5D mark II versus 7D spot focus

[awinphoto]

I'm all for the discovery of the truth, so lets drop the intensity of light debate (even though, all things being equal, the bigger mirror, the more light, even if the center portion that would be collected by a crop sensor may or may not be equal intensities, the extra surrounding information may or may not be enough to affect exposure and light), but lets assume that's all equal... Let's take a 40D and a 5D mark 2 (same AF system)... Assuming the AF points are pretty much in equal or equal like proportion to the frame on the 40D and 5d mark II or at least the individual AF sensor size within the frame, and since the 5D is 1.6x bigger, then that would possibly mean, at least to the layman, that assuming the AF sensor isn't necessarily bigger, but if there were lets say 20 pixels of information per each sensor size on a crop sensor, on a full frame, there would be 36 pixels, hence more information going to the sensor, which allows it to be better in low light, which is the original problem that was in question. Any debates about this thinking?

 

[Loswr]

The focus sensor isn't smaller but the light allowed through to the mirror to the AF sensor is proportional by the diminished light.... Canon isn't going to waste money putting a full frame size mirror into a crop body when it doesn't need it. Plus by doing so, you would see image you wouldn't be getting anyways... When you read reviews about the 5D's viewfinder, you read adjectives such as Big, bright, pictureframe... when you do the math, it makes perfect sense...

Sorry, but that's just not correct. Yes, the reflex mirror is larger in FF camera than in an APS-C camera, which menas a bigger, brighter viewfinder, but that's completely irrelevant for the AF system. First off, the reflex mirror reflects most of the light up to the viewfinder, and allows a small amount of light to pass straight through the mirror to be reflected downward by the secondary mirror. Importantly, the secondary mirror is much smaller than the main mirror, and only a small portion in the center of the main mirror is semitransparent, to allow light through for the AF system.

The key point is that for a lens with a given max aperture, the light per unit area is the same, regardless of image sensor size. The light hitting any one point on the sensor (image sensor or AF sensor) is the same. The FF sensor gathers more light because the total area of the sensor is larger, and likewise the VF is brighter because the total area of the VF is bigger. I don't know the exact measurements in real units for the dimensions of a single AF point, but for the 5DII's AF points to receive more light, the individual points would need to be larger than their 7D counterparts. I can almost guarantee that's not the case for the center AF point, since the f/2.8-sensitive 'X'-shaped (diagonal) cross sensor on the 7D sits outside the standard '+' shaped f/5.6-sensitive center AF point, meaning it is almost certainly physically larger than the 5DII's center AF point, and with a larger area and the same light per unit area, that the 7D's center AF point is getting more light than the 5DII's center AF point.

When I bounce a laser pointer off a mirror onto a wall, the brightness of the laser dot on the wall doesn't change depending on the size of the mirror I bounce it off of.

Exactly - light per unit area is the same.

And I wont argue this matter any futher. any single point in a F5.6 exposure (with the same shutter) would be letting in 2x the light in every part of the frame than a F8 exposure... Same with the mirror.

Well, I'll argue further. The above statement is correct, but irrelevant. An f/5.6 exposure lets in 2x the total light than f/8 over the entire image circle, true. But that's true whether there's a FF sensor sitting in that image circle, or an APS-C sensor sitting in that image circle, or the even smaller AF sensor sitting in that image circle. The smaller the sensor, the less of that image circle is sampled by the sensor. Same light coming through the lens, less light detected by a smaller sensor.

You seem to be implying that the sensor size affects exposure - it does not. If you meter a scene with a given aperture, you should get the same shutter speed for both APS-C and FF (and for MF and a tiny digicam, too). Sensor size does not affect exposure, because exposure is determined by the light per unit area, which is determined by the f/number and independent of sensor size.

Hopefully, the above clearly demonstrates that the amount of light hitting both the 5DII and 7D AF points is the same. Fine, but that's not the only factor. Different AF sensors and different points within those sensors can have different absolute sensitivities. Even two AF points with the same rated sensitivity (e.g. f/5.6-sensitive horizontal lines) can have different absolute sensitivities in terms of the amount of light required to achieve a focus lock. So, even if the same amount of light is hitting the AF point, one AF point may simply be more sensitive (e.g. applies a higher internal gain while maintaining adequate S/N to achieve a lock, much in the same way that newer image sensors can achieve higher ISO values with equivalent noise).



