5D3 and 24 1.4 II L Edit, and other Lenses, owners please read!

[bdunbar79]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

This was suggested last night when he frantically posted first, then deleted the whole thread, only to repost.

 

[alexturton]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

no changes in exposure on my copy.

This lens is known for having QC. Rife with AFMA and AF accuracy issues.

I had to take my first copy back as it had +15afma and was wildy unpredictable for focus.

My second copy is perfect and is now my favourite lens (over my 85l 1.2 and 70200)

 

[tpatana]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

Ok, confirmed.

You get the problem if you don't have all AF points selected, and if you select all AF, then the problem goes away.

So depending on AF/MF and AF method/selection, the evaluation formula based on the measurement data is changed. MF will make it back to same formula as with all AF points.

 

[Louis]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

Thank you for all your replies, its nice to see i'm not going mad, it may not be a big deal, but when i switched to MF yesterday on a shoot, because my model wasn't inline with any 1 of my points, I had to manual focus, and then all my pictures were overexposed when I looked back at the images, I did only think this issue was with my combo 5D3 and 24 1.4 etc, thats why I reposted, but now reading the posts I see its with other lenses also

 

[canikon]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

Hello everyone, fisrt time poster here.
The phenomenon described by the OP is related to the strong metering bias that the 5DmkIII has on the selected AF point in evaluative metering.
This is the first thing I noticed coming from a 40D, and is my main complain about 5DmkIII metering algorithm.
When I shoot at night in urban environment, I always switch from evaluative to mean to avoid any strange behaviour of the camera.
In AF mode the level of brightness of the portion of the scene that is caught in the selected AF point area determines the bias in the exposure value.
When "all points" is selected, or when MF selected, this doesn't happen because the camera has no clue about what AF point will be actually used to focus.
This happens on every lens, and is not a flaw in the body, but a choice of the designers of the evaluative metering algorithm that I personally don't like at all.
I'll provide evidence later.
Bye for now.

Update: see the 4 pics attached, made with 135 f/2 L
the AF point is on the light on ceiling, iso is 320 fixed, aperture f/2.0 fixed
pic 1 has AF + eval metering, resulting in 1/4000 sec (strong bias on the AF point, it results effectively in a spot metering on the AF point)
pic 2 has MF + eval metering, resulting in 1/160 sec (the camera has no clue about the AF point in use, and provides a "true" unbiased evaluative metering)
pic 3 has AF + average metering, resulting in 1/160 sec (no bias on mean metering)
pic 4 has MF + average metering, resulting in 1/160 sec
As you see, in the case AF + evaluative metering the camera attemped not to clip the area of the picture on the selected AF point, thus rendering every other area of the image absolutely meaningless.
The amount of shift in this case is almost 5 stops!
If I were the designer of this feature, I would let the user to activare or deactivate it, or at least provide a "cap" on the maximum amount of correction allowable, say max 2 stops of bias.
Having it permanently operating as it is, without reasonable cap, means that you have a spot metering on whatever AF point is selected, and this is pretty disturbing to me in certain circumstances.
What do you think about it?

 

[bdunbar79]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

Hello everyone, fisrt time poster here.
The phenomenon described by the OP is related to the strong metering bias that the 5DmkIII has on the selected AF point in evaluative metering.
This is the first thing I noticed coming from a 40D, and is my main complain about 5DmkIII metering algorithm.
When I shoot at night in urban environment, I always switch from evaluative to mean to avoid any strange behaviour of the camera.
In AF mode the level of brightness of the portion of the scene that is caught in the selected AF point area determines the bias in the exposure value.
When "all points" is selected, or when MF selected, this doesn't happen because the camera has no clue about what AF point will be actually used to focus.
This happens on every lens, and is not a flaw in the body, but a choice of the designers of the evaluative metering algorithm that I personally don't like at all.
I'll provide evidence later.
Bye for now.

I also believe that the 1Dx's algorithm is different too, than the 5D3, and I have plenty of evidence to at least suggest that as a possibility.

 

[East Wind Photography]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

And dont forget on some models light entering through the top LCD could be a problem.

Just to confirm - everything else is the same with the physical setup? Remember that it's not just what's in front of the camera that matters - light entering through the VF also affects metering so if you're in a different position behind the camera, that can make a difference. Try covering the VF during testing (a lens cap hung over the eyecup does the trick).

