Some unanswered Mk III questions

[Stephen Melvin]

Actually, it seems like these haven't been asked. Anybody own a Mk III, yet?

As someone who's owned a 7D and a 5D Mk II, there are some areas that really need improvement that haven't been mentioned in any of the marketing or brief "previews" from various sites.

1. The 5D Mk III inherits the 7D's iFCL metering sensor. Does it also inherit the low light flaw of the 7D, where the burst mode slows way way down in a dark environment?

2. Has shooting in Live View mode been made any more responsive? Current implementations are like first generation compact digital cameras from over a decade ago. I call it, "click and wait."

3. Has CDAF improved in Live View mode? I find I have to use "quick" AF, which drops down the mirror and engages the phase detect AF module, because CDAF is so slow as to be absolutely useless most of the time. Virtually all of the competition has been ahead of Canon on this. I hope to see a very significant improvement in the Mk III.

4. Any improvements to the ability to take a still in the middle of a video clip? A faster main processor should allow for a shorter gap in the clip.

5. How is the new camera removing moiré? Is there on-sensor binning? If it's binning, then why is the maximum ISO lower for video than for stills? In theory, it should be higher with binned pixels. Is it still line skipping? I haven't seen a word about this.

6. What's the buffer when shooting RAW and sending the files to both cards?

 

[dwischnewski]

According to the manual

4. No, there is a one-second gap
6. It slows down considerably. How much, it does not tell.

 

[PhilDrinkwater]

6. It slows down considerably. How much, it does not tell.

Are you sure?

I was on to Canon themselves and physically heard them shooting a 5diii at their offices and it seemed to be OK. The guy also said that the camera would get back to shooting again quicker than the 5dii due to a faster processor.

This was when discussing writing to two cards...

 

[jrista]

1. The 5D Mk III inherits the 7D's iFCL metering sensor. Does it also inherit the low light flaw of the 7D, where the burst mode slows way way down in a dark environment?

I'm not sure thats just a 7D flaw. AF and Metering sensors have to work with very little light due to the way they work. They start out with half as much light as passed through the lens, since a half-silvered mirror splits that light so some of it passes to the metering sensor and through the viewfinder, and some of it is directed to the AF unit. The AF unit itself splits that remaining light multiple times...for each AF point. Each AF point requires its portion of light to be split again and projected onto two phase-sensing strips.

So the metering sensor is always working with half light, and the AF points are always working with a fraction of half the light. In a dark environment, both need more time to gather enough to do their job. The AF unit is likely to put more of a drag on frame rate than the metering sensor. In the case of the 1D X and 5D III, they do have that 5-point vertical strip of extra sensitive double cross-type points...one standard sensitivity and one diagonal double sensitivity. That should certainly help with AF speed in low light, however they are limited to f/2.8 lenses. If you are using a lens with a narrower maximum aperture, you won't gain the benefit of those extra sensitive points, and you'll still likely experience a drag on frame rate in low light.

 

[akiskev]

Another unanswered question:
What is the live view output resolution of 5d mkIII?

 

[Stephen Melvin]

1. The 5D Mk III inherits the 7D's iFCL metering sensor. Does it also inherit the low light flaw of the 7D, where the burst mode slows way way down in a dark environment?

I'm not sure thats just a 7D flaw. AF and Metering sensors have to work with very little light due to the way they work. They start out with half as much light as passed through the lens, since a half-silvered mirror splits that light so some of it passes to the metering sensor and through the viewfinder, and some of it is directed to the AF unit. The AF unit itself splits that remaining light multiple times...for each AF point. Each AF point requires its portion of light to be split again and projected onto two phase-sensing strips.

So the metering sensor is always working with half light, and the AF points are always working with a fraction of half the light. In a dark environment, both need more time to gather enough to do their job. The AF unit is likely to put more of a drag on frame rate than the metering sensor. In the case of the 1D X and 5D III, they do have that 5-point vertical strip of extra sensitive double cross-type points...one standard sensitivity and one diagonal double sensitivity. That should certainly help with AF speed in low light, however they are limited to f/2.8 lenses. If you are using a lens with a narrower maximum aperture, you won't gain the benefit of those extra sensitive points, and you'll still likely experience a drag on frame rate in low light.