But enough theory. How about the real world?

Overall, when comparing the 5DII AF with the 7D AF, I would rank them as follows: 5DII center AF point > 7D center AF point > 7D off-center AF points >> 5DII off-center AF points. More specifically, the 5DII's center AF point outperfoms the 7D's AF points slightly in terms of accuracy and noticeably in terms low-light sensitivity. The 7D's center AF point is quite good, and the off-center AF points of the 7D are nearly as good as the center AF point. Comparitively, the off-center AF points of the 5DII suck, and I only use them in bright light, preferably with contrasty subjects. For AI Servo tracking, the 7D wins, hands-down.

...for focus just about always use spot towards the center of the frame and recompose.

That is fine sometimes, but if you start using fast lenses (f/2 and faster) shot wide open, that technique can lead to your subject being out of focus. See this linked article for more details on the problem with focus-recompose.

Anyway I guess my question in general is from anyone that's used both would spot AF be noticeably different on either body?

I don't think anyone has specifically addressed this issue. If by 'spot AF' you mean that mode on the 7D, there is no such mode on the 5DII. With Spot AF, the AF system reduces the effective size of the AF points. Most people aren't aware that the actual AF point is significantly larger than the little box that represents the AF point in the viewfinder. That's one thing that leads to complaints about AF - you place the box right on the small feature you want to focus on, but the actual AF point is larger, and if there's a high-contrast feature that's just outside of your selected AF point 'box' the camera will lock onto that instead of what you put the box over. The 7D's Spot AF restricts the region of the AF point used to an area approximately equal to the size of the box in the viewfinder (note: not the tiny inner box that indicates Spot AF, but rather the larger, main AF point box).

All my lenses are f/2.8 or faster, but sometimes I do take some photos in low-light conditions without AF assist where the 7D does hunt a while.

Back to my statement above. The center AF point of the 5DII does better in low light than the center AF point of the 7D. With the center point, my 5DII will get an AF lock in lighting conditions so dim that the 7D would hunt and give up. With a Speedlite mounted, the AF assist comes on sooner and more often with the 7D than with the 5DII.

I presume you avoid the AF assist on the 7D because the strobing of the popup flash is annoying - I know that's the case for me. You might consider using a Speedlite - you can set it so the flash doesn't fire, only the AF assist lamp does, and the AF assist is a much less offensive red grid (just be sure to get one with the dedicated lamp, like the 430EX II - lower flashes, e.g. 270EX II, just strobe the main tube like your popup flash).

Hope that helps...

 

[Meh]

I'm all for the discovery of the truth, so lets drop the intensity of light debate (even though, all things being equal, the bigger mirror, the more light, even if the center portion that would be collected by a crop sensor may or may not be equal intensities, the extra surrounding information may or may not be enough to affect exposure and light), but lets assume that's all equal... Let's take a 40D and a 5D mark 2 (same AF system)... Assuming the AF points are pretty much in equal or equal like proportion to the frame on the 40D and 5d mark II or at least the individual AF sensor size within the frame, and since the 5D is 1.6x bigger, then that would possibly mean, at least to the layman, that assuming the AF sensor isn't necessarily bigger, but if there were lets say 20 pixels of information per each sensor size on a crop sensor, on a full frame, there would be 36 pixels, hence more information going to the sensor, which allows it to be better in low light, which is the original problem that was in question. Any debates about this thinking?

Yes. First, let's just clarify the difference in sensor size so there's no confusion for others. 1.6x refers to the diagonal measurement. The FF sensor is actually 2.5 times larger in area than the APS-C sensor.

You may realize this but it's not clear from your statements and others may not know. The AF sensor is completely separate from the image sensor (it's usually at the bottom of DSLR camera bodies) so therefore the relative size, resolution, pixel pitch, etc. between image sensors in various cameras has nothing to do with the AF sensors in all those cameras. The AF sensors do not need to scale (in terms of size or pixel size) along with the image sensors.

I'm not really sure exactly what you mean by "more information going to the sensor" but that's not really how phase detect AF works.