Since Live View metering is done using the image sensor rather than the metering sensor, might be good to see if the same phenomenon occurs in live view.

 

[luciolepri]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

I confirm what canikon wrote. At first I thought my camera was defective, but after some tests I understood what was going on: the 5D MKIII gives a very high exposure priority to the metering zone where the AF sets the focus. So, in hi contrast situations, exposure metering results can change a lot, depending on which AF point is in focus.
It would be nice a firmware upgrade to let the user calibrate this metering bias...

 

[tpatana]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

Hello everyone, fisrt time poster here.
The phenomenon described by the OP is related to the strong metering bias that the 5DmkIII has on the selected AF point in evaluative metering.
This is the first thing I noticed coming from a 40D, and is my main complain about 5DmkIII metering algorithm.
When I shoot at night in urban environment, I always switch from evaluative to mean to avoid any strange behaviour of the camera.
In AF mode the level of brightness of the portion of the scene that is caught in the selected AF point area determines the bias in the exposure value.
When "all points" is selected, or when MF selected, this doesn't happen because the camera has no clue about what AF point will be actually used to focus.
This happens on every lens, and is not a flaw in the body, but a choice of the designers of the evaluative metering algorithm that I personally don't like at all.
I'll provide evidence later.
Bye for now.

Update: see the 4 pics attached, made with 135 f/2 L
the AF point is on the light on ceiling, iso is 320 fixed, aperture f/2.0 fixed
pic 1 has AF + eval metering, resulting in 1/4000 sec (strong bias on the AF point, it results effectively in a spot metering on the AF point)
pic 2 has MF + eval metering, resulting in 1/160 sec (the camera has no clue about the AF point in use, and provides a "true" unbiased evaluative metering)
pic 3 has AF + mean metering, resulting in 1/160 sec (no bias on mean metering)
pic 4 has MF + mean metering, resulting in 1/160 sec
As you see, in the case AF + evaluative metering the camera attemped not to clip the area of the picture on the selected AF point, thus rendering every other area of the image absolutely meaningless.
The amount of shift in this case is almost 5 stops!
If I were the designer of this feature, I would let the user to activare or deactivate it, or at least provide a "cap" on the maximum amount of correction allowable, say max 2 stops of bias.
Having it permanently operating as it is, without reasonable cap, means that you have a spot metering on whatever AF point is selected, and this is pretty disturbing to me in certain circumstances.
What do you think about it?

Knowing that, you can also use it to your advantage. E.g. if you know you want bright sign to be exposured correctly, use focus points on that one, and also if you want ambient proper but focus, you can e.g. exposure lock somewhere else and re-compose.

So I don't see that as a disadvantage, after knowing how it works. Not sure if that was mentioned in the manual though.

Also for the OP, I wish I had that much confidence on my skills and gear that I don't chimp the histogram every few picture. If it took you after the shooting to notice over-exposure, maybe next time you should chimp some too?

 

[canikon]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

Knowing that, you can also use it to your advantage. E.g. if you know you want bright sign to be exposured correctly, use focus points on that one, and also if you want ambient proper but focus, you can e.g. exposure lock somewhere else and re-compose.

So I don't see that as a disadvantage, after knowing how it works. Not sure if that was mentioned in the manual though.

You are actually right, it is a nice feature when you are doing portraiture in harsh light conditions, provided that you don't AF focus on some dark hair or eyebrow... in that case the exposure is again pretty weird.
All in all, I would prefer at least a "cap" on the amount of shift, otherwise the evaluative metering is not evaluative at all and is more a sort of spot metering on AF point. That is not the way is is marketed by Canon, but it is the way it acts actually.
Moreover, if you are grabbing a shot quickly you don't have the time to AE lock on the scene and then AF and shoot. This is the reason why when shooting after sunset in urban environment I usually use mean metering, just to avoid that the light of some car or some street lamp could ruin an otherwise perfect shot.
I shoot other Canon and Nikon bodies, a 40D, a 600D, a D800, and I must admit that they have much better and more enjoyable evaluative metering algorithms, and I actually trust those cameras much more than my otherwise beloved 5DmkIII, that actually I trust not at all in high contrast situations.

 

[luciolepri]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

Knowing that, you can also use it to your advantage. E.g. if you know you want bright sign to be exposured correctly, use focus points on that one, and also if you want ambient proper but focus, you can e.g. exposure lock somewhere else and re-compose.