You don't understand the nature of the 7D's flaw. I'm not referring to AF or metering, really. If you take a 7D and put it on "M" and the fastest drive mode. Set the lens to manual focus and set your shutter speed an aperture manually.

In other words, take the AF and metering completely out of the equation.

Point the camera in a bright environment and you'll get the full 7fps drive speed.

Point it in a dark environment and it slows way, way down, and there is absolutely no reason for it. None. Nada. Zip.

This is what is known as a bug, and I'd like to know if the Mk III does this, too.

 

[Stephen Melvin]

Another unanswered question:
What is the live view output resolution of 5d mkIII?

1080i with an external monitor.

 

[Loswr]

You don't understand the nature of the 7D's flaw. I'm not referring to AF or metering, really. If you take a 7D and put it on "M" and the fastest drive mode. Set the lens to manual focus and set your shutter speed an aperture manually.

In other words, take the AF and metering completely out of the equation.

Well, actually, you're not taking metering out of the equation - even in M mode, the camera is still going to meter (supports safety shift, etc.), and the low light level forces that process to slow down.

Try this with your 7D - set the C.Fn's so the shutter half-press triggers AE-Lock instead of metering start (and AF+metering start to AF-ON, if you like), then try a burst in low light. You should get the full 8 fps.

Given that the 5DIII uses the same 63-zone iFCL metering system of the 7D, I'd expect the behavior to be the same, but the workaround above should still apply.

 

[nitsujwalker]

Just as a note on the 7D "flaw," mine does not do that. I can put it on Manual and crank up the shutter speed and shoot it in complete darkness at the full burst rate. Not sure if it is only older 7Ds that do it or firmware? But either way, mine doesn't slow down in the dark. Though I'm not sure why I'd want the burst rate in the dark when it is not exposing properly.

 

[jrista]

1. The 5D Mk III inherits the 7D's iFCL metering sensor. Does it also inherit the low light flaw of the 7D, where the burst mode slows way way down in a dark environment?

I'm not sure thats just a 7D flaw. AF and Metering sensors have to work with very little light due to the way they work. They start out with half as much light as passed through the lens, since a half-silvered mirror splits that light so some of it passes to the metering sensor and through the viewfinder, and some of it is directed to the AF unit. The AF unit itself splits that remaining light multiple times...for each AF point. Each AF point requires its portion of light to be split again and projected onto two phase-sensing strips.

So the metering sensor is always working with half light, and the AF points are always working with a fraction of half the light. In a dark environment, both need more time to gather enough to do their job. The AF unit is likely to put more of a drag on frame rate than the metering sensor. In the case of the 1D X and 5D III, they do have that 5-point vertical strip of extra sensitive double cross-type points...one standard sensitivity and one diagonal double sensitivity. That should certainly help with AF speed in low light, however they are limited to f/2.8 lenses. If you are using a lens with a narrower maximum aperture, you won't gain the benefit of those extra sensitive points, and you'll still likely experience a drag on frame rate in low light.

You don't understand the nature of the 7D's flaw. I'm not referring to AF or metering, really. If you take a 7D and put it on "M" and the fastest drive mode. Set the lens to manual focus and set your shutter speed an aperture manually.

In other words, take the AF and metering completely out of the equation.

Point the camera in a bright environment and you'll get the full 7fps drive speed.

Point it in a dark environment and it slows way, way down, and there is absolutely no reason for it. None. Nada. Zip.

This is what is known as a bug, and I'd like to know if the Mk III does this, too.

Well, metering, at the very least, will always occur. Even in M mode, it is essential so the camera can handle nuanced changes in lighting and still expose according to the settings you choose. You can move AF start to a different button, and only activate AF when you want to...that might help improve burst rate to some degree.

 

[jrista]

Try this with your 7D - set the C.Fn's so the shutter half-press triggers AE-Lock instead of metering start (and AF+metering start to AF-ON, if you like), then try a burst in low light. You should get the full 8 fps.