 

[awinphoto]

I'm all for the discovery of the truth, so lets drop the intensity of light debate (even though, all things being equal, the bigger mirror, the more light, even if the center portion that would be collected by a crop sensor may or may not be equal intensities, the extra surrounding information may or may not be enough to affect exposure and light), but lets assume that's all equal... Let's take a 40D and a 5D mark 2 (same AF system)... Assuming the AF points are pretty much in equal or equal like proportion to the frame on the 40D and 5d mark II or at least the individual AF sensor size within the frame, and since the 5D is 1.6x bigger, then that would possibly mean, at least to the layman, that assuming the AF sensor isn't necessarily bigger, but if there were lets say 20 pixels of information per each sensor size on a crop sensor, on a full frame, there would be 36 pixels, hence more information going to the sensor, which allows it to be better in low light, which is the original problem that was in question. Any debates about this thinking?

Yes. First, let's just clarify the difference in sensor size so there's no confusion for others. 1.6x refers to the diagonal measurement. The FF sensor is actually 2.5 times larger in area than the APS-C sensor.

You may realize this but it's not clear from your statements and others may not know. The AF sensor is completely separate from the image sensor (it's usually at the bottom of DSLR camera bodies) so therefore the relative size, resolution, pixel pitch, etc. between image sensors in various cameras has nothing to do with the AF sensors in all those cameras. The AF sensors do not need to scale (in terms of size or pixel size) along with the image sensors.

I'm not really sure exactly what you mean by "more information going to the sensor" but that's not really how phase detect AF works.

I should have used some other form of measurement, but I was referring to size of each AF sensor compared from a crop camera to a full frame So in relation IF a sensor in a crop camera covered an area compard to the same sensor in a full frame, all things being equal, would be covering 1.6x more information

 

[bchernicoff]

When you say: "....AF to start with shutter press" Do you mean : C.FnIII:Autofocus/Drive - AI Servo 1st/2nd image priority [2] set to option 3?

I am at work and don't have my camera handy, so I probably used some incorrect terms. I pulled up the owners manual and was referring to C.Fn IV -1 (pg 217 - 219 in the English owners manual). I have the shutter button set to "Metering Start" and the AF On button set to "Metering and AF Start". This means that half-pressing the shutter does not autofocus it only meters. You must now press AF On button to start focusing. Think of it as having the camera set to manual focus in terms of being able to focus separately from metering or taking the picture, but still getting the benefit of AF, when you press the AF On button. Again, I don't recommend this for moving subjects. For sports, I recommend what I said below:

And, I try to understand the "9 center points".....
I have set mine to: C.FnIII:Autofocus/Drive - Select Af Area selec. mode [6] set to Man. Selec.: AF point expansion

I was referring to Zone AF, with the central zone (9 points) selected, page 90 in the English owners manual. This really helps with AI-Servo tracking for me. I also have set my AI Servo tracking sensitivity to the most sensitive as I am usually shooting motorcycles which are quite fast moving.

 

[Meh]

I'm all for the discovery of the truth, so lets drop the intensity of light debate (even though, all things being equal, the bigger mirror, the more light, even if the center portion that would be collected by a crop sensor may or may not be equal intensities, the extra surrounding information may or may not be enough to affect exposure and light), but lets assume that's all equal... Let's take a 40D and a 5D mark 2 (same AF system)... Assuming the AF points are pretty much in equal or equal like proportion to the frame on the 40D and 5d mark II or at least the individual AF sensor size within the frame, and since the 5D is 1.6x bigger, then that would possibly mean, at least to the layman, that assuming the AF sensor isn't necessarily bigger, but if there were lets say 20 pixels of information per each sensor size on a crop sensor, on a full frame, there would be 36 pixels, hence more information going to the sensor, which allows it to be better in low light, which is the original problem that was in question. Any debates about this thinking?

Yes. First, let's just clarify the difference in sensor size so there's no confusion for others. 1.6x refers to the diagonal measurement. The FF sensor is actually 2.5 times larger in area than the APS-C sensor.

You may realize this but it's not clear from your statements and others may not know. The AF sensor is completely separate from the image sensor (it's usually at the bottom of DSLR camera bodies) so therefore the relative size, resolution, pixel pitch, etc. between image sensors in various cameras has nothing to do with the AF sensors in all those cameras. The AF sensors do not need to scale (in terms of size or pixel size) along with the image sensors.

