It's a long article, with a lot of talk first, but don't skip the heart of the problem below. It can seriously affect the Nikon default TTL BL flash exposure. The first half of this page is about how it ought to work. The last half page is about how it often in fact works instead.
The obvious major function of TTL BL flash mode is to provide automatic balanced flash in sunlight (automatic reduced flash level for proper fill level in brighter light). See Part 4 about that. The story here is about a subtlety, and about accuracy, which necessarily brings up strong basics of flash metering, even more important in their own right — the good stuff we need to know about.
Spot, Center, or Matrix metering choices only affect metering the ambient. The flash has its own system, which does not use Spot metering (the flash exposure does not use Spot Metering, but the ambient metering does). However, TTL BL "balances" the flash exposure with the ambient (TTL BL reduces flash level to be fill flash, to not overexpose the sum with ambient, but TTL mode does not, it comes ahead on). But if Spot metering, only the small spot area matters (to ambient), so there is no concept of background ambient to balance the TTL BL flash with — so the flash system switches to TTL mode if ambient is doing Spot metering. Spot metering is a way to switch the flash to TTL mode — however it may also greatly affect your ambient if ambient is normally bright (and meterable).
Incorrect reporting of D-lens distance in zoom lenses is one common cause of underexposure in TTL BL direct flash. Not speaking of focus, the focus is OK, but the distance reported is not very accurate. D-lens distance data is used by metering of TTL BL direct flash. Bounce flash (tilting or rotating the flash head) does not use (ignores) the D-lens data. TTL mode does not use D-lens data. Commander remotes do not use the D-lens data. Or, if your camera model (with Commander) has the FV Lock function, FV Lock will simply ignore the D-lens data, and will use the direct TTL BL metered value.
Caution: A TTL BL non-tilted flash head (in an umbrella for example) on a hot shoe extension cord (Nikon SC-17, SC-28, SC-29) will use D-lens distance data, even if the TTL BL flash is at a different distance than the lens. Nikon advises to switch hot shoe extension cord flashes to TTL mode (but they no longer provide the obvious way to do that). Spot metering or FV Lock or tilting or rotating the flash head are work-arounds.
Regarding TTL BL direct flash, first how D-lens distance works correctly. Then farther below, how it often works incorrectly.
Next is a quick test about how Nikon TTL BL mode uses (or does not use), the D-lens distance info for hot shoe TTL BL direct flash. This test is NOT about metering, it is about how an incorrect D-lens distance can override the TTL BL metering. All of the Nikon lens made in the last 20 years are a D-lens (with the embedded CPU chip, i.e., a CPU lens). The D-lens can report distance at the focus point (it is not very accurate, but sometimes it is close). This test will show that the D-lens can prevent TTL BL direct flash overexposure, which can frequently occur in direct flash due to dark backgrounds (from inverse square law falloff). But underexposure of TTL BL is not affected by the D-lens. And sometimes the D-lens distance is simply wrong.
The test uses the fact that we know reflective meters will overexpose black scenes, and will underexpose white scenes. Large areas of either black or white typically come out gray, because that is simply what the camera reflected meters do. So expecting that, a pair of black or white scenes were used here, and without being too fussy or precise (no attention was given to compensate exposure), we can consider "proper" exposure to be white that is white, and to be black that is black, instead of both being gray. (Genuinely useful info here would be to also mention that we ought to learn Flash Compensation.)
The D-lens distance can sometimes correct overexposure in direct TTL BL flash. And sometimes its error can cause underexposure.
All that flash exposure can do with distance is to compute Guide Number. The test method here is to compare TTL BL results to the SB-800 GN mode (Guide Number method). If they both use guide numbers, the exposures ought to match. We have to know distance for the Guide Number method, but if the numbers are known, GN is usually a pretty good result, better than some — because, GN is independent of the reflected subject colors, similar to how incident metering is also independent of subject colors. These two methods (GN and incident meters) give an exposure based on the actual light level instead of being about how well the subject's specific colors reflect light (downside of reflected meters).
