eTTL oddity with bounced flash

I recently got a Tamron 70-200 f2.8 and was playing around with it last night on my T4i. I discovered that with direct flash (from my YN-658), eTTL provided a proper exposure but with bounced flash it was severely underexposed (manual mode, 1/100, f2.8, ISO 400). To check that there was enough flash power I set the flash to 1/4 power and bounced it, sure enough the exposure was blown out.

Then I tried the same experiment with my Sigma 17-50 f2.8, all the same settings. Every shot was properly exposed, direct flash and bounced flash.

Should I just chalk this up to shortcomings in Canon's eTTL algorithm?

No, IMO it's third party manufacturers' implementation of it.

If you do the same with a 'distance reporting' Canon lens and a Canon Ex Speedlite, the flash exposure will be very close (given no gel nor modification while in the straight ahead orientation). (Not the same, for reasons intrinsic to any reflected light metering tech.)

All three of my Canon speelites under expose when bounced.

Flashes in *TTL mode have no way to compensate accurately for light that is bounced as the dist is incorrect. You will find this info in about a zillion forums and how-to lighting books.

windpig wrote in post #17236166
All three of my Canon speelites under expose when bounced.

Same here. I usually dial in +2/3 EC on the flash.

Well, in theory flashes used in E-TTL mode has all the possibility there could be to set the flash power correctly, even if you bounce the flash. Since the light reflected from the target is measured by the camera prior to determining the proper flash power, it should make no difference at all.
But tests done on real targets have shown that sometimes it does anyway. Why I don't know.

groundloop wrote in post #17235427
I recently got a Tamron 70-200 f2.8 and was playing around with it last night on my T4i. I discovered that with direct flash (from my YN-658), eTTL provided a proper exposure but with bounced flash it was severely underexposed (manual mode, 1/100, f2.8, ISO 400). To check that there was enough flash power I set the flash to 1/4 power and bounced it, sure enough the exposure was blown out.

Then I tried the same experiment with my Sigma 17-50 f2.8, all the same settings. Every shot was properly exposed, direct flash and bounced flash.

Should I just chalk this up to shortcomings in Canon's eTTL algorithm?

Curious does the auto zoom change when the different lenses are used?

I've had generally good ETTL results bouncing my Canon 600EX's with no FEC applied. So somehow the flash power is being properly metered in spite of the distance reporting issue. No 3rd party lenses though.

The normal procedure for E-TTL is to send out a pre-flash immediately before the picture is taken. This pre-flash is for metering, and it does of course take into account if the flash is bounced, since the actual flash unit is used to generate the pre-flash. Thus the pre-flash travels the same distance and reflcets from the same surfaces as the real flash.

The distance report is disregarded by the camera when the flash is bounced, since the camera knows very well about the bouncing.

Didereaux wrote in post #17236176
Flashes in *TTL mode have no way to compensate accurately for light that is bounced as the dist is incorrect. You will find this info in about a zillion forums and how-to lighting books.

That's curious because this flies in the face of how eTTL II works - a pre-flash is fired, the intensity of the pre-flash is measured by the camera which then uses that information to calculate how much flash power is required.

I've thought about this some more and the only thing I can think of that would cause problems would be if the output of the flash isn't linear (if the camera asks for 50% power and the flash only delivers 35%, for instance).

groundloop wrote in post #17235427
I discovered that with direct flash (from my YN-658),

568 II?

i have no experience with it, but i think this is a known issue. It kept me from buying any of their flashes even though i mostly use manual mode.

see, "NOTE" and Comments:
http://flashhavoc.com/​yongnuo-yn-568ex-ii-canon/

http://www.dpreview.co​m/forums/thread/354140​9

Yep, it's a 568 II.

I just now tested with my YN-565 and there was a lot smaller difference between direct and bounced flash, so it's very possible it's unique to the 568 II.

In one of those links it was mentioned that the problem went away when eTTL metering mode was set to 'average' instead of 'evaluative', and indeed I found that to be the case. I suppose I can live with that. Most of the time when I use a speedlight it's off camera in manual mode anyway, it's just nice to be able to get exposure at least in the ballpark when I don't have the time to set flash levels.

hes gone wrote in post #17237756
=he's gone;17237756]568 II?

i have no experience with it, but i think this is a known issue. It kept me from buying any of their flashes even though i mostly use manual mode.

see, "NOTE" and Comments:
http://flashhavoc.com/​yongnuo-yn-568ex-ii-canon/

http://www.dpreview.co​m/forums/thread/354140​9

I've had time to do some more reading on this issue and must say that I'm pretty disappointed in Yongnuo. In general they've had some pretty good products at a good price, but when they ship products with a defect they're hurting their reputation when they don't take immediate steps to offer repairs.