I'm not really sure exactly what you mean by "more information going to the sensor" but that's not really how phase detect AF works.

I should have used some other form of measurement, but I was referring to size of each AF sensor compared from a crop camera to a full frame So in relation IF a sensor in a crop camera covered an area compard to the same sensor in a full frame, all things being equal, would be covering 1.6x more information

By "more information" I assume you mean more pixels over a larger area for each AF point but that would not improve the AF.

In simplified terms, phase detect AF is just looking for a peak along the strip of pixels that make up the sensor. It then compares that to the peak along the other strip of pixels (each AF point is actually two sensors) and looks to see if they "match", if not the AF system tells the lens to adjust to the correct focus position (phase detect AF determines this precisely, contrast AF does not which is why it is slower and cannot track focus).

There is no need to have more pixels covering a larger area because the peaks don't move very far between focused and unfocused.

 

[Loswr]

...I was referring to size of each AF sensor compared from a crop camera to a full frame So in relation IF a sensor in a crop camera covered an area compard to the same sensor in a full frame, all things being equal, would be covering 1.6x more information

Actually, in relative terms the AF sensor on the 5DII is smaller than the sensor on the 7D, i.e. the spread of the AF points on the 7D covers a larger proportion of the image frame compared to the 5DII. That's one complaint about the 5DII's AF system - there are no AF points anywhere near the 'rule-of-thirds' intersections, whereas there are on the 7D, and on FF bodies like the 1DsIII (sidebar: I love Canon's somewhat deceptive advertising on the 5DII's AF, where they state in the white paper that 9 AF points in the 5DII have the same horizontal spread as the 45 points in the 1DsIII - that's true, but the much-reduced vertical spread means the diagonal points don't reach the thirds intersections).

Ok, so the relative size is slightly larger on the 7D is slightly larger than the 5DII, but the absolute size of the 5DII's Af sensor is larger, so the 5DII's AF sensor gets more total light than the 7D's AF sensor. That's totally irrelevant. Unlike an image sensor, the AF sensor is not a full CMOS or CCD sensor that gathers light over it's entire area. The light is gathered by a small number of discrete regions - the AF points. Each AF point operates independently of the others, they don't 'add up' the light. So, unless there are data to support the idea that the individual AF points are substantially larger or smaller from one model to the next, there is no sensitivity difference based on the amount of light reaching the AF sensor (but, as stated above, electronic differences between sensors can result in different sensitivities).

 

[KyleSTL]

awinphoto,

For a given aperture (regardless of focal length, since aperture is a ratio of focal length and iris diameter) the light PER UNIT AREA is identical regardless of format (1/1.7", APS-C, 35mm, medium format, etc). Your statement of the 5D Mark II's AF sensor receiving more light is only true if (and only if) the individual focus points are larger than the 7D's focus points. The overall amount of light reflected by the mirror (which is dependent on mirror size, and therefore sensor size) is completely irrelevent to the light used for any given AF point.

The relevent question is (all other things being equal):

How big are individual AF points of the 5D Mark II vs. the 7D?

 

[TexPhoto]

Have both. 7D AF is better and faster. It's a sports oriented camera to me anyway.

And as other's have stated, crop factor has nothing to do with AF sensor size/sensitivity. I don't know if the actual sensors in the camera are bigger smaller, etc. but the size of the sensor would not effect the AF system.

 

[Loswr]

Have both. 7D AF is better and faster. It's a sports oriented camera to me anyway.

Agreed that the 7D is better, mostly. In good light, and for moving subjects, the 7D wins. Personally, I find that the center AF point of the 5DII is better in very low light. But that's probably appropriate - in light dim enough to give the 7D AF system problems, the ISO for the shot would probably be higher than I comfortably go on the 7D (unless I have a Speedlite attached, in which case, the AF assist lamp obviates the low-light AF problems).

Do you find that the 7D autofocuses well in low light without an AF assist?