The pictures below: Nikon D300, f/8 1/125 second, ISO 200. 70-200 mm D-lens (86 mm) at 5 feet (this is a good lens case, the D lens distance was 4.92 feet, 98.4% of actual, which is close enough — but some other lenses can't do that, more below.) SB-800 hot shoe direct flash. No compensation, all point and shoot, except the distances were measured for the GN mode. The only adjustment is white balance.
The backgrounds are black and white paper. I always use Center Weighted metering myself, but this is instead Matrix metering for those that think it might matter (it doesn't for flash indoors, because flash uses its own metering area in the center of the frame). Again, this is not about theoretical perfect exposure, as there was no attempt here to compensate anything as would be otherwise expected.
The first case was in fact an all-black scene, and the results below are to show how it is SUPPOSED to work, when things go right (the rest of the page complains often things don't go right). The results seem dramatic enough, and obvious enough, we can see it at a glance. These exposure results are basic fundamentals about how reflective metering works.
The TTL results did overexpose the black scene (as normal and expected of reflective metering). But the TTL BL results came out like the GN results instead. This is a case that worked. This is due to the effect of the system checking TTL BL direct flash with the D-lens distance info, computing the guide number, and saying "Whoa man, that's too much flash power for this subject distance", and limiting overexposure via the guide number distance check. Which can be a good thing (as here), and Guide Numbers are independent of the scenes colors. We might prefer the brighter gray result, but the true color of this scene is black.
The TTL results did underexpose the white scene (as normal and expected of reflective metering), and the TTL BL results came out like the TTL results. This is the same metered exposure, but just with no effect from the D-lens check on underexposure. The D lens check only guards against overexposure of TTL BL direct flash.
But the D-lens distance info obviously DID NOT affect the TTL BL expected underexposure of the white scene (D-lens distance never increases flash exposure). The TTL BL result is slightly more underexposed than TTL mode, as is routinely typical of TTL BL. I suppose it is still trying to "balance" with the dim ambient (a normal indoor scene here, but dim compared to the flash), reducing flash level enough that the sum of the two does not overexpose.
The big point here is that the 70-200 mm lens at 5 feet did NOT report a too-short distance, and so D data was ignored, allowing the TTL BL metering to be used as metered. It was still regular reflective metering, never ideal, but everything came out as expected.
I focused on the horse's nose. Note this area includes the dark direct shadow. We don't see that dark shadow when focusing, but metering of TTL preflash obviously sees it, which is a significant dark area in the center area. But this scene is white overall, which is a strong influence the other way, towards underexposure. The point here is, the meter meters what it can see. White scenes typically underexpose. Or TTL direct flash metered on camera obviously frequently includes preflash shadows and dark open backgrounds, which tends to overexpose. For bounce, direct shadows disappear and backgrounds are lighted better, which may tend to underexpose, because the metering of the TTL preflash sees that too. Photographers can learn to see the scene, to expect this before the shutter. Call it platitude if you want, but experience is literally about expecting the same result as you saw last time in a similar situation (but for experience to help, you do have to think about it).
Which picture is the correct result? Possibly none of them are exactly what we would want. These are extreme subject situations, so we know that some attention may be required. The camera automation is a starting point, but the photographer will need to pay attention, and then do what he sees he needs to do (flash compensation), to ensure getting his desired result. But the photographer did not get involved above, here he relied only on the automation.
This test used an older D300. Nikon sometimes tweaks things, so just to check, the newer D800 produced very similar, but slightly different results (same concept, viewed here). It does not affect the discussion. And problems are not due to the camera anyway. Problems (shown below) are because the lens simply often reports inaccurate distances, and the camera believes it.