Given that the 5DIII uses the same 63-zone iFCL metering system of the 7D, I'd expect the behavior to be the same, but the workaround above should still apply.

Does using AE lock instead of metering start actually adjust exposure properly? Or could that allow improper exposure if you lock exposure before metering occurs?

 

[Loswr]

Try this with your 7D - set the C.Fn's so the shutter half-press triggers AE-Lock instead of metering start (and AF+metering start to AF-ON, if you like), then try a burst in low light. You should get the full 8 fps.

Given that the 5DIII uses the same 63-zone iFCL metering system of the 7D, I'd expect the behavior to be the same, but the workaround above should still apply.

Does using AE lock instead of metering start actually adjust exposure properly? Or could that allow improper exposure if you lock exposure before metering occurs?

You'd want to meter first - it's just like (with default settings) pressing the AE-Lock button, you need to half-press the shutter first to start the metering, then you lock it while the metering is active. In the above case, you'd press AF-ON to start metering (and AF), release that button then press and hold the shutter (which will lock the exposure and keep it locked as long as the shutter is held down, as you'd get with default settings if holding down the AE-Lock button during a burst - and actually, that's another workaround for the slow fps issue).

 

[Stephen Melvin]

Well, metering, at the very least, will always occur. Even in M mode, it is essential so the camera can handle nuanced changes in lighting and still expose according to the settings you choose. You can move AF start to a different button, and only activate AF when you want to...that might help improve burst rate to some degree.

"...it is essential so the camera can handle nuanced changes in lighting and still expose according to the settings you choose."

Let's say my settings are ISO 3200, 1/125 at f/1.4. Those remain absolutely consistent, because I've set my exposure and don't need the meter anymore. No matter what the meter reads, my settings are ISO 3200, 1/125 at f/1.4. I'm in manual mode. The meter should have absolutely no bearing on the speed of operation of the camera. That it does is absurd.

 

[awinphoto]

Well, metering, at the very least, will always occur. Even in M mode, it is essential so the camera can handle nuanced changes in lighting and still expose according to the settings you choose. You can move AF start to a different button, and only activate AF when you want to...that might help improve burst rate to some degree.

"...it is essential so the camera can handle nuanced changes in lighting and still expose according to the settings you choose."

Let's say my settings are ISO 3200, 1/125 at f/1.4. Those remain absolutely consistent, because I've set my exposure and don't need the meter anymore. No matter what the meter reads, my settings are ISO 3200, 1/125 at f/1.4. I'm in manual mode. The meter should have absolutely no bearing on the speed of operation of the camera. That it does is absurd.

Cool... Cant say i've tried shooting in low light rapid fire with my 7d... I guess I've always been paranoid in low light making sure everything is good focus when shutter speeds dropped and using shallow DOF so I tend to shoot slower, but i'll give it a try and see how mine does. A new experiment to try out, woo hoo

 

[Loswr]

Let's say my settings are ISO 3200, 1/125 at f/1.4. Those remain absolutely consistent, because I've set my exposure and don't need the meter anymore. No matter what the meter reads, my settings are ISO 3200, 1/125 at f/1.4. I'm in manual mode. The meter should have absolutely no bearing on the speed of operation of the camera. That it does is absurd.

...and that's the issue, right there. Even in M mode, the meter provides a reading, as you state above. In the 7D manual, Canon acknowledges that the frame rate may drop in low light (bottom of p.93), and although they don't explicitly state it, that's due to the metering, which needs to integrate the light input over a longer time for a more accurate reading in low light levels. As I stated above, I'd expect the same issue with the 5DIII. Note that Canon appears to have taken a different approach with the 1D X - instead of integrating the same number of zones for a longer time, in low light the 1D X metering switches from it's ful 252 zones to only 35 zones, integrating the input spatially instead of temporally, which I suspect is intended to allow the high frame rate to be maintained even with metering before each frame.