One other observation - with some lenses, the 5DII seems achieve an AF lock faster, even in good light. When doing AF microadjustments (where I take several shots starting from MFD and several from infinity to a target at a fixed distance), I notice that with some lenses (e.g. 85mm f/1.2L II, 100mm f/2.8L Macro IS) the 7D tends to slightly overshoot, then go back to the correct focus, whereas the 5DII just seems to move to the correct focus and stop. It might be happening with all lenses - the two above have overall slower AF than lenses like the 70-200mm f/2.8L IS II or the 135mm f/2L, so I may just be missing that occurrence with the faster-focusing lenses.

How big are individual AF points of the 5D Mark II vs. the 7D?

To refine that a bit further, even an 'AF point' isn't a single sensor unit - each AF point is composed of multiple, discrete sensors. Here's what the center AF point of the 7D looks like (along with the smaller points to the right and left):

The light-sensitive area of the AF point is only a tiny fraction of the area we're calling an 'AF point'.

 

[PeterJ]

Thanks everyone for the feedback .

I don't think anyone has specifically addressed this issue. If by 'spot AF' you mean that mode on the 7D, there is no such mode on the 5DII. With Spot AF, the AF system reduces the effective size of the AF points. Most people aren't aware that the actual AF point is significantly larger than the little box that represents the AF point in the viewfinder. That's one thing that leads to complaints about AF - you place the box right on the small feature you want to focus on, but the actual AF point is larger, and if there's a high-contrast feature that's just outside of your selected AF point 'box' the camera will lock onto that instead of what you put the box over. The 7D's Spot AF restricts the region of the AF point used to an area approximately equal to the size of the box in the viewfinder (note: not the tiny inner box that indicates Spot AF, but rather the larger, main AF point box).

That I didn't realise and probably isn't something I'd enjoy, as the article you linked mentioned focus and recompose does have some things to be wary of but I've always preferred the 7D spot mode over single-point. Having said that I'm sure it's something I'd get used to and just a matter of knowing the limitations.

Back to my statement above. The center AF point of the 5DII does better in low light than the center AF point of the 7D. With the center point, my 5DII will get an AF lock in lighting conditions so dim that the 7D would hunt and give up. With a Speedlite mounted, the AF assist comes on sooner and more often with the 7D than with the 5DII.

I presume you avoid the AF assist on the 7D because the strobing of the popup flash is annoying - I know that's the case for me. You might consider using a Speedlite - you can set it so the flash doesn't fire, only the AF assist lamp does, and the AF assist is a much less offensive red grid (just be sure to get one with the dedicated lamp, like the 430EX II - lower flashes, e.g. 270EX II, just strobe the main tube like your popup flash).

Hope that helps...

I agree the popup is vastly annoying. I've got a 580EX II and do find the AF assist beam is good, but for some shots prefer them to be candid rather than that "oh look we're about to have out photo taken" look that you always end up with using AF assist. Anyway it sounds like for for those situations the 5D will win hands down, I was aware of all the IQ advantages in low light but didn't know how the AF held up there.

 

[PeterJ]

Just thought I'd post back and thank everyone for their advice, as of today I'm a proud 5D Mk II owner ;D. I haven't had a chance to take any especially interesting shots but gave it a test under varying conditions and I can't find much difference in the center focus point speed / low-light perfomance versus a 7D. As some commented the outer spread of focus points is not great (well pretty bad), and I tried to repeat a shot I'd taken recently through a bird cage using spot focus that it couldn't match, but I expected as much.

Now for the good part is that banging on a tripod and taking a few food shots the colour reproduction was fantastic. Outside during the day the FF gave my 70-200 a much better focal range and shallow depth of field for the style of photos I often take. Got some nice shots at night too using a 50/1.4, I expected a nice ISO boost but it seems to handle dramatic changes in contrast much better too. I'm a happy camper and look forward to giving it a good run over Christmas. It will really complement my 7D well rather than replace it .

About my biggest gripe so far is the location and type of power switch, I saw a post recently by someone asking about how to extend the standby time. I didn't reply but thought why can't you just turn it on & off between shots but now see where he was coming from. I put mine of a tripod earlier and just about needed a screwdriver to get at the switch :.

 

[Cyclops]

I just got my 5D a few weeks ago, and comparable the button layout (power switch) is much better on the 7D. Took me a long time to find the quick menu on the 5DIII, and takes a while to activate, and that menu is why i like my 7D.