Urgent bounce flash question 760D owners

tr573 wrote:

walkaround wrote:

Thank you for posting these shots, it's actually very helpful. You don't mention if you are using a modifier (soft box, sto-fen, etc) on the flash, and if you are in HSS mode or not. I also would've loved to have seen Evaluative metering on the E-TTL as opposed to Average... but from some other shots on this site posted years ago, I'm convinced that if you had used FEL before the gray card shot, it would've turned out fine.

I am not using a modifier.

I would also say that, in the prime lens shots, you are at a much wider aperture, and therefore the ambient light is at a very different value than in your 300mm shots. For those you also have the framing the same on the 6D and 70D, so the 70D was significantly further away from the subject. Light power falls off rapidly with distance. This may somewhat explain the difference between those two cameras in this test.

I am using a 430EXII , in a room with an 8 ft white drop ceiling, and the bookshelf was all of 10 feet away with the 70D shot. There is no way the flash does not have enough power to light that - yes, light falls off with distance, but the whole point of "automatic" flash is to compensate for that automatically. It did not do so, and lit the scene almost entirely in zone 2.

Do you want me to take an FEL version of that shot just to prove to you that the flash has enough power?

I didn't say your flash didn't have enough power. I'm sure it does. But what happens when you use automated flash is that it fires a 1/32 power preflash to meter the subject. That's what I suggested was impacted by the distance difference. How does it do when both cameras are the same distance?

Regardless, I don't think this has anything to do with the lenses, per se. More the aperture at the time of exposure. And in the case of bounce, distance information is not used anyway.

I see why you are expecting a different result, and only a Canon engineer can explain exactly what the E-TTL algorithm is doing here, but obviously when there is a low ambient light exposure setting on the camera, the flash metering is being fooled by the lack of a subject in your photo. There is no "foreground", and no change in any one area of the frame for the camera to determine the subject, so it may be unrealistic to expect a perfect exposure here - although it seems obvious to you that a gray card is an easy subject to get right, perhaps not for the flash algo under these conditions.

Yet the 6d managed to get this almost perfect, while the 70D fell flat on it's face and did not.

"Almost perfect"? Nah, not really. As you yourself said, it underexposed also.

But really, you are just the latest in a very long line of people complaining about how E-TTL works, and there is nothing about your photos to indicate that this is anything unique to the 70D.

You would think Evaluative metering would be smart enough to recognize a snow scene and not turn it all gray - but it doesn't. I still have to add +1 or +2 EC. Is that a "firmware bug"? Is Canon "refusing to acknowledge" the well-known Snow Scene Problem? I'm not being fatuous here, I'm suggesting you just need to adapt, learn when to use FEC and FEL, and when the metering will be fooled like this. It's the same as any other automated functions on the camera. They work great 80% of the time. The other 20% we have to tweak them.

There is no famous snow scene "problem" to be acknowledged - that is a well understood limitation of a meter that sees everything as "18% gray" - this was an 18% gray card scene. It's the EXACT THING that it's supposed to meter perfectly. And it failed miserably.

Edit: BTW, the ambient was about 5 stops under on the 2.8 exposures, and thus 7 under on the 5.6 ones. Not enough ambient light to affect either really.

The ambient light EV has a direct impact on the flash power used, and the calculations when using E-TTL. It's not irrelevant.

Anyway, we can go around and around on this, which is interesting, but regardless no one has demonstrated that this is only a problem on the 70D. To say otherwise is to willfully ignore page after page of past history.