5D3 & 600ex-rt + AF assist beam = slower focus

[Dave_NYC]

I have a 5Dmk3 that I sent in to the California service operation this past summer because I lost shots due to slow focus without flash (assignment: dress rehearsal under stage lighting. Focus delay with 70-200 f2.8 v2 even when using lit part of face or collar, etc.) Canon wrote that they could not recreate my probem.

Actually, just realized there would be a caveat for this particular situation (where you are not using flash which would overwhelm the red AF assist light from the 622) - you would have to half press to focus lock then take the shot, because until the camera has focus lock in low light, the 622 will put out the AF assist red light pattern, and if you do a full press only that red pattern will probably end up in your shots.

The good news is if you wanted to go that route, that is what I was testing - I was getting half press "beep" focus lock on just the AF assist the 622c puts out in typically a second or less. Just don't hit the button all the way straight off the bat or you might get a lot of jokes about incoming airstrikes...

[gilmorephoto]

I finally had a chance to compare both my bodies (5D3 and T1i) with two flashes (600EX and 430EX II).  I used the 40mm f2.8 for both (at first) with center point selected to keep things as consistent as possible. I shot into my walk-in closet, so it was dark enough to trigger the AF Assist light but with the door open so some ambient light would trickle in.  I turn the focus confirmation beep back on to have another indicator of when the camera thought it was focused and when it fired. I tried both camera with both flashes.  I did not time anything with a watch because if it's close enough not to be able to tell the is a significant difference, I really would not be bother which one is technically faster. That said, here are my observations:

5D3--regardless which flash used, the body had the marked hesitation that I mentioned earlier. This is not just a little shutter lag--this is a full second or two, the moment is gone, I didn't get anything and, if I did, it is not in focus.  This is beyond annoying, this is failure when mission critical.  There was no perceptibly meaningful difference between the flashes, so this lead me to believe it's the way the body acquires focus.  I noted the two or three step confirmation process someone noted earlier. This was the behavior no matter which focus point or AF method was selected (excluding SERVO).

T1i--press, shoot.  find another spot, press, shoot.  repeat, repeat.  All in focus, all virtually instantaneous and at worse considerably better than the best attempts with 5D3.  If anything, 600EX performed better than 430 EX II (thank goodness for that) on this body. This is what I was expecting from the 5D3 and it's not even close.  (I repeated the test with my 15-85mm zoom and the results were consistent regardless of the focal length and aperture setting).

I'm now even more convinced it's a serious with the 5D3 and not the flash.  There is no way my T1i (which I still love) could/should outperform the 5D3 under these circumstances, especially given the type of photography the 5D3 is geared toward.

I'll repost this on the Canon forum.

[dlleno]

nice test gilmorephoto. emperically, very strong evidence that the 5D3 AF is not very good at utilizing the AF assist beam at least in the factory 'out of box' configuration.   By they way --  for a reality check  EV -1 is enough light to see where you are walking... and within the 5D3's ability to focus by itself without the AF beam.  With flash on board, the camera should be able to utilize the AF asset beam in  total darkness. 

the fact that a modern camera takes several second to acheive focus  using the flash AF assist beam  is rather astonishing to me.    I have tested the lowly 40D in this situation with the 580 EX ii  (70-200 f/2.8 and 17-55 f/2., and find that this body utilizes the AF assist beam quite handily, and acceptably acheives AF in total darkenss.   But then, the 40D AF center point isn't as precise as that of the 5D3 -- it may "see" more of the AF assist beam!

is there a geometry problem here - I mean is the AF system so precise that the AF assist beam doesn't nail the exact location of the center AF point?   is the hot shoe manufacturing dimmentional tolerances such that one camera will aim the flash better than another?  Has anyone mounted the flash off camera or at various distances and experimented with this?

If geometery is unrelated to this issue, then perhaps the AF sampling rate of the 5D3 does not live well with the pulse rate of the AF assist beam

[echelonphoto]

Just did an evening engagement shoot at Longwood Gardens . Mrk3 really struggled with focus in low light...using
my af assist...600 rt and 70-200 f4 is lens...sometimes would not acquire focus at all...usually at a distance of about 20 ft. Didn't have my mark2 to compare...but this is really not what Iwould call improved AF. If I turned
off af assist...did not focus at all. this is using the center spot one shot af.

[dlleno]

echelonphoto:  can you be more specific:  what level of light were you shooting in (tell us the ISO ,shutter, and apertuture) -- did you capture any (even out of focus) photos where the EXIF data would reveal this information?

if you put your 5D2 side by side with the 5D3 perhaps you could capture some photos (without flash or AF assist) and we could know what levels of light we are talking about.

[emko]

has canon said anything about this yet?

[dlleno]

has canon said anything about this yet?

no.  In fact, we have one example, cited on this forum, where Canon repair indicated that they could not reproduce the problem. 

[teedidy]

Hi Everyone, First post so forgive me if some (or all) of this is redundant.

I have (and love) a 5D3 and few points I have not seen anyone make yet.

One: The assist beam only turns on when single shot af is used. AI focus does not trigger the beam. (DOF button reprogrammed to switch from AI to single shot af)

Secondly I have some older flashes 580ex (not II) and 430ex (not II). Both of these work extremely well as an assist beam for focusing. (Better then the yongnuo 622 since they adjust the pattern based on where the assist point is located). In complete darkness (bathroom, no lights, door shut) I achieve focus effectively fast using either of the 3 above assist beams. With both the 430ex and 580ex the size and shape of the pattern changes depending on what AF points I am using.

Thanks everyone. Keep up the good work!

-T

[dlleno]

wow teedidy that is extraordinarily different results than folks using 580 ex ii and 600 ex-rt.  so perhaps we ARE talking about a marriage problem between the AF system and the AF assist beam supplied by the newer flashes.   

[neuroanatomist]

One: The assist beam only turns on when single shot af is used. AI focus does not trigger the beam. (DOF button reprogrammed to switch from AI to single shot af)

It doesn't work in AI Servo, so I suppose it makes sense that it would not work in AI Focus, either (although I never used the latter, and now it's not even an option on my most-used camera).

[gilmorephoto]

I believe AI Focus DOES trigger the beam, but let me triple-check tonight. I recall switching through the three options (One Shot, AI Foucs, and Servo) to determine that Servo (and only Servo) did not generate beam but I could have been mistaken.

[digital paradise]

AI Focus does. AI Servo does not. 

[digital paradise]

Post #85

http://forums.usa.canon.com/t5/EOS/5D3-AF-assist-beam-slower-focus/td-p/2277/page/9

[RustyTheGeek]

That's good to see, Canon says they are listening, in the forums at least.  After months (or years) of testing and then 9 months on the market, I really wish they had a little more to offer.  But something's better than nothing I guess.

Thanks for the heads up!!

[jaayres20]

I sent in my 5D3 for a new shutter and I specifically complained about the AF assist beam causing slower focus.  They did not admit there is a problem and after I got my camera back I looked through what they had done and they mentioned that they made electrical adjustments on the AF.  I finally used it at a darkish event and at times I thought it was significantly better but after using it for two hours I came to the conclusion that it has the same old problem.  I went around and found about 50 different places where the camera would focus instantly without the AF assist on my 70-200.  I would then immediately switch the AF assist back and and try to take the same exact shot and noticed a significant lag (sometimes a second).  After it is dark enough where AF is not possible then the AF assist is a benefit to me even if it does take over a second or two to focus so I would just like to keep it on.  Instead I am turning it on and off several times during an event and it is a waste of time and not very efficient.