TTL BL mode is the "only" automatic flash menu choice available for the Nikon internal flash, or the SB-400, SB-500 and SB-700 external flashes, or the Commander menu for Remote flashes. Plain TTL mode is not available. Their menu may be shown generically abbreviated TTL, but is a TTL BL system, and it is TTL BL by default — Unless Spot Metering is selected, which always switches to TTL in every case (no concept of background or balancing with Spot metering, but Spot metering introduces its own issues). Only the current SB-910, and discontinued SB-600/SB-800/SB-900 flash models have a menu to inform the camera to select TTL to override TTL BL.
Theory: Nikon only says a few vague gee whiz marketing words, and does not tell us how the system works. Automation mysteries: matrix, databases, balanced flash, etc, are just labels, but no details. They say 3D Matrix metering takes into account the subject's distance, but never say how and why, or when and how often. Cameras don't even know distance. The D-lens is a crude tool attempting to approximate it from lens focus rotation. The subject is assumed to be at the focus sensor, so they already know where that is, and distance does not matter anyway, there is nothing sunshine can do about it. Subject distance only matters to direct flash, when it is very important. The D-lens system is flawed, accuracy is too poor to count on, especially flawed by zoom lenses that report focus distance incorrectly at different zooms.
My notion of the concept is this: Direct flash into an open background often sees a lot of dark background at distances greater than the subject (inverse square law, it is dark back there, the flash does not reach that far). Metering sees that too, and its goal is to try to make things brighter, which frequently causes overexposure of direct flash. So the D-lens distance and communication with the flash can compute guide numbers, and if it thinks the metered flash power exceeds what is reasonable for the distance, it cuts back on the flash power. We can see this happen, shown here, above and below (it is only a factor for TTL BL direct flash, plain TTL mode is as metered with D lens distance not used). The D-lens distance reported by zoom lenses is not always near correct however.
D-lens: Nikon lenses (since 1992) have a CPU chip in them, which can report focused distance to the camera (judged by the rotation of the focus ring). Those with it were called a D-lens (distance). They all have it now, but they don't bother to say D if they have later features to tout instead. Later lenses are called other names (G, AF-S, etc), but they are still D-lenses too. The D-lens distance info can be a watch guard for hot shoe TTL BL direct flash, protecting against overexposure for direct flash (TTL BL mode), with the flash head is level straight ahead. There is an electrical switch to indicate when the flash head is tilted or rotated, because the GN distance applies only if direct flash. Guide Number is not practical for bounce. We can't measure the bounce flash path distance very easily (up and down), and there are losses at the ceiling reflection surface too. So TTL simply meters a preflash to measure the reflected light directly (instead of the distance). But here, we are speaking here only about direct flash.
GN mode: (NOT part of TTL BL) The purpose of including GN mode here is to explain the D-lens distance info effect, for the obvious comparison above (SB-800 menu shown here). GN mode is a manual flash mode, but it computes the manual power level from Guide Number. The hot shoe flash knows aperture and ISO and zoom from the camera, and it knows the guide number chart too, so mode GN looks up the right flash power level for this distance we enter (the D-lens distance is too inaccurate for GN mode to use). GN mode is basically just a calculator for the Guide Number system, and it is pretty good stuff for direct flash, if you know the distance. Again, it only works for direct flash, and you enter the distance manually. Nikon calls the mode: Distance Priority Manual Flash. Unlike the TTL modes, a really wonderful advantage of GN mode is that it is independent of the subject colors (independent like incident metering in that way), so that generally, if the light itself is made right for the distance, black things will come out black, and white things will come out white. But GN mode is direct flash only, and you have to enter the subject distance. It works.
The D-lens Logic: In this way, the TTL BL direct flash power level is compared to the level suggested by the flash distance (guide number is the inverse square law requirements). If the metered direct TTL BL flash value seems excessive in comparison to the reasonable logic of Guide Numbers, then the TTL BL exposure is reduced, to avoid overexposure (could be caused by metering dark direct backgrounds). Rhetorical, but how else could metered flash use distance?
For flash, the D lens distance is considered only for TTL BL mode direct flash. NOT considered for TTL mode, and/or is NOT considered if the flash head is tilted or rotated (bounce flash, which is a very different flash path distance).
