Hello...
Just to throw some more oil unto the fire, I've made my own set of tests...
I wanted to see if the lens aperture had an effect. After all, metering is done with the lens wide open, and that would also explain the different behaviors we see depending on the user : each of us has its own set of lenses so they surely have an impact...
OK, so here is what I've done : I've used only one lens (a PK-m Porst 135/1.8), but with the aluminum foil and halfway mount tricks to fool the camera so it thinks it always has a PK-A f/1.2 lens used at f/1.2, whereas the actual aperture varies (so I can actually choose what is the "wide open" aperture).
These tests were made using a EE-S screen.
First, straight flash: results are perfect from f/1.8 down to f/16 (with a slight variation, but you really have to look for it, I'd say maybe half a stop)... So, using an EE-S and any lens, you'll always get the same direct flash exposure (I'd suspect that using the stock screen would result in a -1 Ev variation with fast lenses, just like with stop-down metering).
Now, with bounce flash, results are really strange...
First, in my test conditions, between f/1.8 and f/2.8, exposure is good, with no visible variation...
Then, at f/4, a major overexposure occurs, followed by a seemingly constant flash power from f/4 down to f/16 (each pic being gradually darker than the previous one thanks to the decreasing aperture)... What is crazy is that the camera has no way to know that it reached f/4!!! So why the sudden overexposure? We already determined from the straight flash test that the focus screen gives accurate readings on the whole aperture range!
I'll go on with my tests, but it seems to me that in bounce mode, below a certain pre-flash metering reading, the camera loses its mind and fires the flash at a fixed output...
That would explain why some persons experience the problem, and others do not... For instance, I often use fast lenses only (between f/1.4 and f/2.8) at moderate distances, and so the pre-flash metering certainly stays above this threshold, and P-TTL works for me... But take someone using slower lenses, on longer distances, right at the edge of this threshold value... One moment the picture would be fine, then it would overexpose (or underexpose) for no reason, the flash being fired at this fixed level...
Now, the only time where I had trouble with bounced P-TTL and had to switch to manual was while snapping candid pics from afar in a dark venue, bouncing against the ceiling. Thinking back to this moment, I recall that I was baffled by the P-TTL behavior: it felt like it was indeed firing at a fixed power, as changing the iso was having a direct effect on flash exposure, even while I was far from the flash' maximum power and flash exposure compensation was useless (and switching to manual flash solved the problem anyway, so P-TTL should have handled it).
|