The bottom line is, as the software guys like to say: "It's not a bug, it's a feature." Most people shooting in M mode would like the camera to meter. Most people would like the camera to meter even in low light levels. So, it meters before every shot, and that takes fractionally longer in low light. The camera doesn't know if you're going to shoot one shot, or mash down the shutter button for a long burst. It just knows it's supposed to meter before the shot, and not release the shutter until the metering is done...unless you tell it otherwise.

As suggested above, using continuous AE Lock (whether holding the AE Lock button during the burst, or reassigning the shutter half-press to AE Lock) will stop the camera from metering between frames of the burst, giving you the full frame rate.

 

[Stephen Melvin]

I guess we'll tack that one up to, "probably still broken."

What about my other questions?

1. The 5D Mk III inherits the 7D's iFCL metering sensor. Does it also inherit the low light flaw of the 7D, where the burst mode slows way way down in a dark environment?

-- probably still broken, but no real world confirmation.

2. Has shooting in Live View mode been made any more responsive? Current implementations are like first generation compact digital cameras from over a decade ago. I call it, "click and wait."

3. Has CDAF improved in Live View mode? I find I have to use "quick" AF, which drops down the mirror and engages the phase detect AF module, because CDAF is so slow as to be absolutely useless most of the time. Virtually all of the competition has been ahead of Canon on this. I hope to see a very significant improvement in the Mk III.

4. Any improvements to the ability to take a still in the middle of a video clip? A faster main processor should allow for a shorter gap in the clip.

-- according to the manual, still takes a full second.

5. How is the new camera removing moiré? Is there on-sensor binning? If it's binning, then why is the maximum ISO lower for video than for stills? In theory, it should be higher with binned pixels. Is it still line skipping? I haven't seen a word about this.

6. What's the buffer when shooting RAW and sending the files to both cards?

 

[jrista]

Well, metering, at the very least, will always occur. Even in M mode, it is essential so the camera can handle nuanced changes in lighting and still expose according to the settings you choose. You can move AF start to a different button, and only activate AF when you want to...that might help improve burst rate to some degree.

"...it is essential so the camera can handle nuanced changes in lighting and still expose according to the settings you choose."

Let's say my settings are ISO 3200, 1/125 at f/1.4. Those remain absolutely consistent, because I've set my exposure and don't need the meter anymore. No matter what the meter reads, my settings are ISO 3200, 1/125 at f/1.4. I'm in manual mode. The meter should have absolutely no bearing on the speed of operation of the camera. That it does is absurd.

Well, not really. Actual ISO settings, vs. selected ISO settings, are not always the same. The camera has to account for light transmission differences in each lens as they are not all the same, and different element types transmit light differently. Diaphragms don't always stop down "neatly", and the actual light that passes through them is not as entirely uniform as one would think given the "ideal" settings available in the camera. Most Canon lenses are chipped to provide the camera with statistics about exactly how they transmit light, and the meter is still used to make fine adjustments to the actual ISO setting that is used when you expose to account for small differences. Even taking all those extra nuances into account, an advanced camera like the 7D still can't always correct deficiencies in specific lens designs. Some lenses always under or over expose by around 1/3rd of a stop on some cameras, while on others they may expose exactly correct.

If you were to take 10 shots of the same scene, all 10 shots probably wouldn't come out exactly identical. Even if they appeared identical, if you were to do a scientific measurement of the ISO of the shot based on the RAW pixels, its doubtful it would always come out at exactly 3200 or even be the same every time. The camera is still doing a lot of work behind the scenes to help your exposures come out as "correctly" as possible.

 

[Stephen Melvin]

We're getting a bit off track here. Any concrete answers to my questions? So far, the only concrete answer is about taking stills during video recording, and it's a disappointing one, from a "things should be improved" point of view.

 

[Mt Spokane Photography]

Another unanswered question:
What is the live view output resolution of 5d mkIII?

Assuming you mean video, its 720p. Lots of complaints that it does not output full resolution.

 

[Stephen Melvin]

Another unanswered question:
What is the live view output resolution of 5d mkIII?

Assuming you mean video, its 720p. Lots of complaints that it does not output full resolution.

As noted earlier, it's 1080i.