The issue causing attention to the lens distance is that direct flash often sees an open background as being dark, due to the inverse square law falloff. TTL BL is a metered system, and it is affected by the dark it sees (intends to do its best to lighten it, but some things are supposed to be dark). But with D-lens data, the system can also still do the same Guide Number check, and for direct flash, TTL BL watches distance. And when this says "What in the world is that meter doing now? This flash level is too much for this distance that is perceived", then direct flash can reduce the flash level to the GN value, to prevent probable overexposure (with good results seen in the black pictures above). Using this D-lens distance does make the assumption that the subject (where we probably want the flash to be correct) is same as the focused distance. In contrast, the actual GN mode does not know about the subject, and we enter its distance manually — which is less convenient, but more versatile, and has advantages.
Use of a Hot Shoe Extension Cord: If you put the flash off-camera on a hot shoe extension cord (like the Nikon SC-28 or SC-17 cord), the camera still thinks the flash is connected to the hot shoe (CLS communication still exists). Then TTL BL direct flash can be adversely affected by the camera lens distance, instead of just using the metered flash from the actual flash distance. The flash and camera distances are the same when the flash is on the hot shoe (and if the reported distance is correct). But if using the hot shoe extension cord, the flash and camera are probably not at the same distance. If the flash is closer than the camera, probably no big problem, the system should use the TTL BL metering then. But the flash being farther is a real problem (for TTL BL direct flash). Actually, it is not so much about the flash being farther than the lens, it is about the flash being farther than the distance the lens may inaccurately report. It can work well if the reported distance is correct. But if the flash head is not tilted or rotated (still at regular 90 degrees to the body, for example even if in an umbrella with flash head aimed away from subject), then TTL BL exposure can be affected badly by an incorrect D-lens distance (camera distance). The D-lens report says "Whoa, that is too much flash power for this reported lens distance, and then it cuts back on the flash power (to prevent possible overexposure). It doesn't understand the situation. It affects TTL BL direct flash mode. The fixes include using Manual flash mode, or a tilted or rotated flash head (bounce), or actual TTL flash mode, or Spot Metering to convert TTL BL to be TTL mode. Or using the wireless Commander mode knows not to allow the off-camera remote flash to respond to D-lens distance. Or using FV Lock will ignore the lens distance info too.
Some claim TTL BL mode "always uses distance" to compute its exposure. It doesn't, but there are some big Ifs and Buts. Distance is exactly what GN flash mode is, but TTL BL mode is in every sense a TTL metered system. Obviously it computes flash exposure based on metering the preflash reflected from the scene. Bounce flash never uses the D-lens distance. However, in some cases of direct flash, yes, hot shoe TTL BL automation can watch distance and sometimes does reduce the flash power level, trying to prevent the overexposure expected from double checks of high guide number calculations (caused by common dark backgrounds affecting the direct TTL BL metering). This feature is intended to be a safety check on an abnormal situation, but seeing that happen a lot could cause us to imagine it always uses distance. TTL BL obviously is adversely affected much of the time on some lenses due to their D-lens distance errors (direct flash only). But lenses vary, all do not report the same distance errors.
|Inverse Square Law|
The Focused Distance Reported
|24-70 mm f/2.8 FX lens (not VR)|
|70-200mm f/2.8 VR FX lens|
|24-120mm f/4 VR FX lens|
|14-24mm f/2.8 FX lens|
|60mm D macro & 105mm VR macro|
|16-85mm f/3.5-5.6 VR DX lens|
|12-24mm f/4 DX lens|
|18-70mm f/3.5-4.5 DX lens|
Problem Concept: Assume the camera meter correctly meters the TTL BL direct flash. However (only for Direct flash in TTL BL mode), a Guide Number check is also done with D-lens data that may for example incorrectly report only half of the correct distance. That would imply the metering of the flash must be 2 EV overexposed if for only half of the actual distance. So the system overrides, and reduces the flash power 2 EV, and our result is a pretty dark picture. This definitely sometimes happens. Guide Number is Real Good Stuff, but it needs accurate distance data. The lens distance number reported is not necessary or useful to the lens job. The problem is that if the metering system chooses to use it anyway, when it can in fact be randomly in serious error.