Well, if you have to do that, what's the point in a "63 Zone Dual-Layer Metering"? Just use the spot exposure!

 

[dr croubie]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

Theory!

Flipping the switch triggers metering. Since evaluative metering invokes code that looks at multiple metering points and tries to intelligently guess what to do, it's somewhat reasonable to assume that it might a different decision on what to do each time it's invoked if the light entering the lens in non-uniform across the image.

A way to test if this is happening is to point the camera so the image is completely uniform in brightness across the image — at the sky, or a flatly coloured wall, etc.

This makes perfect sense. Evaluative metering is linked to the selected AF point. Switching to MF deactivates all of the AF points, which means evaluative metering can no longer be weighted toward a selected point. The fact that you're seeing this only with a wide angle lens is not unreasonable. With such a wide FLV, there's more in the scene to affect the metering when it is no longer weighted toward the selected point.

Theory confirmed, on my 7D & EFs 15-85 as well. Standing outside, with single AF point on the top of the frame, focussing on some bright clouds, in the bottom half of the frame is my neighbour's relatively dark roof. On AF, I get 1/500s. Switch to MF, I get 1/320.

What Neuro said is also applicable to the 7D, metering is weighted more towards what is in focus (in my case, bright sky at infinity). If it doesn't know what's in focus (from being in MF), then it will give even weight to the whole frame and meter more for the dark roof. This is pretty much what the user manual states.

It's not a bug it's a feature.

 

[luciolepri]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

It's not a bug it's a feature.

Yeah, that's just how the exposimeter works with zone metering & AF mode selected, it was like that even with analogic cameras, the point here is that in the 5D Mark III the metering bias is too high.

 

[Louis]

Great response and i've learnt something thank you! I've never noticed this as i hardly ever shoot manual, anyone know why it seems to be more erratic on my 24 1.4 lens and not my 100mm, ? more testing needed etc

It seems then that the evaluative metering will start from the point of focus, but when switched to manual it will whole scene giving a brighter exposure

 

[Loswr]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

...in the 5D Mark III the metering bias is too high.

I wonder how all the people clamoring for the 5DIII to have AF-point linked spot metering would feel about that observation?

 

[Pi]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

What Neuro said is also applicable to the 7D, metering is weighted more towards what is in focus [...]

This is how Canon's evaluative metering worked since the 300D (at least). I noticed however that my 350D and my 50D were more sensitive to the AF point than my 5D2.

 

[Pi]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

I shoot other Canon and Nikon bodies, a 40D, a 600D, a D800, and I must admit that they have much better and more enjoyable evaluative metering algorithms, and I actually trust those cameras much more than my otherwise beloved 5DmkIII, that actually I trust not at all in high contrast situations.

What do you mean by "trust"? You are the one responsible for the exposure. You just need to understand the algorithm and use it appropriately. There is no perfect algorithm and the camera cannot know how you want to expose.

 

[LetTheRightLensIn]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

Theory!

Flipping the switch triggers metering. Since evaluative metering invokes code that looks at multiple metering points and tries to intelligently guess what to do, it's somewhat reasonable to assume that it might a different decision on what to do each time it's invoked if the light entering the lens in non-uniform across the image.

A way to test if this is happening is to point the camera so the image is completely uniform in brightness across the image — at the sky, or a flatly coloured wall, etc.

Please bear in mind that I don't own this lens or camera, but I do have experience in programming. In algorithms like this, it's often the case that there's some guesswork involved — hell, I once implemented an algorithm that would just randomly choose one of tho values if it couldn't determine one value over another with any degree of certainty, and it worked just fine.

Since I doubt the camera saves the reasoning behind its metering decisions between each metering, I can easily see this sort of thing happening in something that's supposed to be "smart" like evaluative metering. If you require absolutely consistent metering, I guess evaluative isn't the right mode to choose.

+1

 

[risc32]

Re: 5D3 and 24 1.4 II L owners please read!!!!!!!!!!!!

...in the 5D Mark III the metering bias is too high.

I wonder how all the people clamoring for the 5DIII to have AF-point linked spot metering would feel about that observation?

To everyone who wanted AF linked spot metering, surprise, you have it!

 

[Hobby Shooter]

It seems then that the evaluative metering will start from the point of focus, but when switched to manual it will whole scene giving a brighter exposure

Makes sense