation will help any of these difficult pictures.
5. This is TTL mode Bounce. Similar to 4, but slightly brighter than TTL BL, which is not unusual. I would still boost it maybe 1/3 stop. No bounce card used.
TTL BL often needs some +EV flash compensation, more than TTL mode, which might not. Just how life is. Both modes see the preflash when metering, it is what is metered. The goal of TTL mode is to set the flash exposure level as seen, regardless of any ambient. Where the goal of TTL BL mode is to reduce the flash to balance flash to the ambient (specifically, to not add enough to the ambient to cause overexposure). Indoors ought to be different, since the indoor ambient is usually low enough to be insignificant and ignored (unless using high ISO to bring it back up). But sometimes TTL BL needs a little more help than expected indoors.
Switching to Spot Metering mode will shift all Nikon iTTL flashes to be TTL mode. The Nikon Spot can be moved with the focus sensor, and it is fairly large, around 16% (FX) to 20% (DX) of frame height. Center Weighted defaults to 50% of frame height, but many models can resize it. I don't see much difference in these with camera in Manual mode for indoor flash, but switching to TTL mode (which will also give up any D lens help), can make a change.
4B. This is TTL BL Bounce, now with bounce card pulled out halfway. Compare card to 4, not 5.
The bounce card adds slight frontal fill (to lighten shadows from other sources). Adds catchlights in shiny stuff, we see an extra direct frontal catchlight just below the larger reflection from the ceiling. More said in the white background case below.
5B. This is TTL mode Bounce, now with bounce card pulled out half way. Compare card to 5, not 4B.
6. This is the same GN mode, direct flash. One beauty of GN mode is that the two GN shots (1 and 6) are the same manual exposure, independent of subject. The metering is affected by seeing this scene difference, but GN mode is not affected by the scene. Guide Number only computes the light level, knowing without seeing anything. Both background cases are difficult for the flash metering.
7. TTL BL mode, direct flash. The meter sees white instead of the dark now, and we know the white is going to meter brighter, and we know automatic exposure is going to compensate by turning it down to be darker (underexposure is expected). Automatic metering strives to make all scenes come out an average middle tone (gray, not too bright, not too dark). Therefore, (large areas) black is overexposed, and white is underexposed. It's just how things work, and flash compensation is how we fix it. Just saying, we know when we add the white background that +EV flash compensation will be necessary. I would add +1 EV flash compensation here, but didn't. Again, the point is, we know beforehand that this will be necessary here.
TTL BL came out even more underexposed than TTL in 8 next below, or 6 above, so the point is, the D lens was obviously not a factor, not used for underexposure for TTL BL.
8. TTL mode, direct flash. TTL is a little brighter than TTL BL, which is not unusual. The scene is white though, and a bit of +EV flash compensation would still be good.
The meter obviously thinks this is correctly exposed. The dark and the light areas average out to the middle tone goal always sought. The meter is far too dumb to know what it is seeing. It did do its thing. This is what meters do.
If you think all these pictures ought to always be correctly exposed, and if your goal is to simply trust the automation to always be right, then it's going to be sad. The light meter just sees some light, which it can measure, but it has no human brain to recognize what the scene is, or how it ought to be. That is what the photographer does. Get involved, learn to watch, and just do what you see you need to do. Soon you will already just know ahead of time. It is usually pretty much like last time in same situation — called experience.
9. TTL BL mode, Bounce. Not a great result (while uncompensated), but we expected to need compensation due to the choice of background. +1 EV compensation sure would help it.
TTL BL is often down a little from TTL. Which is just a technical point, this little difference in the modes (TTL BL vs TTL) is not very important. Every scene might be a little different situation, so we must always watch for need of flash compensation anyway. Does exactly how much really matter? Our goal is to fix it, whatever it takes. We know this will happen, so just start TTL BL with a bit of +EV flash compensation, ahead of time, same as last time for starters. For TTL BL bounce indoors, I would usually start with +1 EV flash compensation, the first time. It's no big deal (but always watch for special cases).
No bounce card here for frontal fill. That might be a big deal.
But the bounce lighting is more even, more natural lighting, less extreme contrast. Better due to bounce, but a bit underexposed due to the white influence. It is just a matter of applying +1 EV flash compensation, as necessary (not unexpected, due to the white background).
10. TTL mode, Bounce. TTL is slightly brighter than TTL BL in 9, but it is a tough situation. But bounce modes are more the same than is direct flash, between the dark and bright background pictures, because bounce makes the distant background less dark, and the foreground less bright. Bounce is affected, but Direct is more affected.
No bounce card used for frontal fill above.
But, The Big Point: The SAME meter was used in all the other cases, direct flash or bounce, and all dark and light instances. This is Not about meter accuracy. It is about the scene being metered, how it varies from a typical scene that actually averages out to be more middle tone. This scene is lighter, so we expect result to be darker (we expect to need flash compensation).
Repeat: It is correctly underexposed because it is a white scene, and the single goal of all reflective lights meters is to make all scenes become gray scenes.
When and if you want a different result, simply do what you see you need to do (flash compensation is how we control automatic TTL flash).
9B. TTL BL mode, Bounce, with bounce card pulled out half way. Compare card to 9, not to 10.
These pictures are not compensated, but the bounce card helped to fill it in. Five feet is a little too close for bounce from a high ceiling — the light must come straight down, without any forward angle to help light the front of the subject. Eight feet would be a better angle, and then zoom in if desired.
The main problem here is that the exposure is too dark, period. But a bounce card adds slight frontal fill. Adds catchlights in shiny stuff (specifically including human eyes). The card is direct flash, and it also adds direct shadows itself (easier to see here on the light background). We see the direct shadows come back (on a very near background), which are repeated here due to the bounce card. But they are weak, and the bounce fills them, so they are relatively insignificant now.