The Evidence: The D-lens distance check is great in theory, when it actually correctly works as a TTL BL direct flash automation. But the D-lens distance data is often not correct, which can cause problems for direct flash (approximating distance from zoom lens focus rotation is very difficult, not at all the same thing as actual subject distance). Prime lenses are better at it, but zooms are a crapshoot (a dice game of chance with a risky unpredictable random result). Zoom lens reported distance accuracy is very suspect, but would do no harm if ignored). This feature is sometimes good or bad (the photo results above are good. Below is the bad.)
The table at right below is my check of nine Nikon lens (of various prices). Carefully measured distances were set up, and then compared the D-lens focus distance actually reported in the Exif (at minimum and maximum zoom). (See a better Exif Tool that can show focus distance in the Exif). Generally, none of the D-lens distance numbers are very accurate. The numbers shaded in pink are problems for the TTL BL flash default.
For those pink incorrect lens distances reporting less than the metered exposure, the Guide Number plan attempting to detect and prevent flash overexposure can override and reduce flash exposure, but a too-short incorrect lens distance can really mess that up. For example, metering a common distant empty black background can cause metered overexposure of the subject, and this plan can work great to prevent it (if the metered exposure is brighter than the Guide Number would expect for the distance). But if the lens distance is simply inaccurately too short, this plan screws up the exposure, often resulting in a pretty dark picture. I gave grace to 98%, as reasonably accurate, and not causing much problem.
If green lens distance is reported greater than the metered exposure, flash overexposure is not expected, and the D lens distance is ignored. It may not be accurate, but ignoring greater distance does not cause a flash problem. Thankfully, such cases don't try to correct to increase the metered TTL-BL flash exposure.
The far right hand columns show the percent distance difference (with 100% meaning accurate distance is reported).
Some lens zooms have more pink than green, but the bottom line is the default TTL BL direct flash result is unpredictable, and often bad results. For example, the first 24-70 mm lens zoomed at 24 mm and focused at 4.6 meters reports 2.5 meters, and this correction of incorrect overexposure prediction in fact causes a couple of stops of underexposure error for TTL BL direct flash. The D-lens data is in steps, maybe a little crude (possibly a bit better up closer). The chart shows some surprises. Even on a tripod, the D lens distance values were not always repeatable each time in every case, sometimes could vary each time the shutter button was pressed (focus seek slightly different when lens refocuses). At 15 feet (4.57 m), the 16-85 at 16 mm zoom might say 1.41 or 2.82 m. The 12-24 mm at 12 mm might say 2.0 or 2.82 m.
It certainly can cause TTL BL direct flash problems when incorrect. If the D-lens distance number reported IS LESS THAN the actual measurement (the cases marked in pink in the table), it will reduce the actual metered direct flash power in TTL BL mode. If the D lens reports a distance greater than actual, the accuracy is not better, but the check is ignored and has no effect. For example, TTL BL direct flash with the 12-24 lens at 12 mm at 20 feet (6.1 meters). The chart shows the lens reports 2 meters at 6.1 meters. This is variable, sometimes reported distance says 2 meters, and next shutter button focus seek may report infinity. Focus was not an issue, the report was just choosing between two steps I assume. But the system using that report of 1/3 distance will cause about 3 EV of underexposure, while infinity is ignored and does not interfere. This extreme underexposure seems a serious issue, not a great plan.
Below are some quickie flash picture examples in the dark of night with the Nikon 16-85 mm lens, to show the evidence of the concept discussed. All are point&shoot on a tripod at same distance, all everything the same except for flash mode, which was as indicated. Focus is always on the hanging white card accurately measured at 19.0 feet. Nikon D300 and hot shoe SB-800 direct flash. Camera Manual mode, Matrix metering, ISO 400, f/8, 1/160 second. Cropped slightly, but no exposure compensation in any.
Pictures below are the D300 with 16-85 mm f/3.5-5.6 DX AF-S lens at 16 mm zoom and f/8 at 19 feet.
That seems a very dramatic picture of a real problem (again, only TTL BL direct flash is affected, but which is default flash mode). Only Nikon brand flashes. Other brands of flashes like Yongnuo are immune to this D-lens problem (on the same camera).
This is a Nikon 16-85mm AF-S zoom lens, at 16 mm, focused at 19 feet (5.8 m). The chart shows the lens reports 2.8 m there. The Inverse Square Law says 2.8 m to 5.8 m is -2.1 EV.
This is the same image file above in 3. but now adjusted +2.1 EV brighter in Photoshop ACR. It is about 0.1 EV brighter now, but Adobe Camera Raw is another possible factor, and that's pretty close.
It is a real problem which occurs very repeatedly. And also, the reported distance can also vary somewhat at each shutter button, meaning at each focus seek. The distance is reported in steps, not continuously.
Fortunately there are workarounds (below) if you know, but otherwise, if this was the shot you wanted, with this lens and with TTL BL, then it may not have a chance. This lens changes with zoom, from reporting distance too short to instead be too long, and possibly it might even average out better, but some shots don't have a chance. This is Not exactly a metering issue, and Not exactly a flash problem, but clearly seems a system design problem, using flawed data for a serious purpose. The D lens distance is a serious problem for TTL-BL (for TTL BL direct flash) when the D lens reports a wrong distance that is too short. The camera does the metering, and does it very well, but here in #3, an incorrect D-lens distance got in the way, meddling with the TTL BL metered result (which Only affects TTL BL direct flash, but TTL BL is the default mode). It is not a bad plan, but the hardware obviously cannot support the plan. With certain lenses at certain zooms and focus distances, it can be demonstrated to be very repeatable. Apparently no one else gets concerned about this, but my cameras do these wrong things which I sure wish they didn't.
What happens is this: The white card subject was measured to determine the precise distance (there is a yellow measuring tape from camera to the point on the ground directly under the white card, but it barely shows here). However, the camera zoom D lens can report a very different guess, often significantly less than the actual distance. That's a problem, so while the preflash was metered to correspond to the actual exposure (needing nearly maximum flash power here), the D-Lens data said "Whoa man - that meter reading is obviously too much flash for a much shorter subject distance". Which is not a bad plan if the lens distance were correct. But unfortunately, the TTL BL flash power is reduced to prevent the supposed overexposure (at least the way it imagined it, described above). Not a plus unless it is correct. Zoom lens distance reports are often not correct.
This 16-85 mm lens is not the only lens affected this way, and the problem is extremely repeatable, with this lens at 16 mm. But we simply do not know when this will happen, which lenses at which zoom and distance values. We just know the flash system is squirrelly sometimes, due to this.
Again: This topic is only about direct flash with TTL BL flash. Not about bounce, not about TTL mode. But the internal flash, the SB-700 and SB-400 and SB-500 only provide TTL BL mode. However, third party flash seem unaffected by this problem.
This problem seems BIG. This very easily repeatable. Why there is no uproar about incorrect Nikon D-lens distance screwing up TTL BL direct flash mode (the default, for example the internal flash) is a mystery to me. We try to work around it, but at Minimum, the camera ought to have a menu to NOT DO THAT, to be able to ignore D lens data in zooms.
4. TTL BL was still selected here, but now using Spot Metering to switch metering to TTL mode, which does not use D-lens distance data. The camera manuals mention the TTL switch, but don't mention this distance problem. TTL is Not balanced flash mode. White card tone is 239.
Make no mistake, the flash system never uses Spot Metering. Flash has its own metering system it always uses. However Spot Metering is a very big deal greatly affecting brighter ambient metering, outdoors or with high ISO indoors. Which is no issue here at night, but if in significantly brighter ambient, Spot Metering users must seriously learn to evaluate the selected spot color, as it can drastically affect the exposure of brighter ambient (only ambient, the flash system has its own metering system). Unless high ISO, indoor or night ambient is typically well underexposed when using flash settings.
So when the ambient is too dim to register much anyway (normal indoors situations), Spot metering won't matter to dim ambient anyway, it remains underexposed regardless (unless high ISO). The ambient being dim and not in play (needing flash) does then allow Spot metering to be an acceptable choice for flash indoors. But don't forget to reset it Off when you go back into bright ambient, when Spot Metering can have a big effect. Spot Metering mode may seem very counter-intuitive, but it works.
5. TTL BL again the same, but now simply using FV Lock, which does not force TTL mode, but it does suppress using the D lens distance data with TTL BL. White card tone is 230.
So FV Lock can be an easy work around for the D-lens zoom problem. FV Lock does NOT select TTL metering, but it does ignore the D-lens distance. Not all camera models have FV Lock, but the camera models with Commander have it.
Nikon manuals say of FV Lock: "This feature is used to lock flash output, allowing photographs to be recomposed without changing the flash level and ensuring that flash output is appropriate to the subject even when the subject is not positioned in the center of the frame."
But recomposing the picture is not a requirement, it can still be the same shot. FV Lock is not a fully automatic method (it is two steps, trigger FV Lock preflash first, and then trigger shutter button). It may seem counter-intuitive, but it works.
This D-lens distance issue is only a problem for TTL BL direct flash, but never for bounce flash nor for any actual TTL mode. However, TTL BL is the only choice for the Nikon internal flash, and the SB-700, the SB-300-400-500, and almost all third party flashes. Only the Nikon SB-600, SB-800, SB-900/910, and a few Metz models have a menu to choose TTL or TTL BL. But Nikon removed that menu choice from the newer flash models, and does not provide it in the camera menu.
To prevent these expected D lens distance errors from causing flash underexposure, we can use bounce, or FV Lock, or even Spot Metering to get out of TTL BL mode. Only the Nikon SB-910 flash is still in production with this TTL override. If in significantly brighter ambient, Spot Metering users must seriously learn to evaluate the selected spot. This is all very non-intuitive, we need a straight-forward camera menu to do this, since it is the camera that does this metering. This seems very fundamentally basic, yet is ignored. TTL mode is perhaps not always exactly right either (is affected by the reflection of colors from the subject), but at least it is more predictable and consistent, less unknown automation mucking around in it. But many flash choices provide only TTL BL, with no other choice. The system default is TTL BL now. Why this kludge that requires selecting Spot Metering?
Zooms and internal focusing can mess up the concept of X focus ring rotation degrees is Y distance. This causes elements to be shifted inside the lens, changing geometry dynamically. This really messes up the D-lens concept. I couldn't care less if the lens distance is inaccurate, since we don't need it to do photography with TTL. But the camera should not insist on using it anyway to spoil our flash pictures in default TTL BL mode. It seems a good plan if it worked, but the D lens distance is inaccurate and inadequate and unacceptable for a flash metering system to actually use it.
If the reported D-lens distance error is less than the actual distance, it limits the TTL BL direct flash power that was properly metered. If the reported D-lens distance error is greater than the actual distance, then it is simply ignored. The system only tries to protect against flash overexposure, not underexposure. My experience is that we often see significant problem at the wide end of zooms. The D-lens is intended to help, but it is implemented too poorly in many zoom lenses, instead we get random wrong results.
The D-lens distance data simply is wrong sometimes (especially in wide zoom lenses), and it is one cause of TTL BL direct flash underexposure problems. You can always compensate the result, or use TTL mode (TTL mode does not use D-lens data).
This D-lens distance can even vary a little between two shutter presses (focus-seeking at each shutter focus), especially on these worst cases. A 20 foot case (6.1 meters) at 16 mm zoom might indicate 1.4 to 2.8 meters on this 16-85 mm lens (but neither of which is the accurate 6.1 meters). It seems some people are not bothered by this, but I think it is a real big deal, adversely affecting TTL BL direct flash exposure. Depending on which lens (and which distance and which zoom), it can cause severe underexposure, shown here. The simple solution for Nikon is to just "don't do that".
The D-lens focus distance data only affects direct flash in TTL BL mode. It does not affect bounce or TTL mode. But many users use only direct flash, and TTL BL is the default mode, so they are going to see a lot of this. No wonder lots of people dislike using flash, I would too, if I had to live like this. Switching out of TTL BL mode, instead to be TTL mode, easily solves this (ignores D-lens data).
However, Nikon is not shipping many flashes today that allow that, the SB-910 is the only current model with the menu to override TTL BL mode. If no menu, switching to Spot Metering reverts back to TTL mode (but it also does Spot metering of the ambient, which if significantly bright, introduces new issues for novices). Or, FV Lock can improve TTL BL (but FV Lock is only on cameras with Commander). FV Lock also ignores TTL BL direct flash D-lens checks, and uses the metered value. For me, FV Lock can help TTL BL considerably (direct or bounce), but has less effect on real TTL mode. I use bounce whenever possible, and avoid TTL BL all I can (it is good in sunshine though). Depending on which lens, it is very bad.
The Nikon 16-85 and 12-24 are fine lenses optically, but their D-lens distance data is abysmally poor. Not only is the D-lens distance data wrong, but it varies, any two shutter presses may report a different distance, at least with wide zooms. The other lenses in the chart might not see so much issue (only a problem when the D-lens data is LESS THAN the actual distance). Reporting infinity distance is great, that means this feature turns off. These are all fantastic lenses, and the only real issue is that the TTL BL flash system design wants to look at inaccurate distance numbers. The D-lens distance data seems a real detriment for zoom lenses that cannot do it. For example, the GN mode is all about distance, but it does not use the D-lens data (how could it?) The system works just fine if not using garbage data. Nikon marketing could tout a new menu disabling D-lens data was to allow TTL BL mode off camera on a SC-28 hot shoe cord, instead of having to admit the zoom lens distance was poor enough to screw up flash exposure.
Here are some pictures with same setup as on next page. D300, 16-85 mm lens, 16 mm zoom, 7.5 feet (2.3m) actual measured distance, and direct flash. All that changes is flash mode. D-lens reports different distances at each focus try. White Balance is the only adjustment made. WCV = White Card Value from eye dropper.
Note that the third picture (TTL BL without FV Lock or Spot metering) is the only case here that D-lens distance info will have any effect. And that effect seems a poor result, at least with this lens choice. The Exif for FV Lock does still say TTL BL. The camera manual shows the FV Lock metering area which is much larger than Spot metering, but not quite Center Weighted. The D-lens distance info must be ignored by FV Lock, and TTL BL often improves significantly (see more). Try it, and decide for yourself.
The bottom line is that there are two exposure skills to develop for using TTL flash. The necessary one is to learn to use Flash Compensation, which is the only control you have over the automated metering. If you don't like it, simply fix it. If it is off a little or a lot, still simply fix it. The other skill is a helpful shortcut — it is to develop the ability of simply looking at the scene, and "seeing it" (and recognizing its dark or light subject, and its dark or white background, etc) to learn to recognize in advance how its dark or light areas are going to influence the metering, affecting the compensation you will need to do (point is, it can largely be done in advance now). Both are quite easy after slight experience (usually just about like last time in the similar situation, i.e., experience). You simply have to think about it. Then regardless if the automation dances around or not, you will know what to do to get the result you want.
Next is Comparing TTL BL with two lenses, more data in more detail about this D-lens/TTL BL direct flash situation (image intense).