This is Part 2 of the preceding TTL BL page.
TTL BL exposure is influenced by four factors, other than any normal metering of the scene:
In contrast, TTL mode is only affected by #4, i.e., TTL mode does NOT watch the others automatically (and someone has to watch 1-3, but it helps to be there, and see it). The others (1-3) are considered helpful automation for TTL BL mode (point&shoot), but sometimes cause their own problems.
This page looks at #3, D-lens distance. Sometimes D-lens data reports shorter than the actual distance.
The TTL BL D-lens algorithm is that when the metered flash attempts to use more power than Guide Number says is appropriate for the distance (maybe TTL BL metered a distant dark background), TTL BL pulls the flash power back to match the D-lens reported distance (a safety factor preventing overexposure, computed with Guide Number). Problem is, D-lens distance is crude, and some lenses are simply not correct, and none are precise. TTL BL exposure depends on metered value, but if the Reported lens distance is less than actual distance, then flash level will be pulled back (TTL BL direct flash only). If Reported distance is Infinity, or any value greater than actual distance, then the undisturbed metered value is used. I am making this next part up, but FWIW, my notion is thinking there may be a limit to the maximum correction allowed due to D-lens. Some of the results are bad, but not quite as bad as the numbers would suggest.
The SB-400, SB-700, the internal flash, and Commander remote flashes do only TTL BL by system default (when their menu selects the generic TTL (Through The Lens metering), it is TTL BL. The Exif reports TTL BL mode (in the Exif Flash Data section, at the end — but Adobe Saves omit that. Check your original file with another Exif viewer.) Spot metering does override any TTL BL to be TTL (but it also does Spot metering too).
Flash GN mode is shown as a comparison, because it is usually about accurate exposure for direct flash, when the accurate actual measured distance is given to GN mode. My notion is that GN mode is what it probably ought to look like. GN mode and TTL mode are not affected by D-lens distance, but the D lens reported value is still shown here, to indicate that sometimes it is variable at each shutter press (focus seeking I think). The camera focuses by rotating the lens for the best appearance in the pixels.
Reports on two lenses are here. Both are prized as great lenses, but the Nikon 16-85 mm is no prize regarding D-lens distance. Nor is the Nikon 12-24, but the Nikon 14-24 mm is, because it generally just ignores it and reports infinity. One point is to show the obvious difference (below) that it makes to TTL BL direct flash (same camera, same flash, same subject, same situation). It was carefully done, however there are variances, doing any of it twice may see two distance results (focus seeking). Also there are few unexplained variations, so not every sample fits the plan, but most do.
Situation for all the pictures: Nikon D300 (DX), f/5.6 ISO 200, 1/200 second shutter, matrix metering. SB-800, maximum range is 25 feet at ISO 200 f/5.6. Hot shoe direct flash, focus on white card edge (on the edge, half of the metered area is not white card). 18% gray card also included. The actual distances were carefully measured. The only changes made are as reported, to distance, zoom, flash mode. White Balance is the only correction/compensation done. WCV just means White Card Value from eye dropper.
The D lens distance is NOT used for GN or TTL flash mode, but can override TTL info for TTL BL direct flash mode.
This lens is the best case I have seen, of non-intrusive D lens info (less harm done to TTL BL direct flash).
Clearly better D-lens data (#3 at top of page) gives TTL BL better direct flash results. There is still some built-in bias against TTL BL (#1 & 2 at top of page), but crude D-lens data greatly exaggerates the problems.
Bottom line though, it really doesn't matter much. TTL has other issues too (light reflected from the subject's colors, #4 at top of page), so in all cases, we just watch results and use Flash Compensation as necessary. Just do what you see is necessary. Knowing why helps predict results, but does not change the need.
Then two old non-D lenses. Frankly, my theory was expecting excellent results, and I was greatly surprised to see maybe the worst case yet. These old lenses provide no D-lens data. PhotoME says the 50 mm lens "reports" every distance as 0.01 meters, and the 55 mm lens reports 0.06 meters (less than 1/4 inch). These are impossible numbers, and you'd think this would be known and planned, and expect the system to ignore them, but there is some strong effect. Sorry, no explanation is known, the obvious is unthinkable. Again, same camera, same metering system, same flash, same subject, etc. Only the lens changed.
This lens is from 3 to 5 years before the D-lens era, it does not report it. TTL BL has trouble with this lens.
This manual focus lens is from 25 years before the D-lens era, it does not report it. It is a non-CPU lens, setup in D300 menu. It has been modified to AI mount, to report aperture. TTL BL has trouble with this lens, and TTL is down some too.
My own notion is that we seriously need a menu option to disable effect of D-lens data on TTL BL direct flash.
TTL mode was such an option, but it is disappearing now.
Sometimes just changing to a lens with more usable D-lens data can make a big difference (see the chart on Previous page, of a few lenses, for example).
Bounce flash is far better lighting anyway (when possible), and bypasses D-lens issues.
If TTL is a choice (SB-600, SB-800, SB-900), it is good stuff. Not always right either (because it does meter preflash reflections from the scene), but is consistent, predictable, no unknown programmers messing around mysteriously. Easily predicted and compensated. We do have to know to watch #1-4 at top of this page.
If TTL BL is the only choice, we can compensate results. Or Spot Metering does override TTL BL flash mode to instead be TTL mode, which ignores D-lens distance. However, it is still Spot metering, not simple point&shoot, instead requiring some thought about the spot selected. Spot metering is not general purpose, it does NOT meter the scene, it meters only the spot, and it is an advanced technique requiring more thought.
On cameras with commander, then FV Lock is a feature (for other purposes, to minimize subject blinking for example). But FV Lock can improve the worst TTL BL issues (sometimes giving different and substantially better results), and so also might do a lot to minimize this inaccurate D-lens data problem.
But the bottom line in every case is always Flash Compensation. If you don't like the results, simply fix it. It's pretty easy.