This is what the bounce card does, it provides mild direct flat frontal fill, typically to lighten shadows from another light source, here the bounce from above. The idea of the card is definitely NOT to blast away and obliterate all evidence of the bounce lighting.
10B. TTL mode, Bounce, with bounce card pulled out half way. Compare card to 10, not to 9B. And see next picture below.
Bounce card: Here the two 10 and 10B images are alternately shown, to show the idea of what the bounce card is trying to do, and does do, and to help see the difference the small bounce card adds. However, to look better, I used Adobe Raw to first boost the exposures by +2/3 stop, to both images (it's a white picture, they needed it). More flash power brightens the bounce card too, so I think this is fair.
The bounce card adds fill, direct frontal flat fill. The purpose is to lighten shadows. The fill should NOT be strong enough to eliminate the bounce shadows. It should just be enough frontal fill to lighten them a little, not so dark. Fill brightens the exposure of the entire image area, from frontal camera position, same as what the lens sees. More flash exposure brightens both bounce and bounce card, but which are from different directions. Size of the bounce card varies only the fill. Bounce makes soft and natural shadows below things, and the bounce card fills them somewhat.
See the bounce shadow on the statue's neck become lighter? True of all the bounce shadows everywhere, like in the cloth folds. Lighting is about the shadows (the natural gradient tonal shading which shows curves and shapes), so the fill should not be strong enough to obliterate the other lighting or the desirable gradient shadows. Are you aware of the jillion different shades of white on the statue? The shadows and tones and shading show detail and shape. We do NOT want to blow all that out with excessive flat frontal light (see #3 above). Ideally, fill just lightens the existing shadows somewhat, without being apparent itself, never recognizable as fill itself. If apparent, it's too much. The real purpose of showing this superimposed image is just to give the future mind's eye the concept of what the bounce card fill adds. We might debate more or less fill (a preference about contrast), but it's a good thing. See the catchlight appear? Catchlights are often a good thing — especially if in human eyes, where they add a desirable sense of vitality inside there, and doesn't need much power to do it. The catchlight alone is well worth it in the eyes of pictures of people (but usually missing in hot shoe bounce flash, unless the bounce card is added).
The bounce card adds some bad effects of its own, like the direct shadows came back. But hard to notice if kept mild overall, and the background certainly ought not be this close anyway. The bounce should remain the brighter source (with darker shadows) than the direct card (otherwise, you just have direct flash again). Bounce should be the dominant main light, and fill should be fill.
Ratio: The little pullout card can add more fill effect than you might think. It is not strong here, but the card is only pulled out halfway (five feet to subject, 14 feet to the ten foot ceiling and back at the angle, less the reflection surface loss. Point is, a little bit compares pretty well.) The ratio between bounce and card is the important factor. Bounce shadows (main light) should remain darker than direct shadows (fill). Tests of ratio are difficult, with so many other factors (the size and location of the spot on the ceiling, and angles of both the bounce and the card, etc. Tilting the head down for bounce distance changes the card angle too). My notions are that subject distance does not much affect the ratio because the bounce and the card both travel the farther distance, and they share the one flash power level. But ceiling height affects ratio. Bounce needs more power on higher ceiling just to stay even, but this power increases the card. However, not that simple, the angles have even larger effect. But card size itself clearly affects only the direct fill. Fill is about this ratio of the two. It warrants a few minutes of practice, with and without the card, to see what you get. You will soon have a pretty good idea what "just works".
Cards too big for bounce are my pet peeve. If you use a larger bounce card, you get more frontal fill. But big is Not better for the lighting. Big is flat frontal direct flash again. Subtle is better. Let the bounce work. We like what bounce does, and it would be good to be able to see it. Meaning, not much point of using bounce if we are going to obliterate its effect with excessive flat forward direct fill. The big cards may try to act like an umbrella, that's one thing, but note that the Big cards are outside or somewhere, and are not bouncing from the ceiling. Or ought not, IMO.
The above cases are not unusual flash situations, (distant backgrounds or white walls), and these are typical factors affecting flash exposure. Flash Compensation is quite important, standard procedure. It may not always be needed, but it should always be watched. Flash compensation is really Not about calibrating the flash or the camera, but instead is simply required by the different scene in front of the camera. So, we watch results, and we simply do what we see we need to do, to get the result we want. And this is real easy, maybe all but the first time. We can see the scene too, and can recognize dark/far vs. light backgrounds beforehand, so a little experience makes our first try usually pretty close.
The point again about the D lens is that direct flash TTL BL sometimes substitutes GN mode (if dark scenes or dark backgrounds) to prevent the overexposure which is frequently seen in direct flash. We should expect these things too, but the D lens check does not affect TTL mode. And it cannot affect bounce flash either, but bounce preflash more nearly illuminates the room background anyway, to not meter so dark.
White Balance: It is important too, and easy to mention here. I included a white card at the edge of the scene, and clicked it to correct all frames. Otherwise, they looked like this (uncorrected), so it is pretty important. After white balance correction, each one varies a little in temperature number, but Adobe Raw says the direct flash were around 6100K (blue) and the bounce were around 5000K (red, due to using near maximum speedlight flash level). That is expected too.
I tend to use this Porta Brace White Balance Card. 5x7 inches, plastic, durable, washable, inexpensive, accurate, about $5 at B&H or Adorama, and it is all we need. The Whibal brand white balance card is possibly technically better, as they claim to test each card, and it costs a little more. Not a thing wrong with either, but I have two of each, and I usually use the Porta Brace card.
There is more about metering on other pages here: