add to that the AF cutting off with lenses over f5.6 (such as with a TC) when often the camera can still AF or at least make the attempt. I think I once heard of a fix involving blocking one of the contacts though I don't know where I saw it.

mtl337 wrote:
If the precision AF is activated by a switch, wouldn't it be nice if there was a hack that would 'trick' my 20D into activating the cross-type sensor for my f/5.6 or larger aperture lenses ....?

Why does it have to be f/2.8 lenses (besides the effort on Canon's part to force me to purchase $$$ glass)?

When that happens, though, the camera is using an erroneously shallow depth of focus figure for its calcuations. There would certainly be a slow-down of focusing at the very least because it will be trying to determine a higher degree of precision at a lower level of illumination that the system was designed for. If the lens starts out greatly defocused, it will probably do the in/out rack, fail to find the focus point because it's calculating too slowly, and give up.

oldsouth wrote:
Which brings us to one of the earlier questions. Which method is best to focus. Center point and recompose, all point, or select the closest point manually to what we want in focus without recomposing.

If the center point is the most accurate with 2.8 lenses, looks like that would be preferred. But then you get into the angle error like was previously explained.


WOW! Thanks for all the excellent contributions to this incredible thread.
Anyhow, as to the best mode, if the 20D had a split-prism I might be all for manual focus, but I'm just not convinced it is worth going with the third party screen either.
However, having read this thread and experimented with what the sensors respond to horizontally or vertically, I am much more proficient with the individual focus point selection now.

First of all many thanks for the article very illuminating. I wish to report my experience, too.

I shoot in TV shows, so low light conditions, fast actors movements, very tight timings for focusing.
From this you need to work at least at 800 ISO, autofocus (no time to manual focus at all), focusing button separated from shooting button (I don't remember wich custom function is) and center focusing point manually selected, no time to let the camera decide what to focus on.

I started last season with my 17-85 IS f/4-5.6 and I achived average results due to the lens too slow. Thanks to some lucky strikes and the IS I got some good pictures.
Anyway the focusing precision of this lens was good: 80% of picture had the focus where I wanted.

This summer I decided to invest some money and buy the 70-200 f/2.8 L IS to have a faster lens, pro picture quality and sharpness.
I bought it one month ago and I'm very disappointed with it!!!

I've made a lot of shots of friends of mine to test the lens and I must say that its precision on focusing is far from being satisfactory! The proportion is the opposite now: the 80% of shots are out of focus because the camera focuses 5 to 15 cm beyond or before the point I aim.
The condition of the test shots are much more easy than those of studios!!! I've tryed it outside, in daylight conditions, and even with sitting subjects! Of course it's always at f/2.8, otherwise there is no point in such a lens!
Having seen this, I tried some tests with the 50 f/1.8 and I got the same results (I didn't noticed this before with it because I never shoot that wide open).

Now I understand why! It seems that the "high precision" sensor that turns on when you put on the 2.8 lens is a lot less precise than the normal one!

Now I'm desperate! I'm starting to work for the new season this week and don't know how to solve this problem!
Any suggestion? Many thanks.

I would suggest giving the lens and camera to Canon to take a look at it. Most people find the 70-200 2.8L IS very satisfactory, as do I. I had to send the lens and 20D to Canon for focus adjust when brand new. They did a great job. BTW they went to Irvine.

howgus wrote:
I would suggest giving the lens and camera to Canon to take a look at it. Most people find the 70-200 2.8L IS very satisfactory, as do I. I had to send the lens and 20D to Canon for focus adjust when brand new. They did a great job. BTW they went to Irvine.

So they solved a similar problem? mmmh , I think I'll do the same... I live in a big city, so I think canon assistance is close at home. I'll keep you informed...
Thanks!

i may have the same problem. I'm going to run some tripod tests, but I've noticed all my lenses not giving me the chrisp I want in certian areas of the picture. I will do tests to figure this one out.

adam

Any chance for fine focus at f4?

My understanding is that the 1-series currently has higher focus sensitivity at f2.8. Is it technically feasible that this feature could be extended to f4? IE for use with my 500f4.

AF is my single biggest problem with my 20D. I am planning to upgrade to the 1D-mk2 replacement when that is available, and primarily for better AF. I am hoping it will have improved AF even by 1-series standards. In my situation, AF is the single most important upgrade.

I also hope that new camera remains 1.3 crop and has 12mp.

Update on my 70-200 f/2.8 IS: finally I got to give the lens to canon service to tune autofocus. I'm getting it back in the afternoon.
This week end big tests and monday I'll report you any changes...
Let's hope...!!!

Riccardo, any news on your lens?

Lens is back!
I haven't had much time to test it well, just took few shots, but it seems to me that now it's pretty much precise than before. Actually the fine tuning did something. As soon as possible I'll take and post some test shot.
Imho, anyone having such a problem should send the lens and the camera to the service. I payed (warranty is over) 25 euros for the job, not that much!

RDKirk wrote:
mtl337 wrote:
If the precision AF is activated by a switch, wouldn't it be nice if there was a hack that would 'trick' my 20D into activating the cross-type sensor for my f/5.6 or larger aperture lenses ....?

Why does it have to be f/2.8 lenses (besides the effort on Canon's part to force me to purchase $$$ glass)?

When that happens, though, the camera is using an erroneously shallow depth of focus figure for its calcuations. There would certainly be a slow-down of focusing at the very least because it will be trying to determine a higher degree of precision at a lower level of illumination that the system was designed for. If the lens starts out greatly defocused, it will probably do the in/out rack, fail to find the focus point because it's calculating too slowly, and give up.



OK, so lets say it's an f/4 lens instead of a 5.6 - would focus be significantly slower than it would be with the 2.8 lens the precision sensors were designed for? What about the case of, say, a 200 mm f/2.8 lens on a 1.4x TC? A Canon TC will report the combination as 280 mm f/4 and the camera will focus based on that full dof. A third party TC will report 200 mm f/2.8 but the camera will think there is 1 EV less light. Will it succesfully focus based on 1/3 dof of 200/2.8?

All this AF is new to me - I'm in the process of transitioning from FD to EOS on a budget, so I appreciate these opportunities to learn.

There was some question about why it needs to be f/2.8 for the higher precision sensor arrays. Keep in mind that the max aperture determines the actual amount of light hitting the sensor before the shutter fires and the aperture is closed down for the shot. So during the focus the need for 2.8 would be determined by the sensitivity of the sensor array. Presumably the higher precision sensor needs more light to make that more precise determination - perhaps in relation to sensor noise, for example.

I mean, failing that, it's a great way to move people to more expensive glass for better results but that can also result in people buying 3rd party glass to get there at the expense of losing sales in the slower glass. I would never that doubt companies make options in their best interest but there is also plausible reasons why it may be so just due to the electronics.

csavery wrote:
There was some question about why it needs to be f/2.8 for the higher precision sensor arrays. Keep in mind that the max aperture determines the actual amount of light hitting the sensor before the shutter fires and the aperture is closed down for the shot. So during the focus the need for 2.8 would be determined by the sensitivity of the sensor array. Presumably the higher precision sensor needs more light to make that more precise determination - perhaps in relation to sensor noise, for example.

I mean, failing that, it's a great way to move people to more expensive glass for better results but that can also result in people buying 3rd party glass to get there at the expense of losing sales in the slower glass. I would never that doubt companies make options in their best interest but there is also plausible reasons why it may be so just due to the electronics.


But an f/4 lens on a sunny day gets more light to the sensor than an f/2.8 lens does on a cloudy day - let alone indoors. So I don't think it's a matter of available light. I know that split-image circles in a manual focus screen go dark when the aperature is too small; this can be a problem with long telephotos. I'm wondering if it's an optical effect similar to that - the high resolution sensors with their longer baseline just can't see anything unless the lens is at least f/2.8. If it's just software it should be hackable.

WillWeb wrote:
But an f/4 lens on a sunny day gets more light to the sensor than an f/2.8 lens does on a cloudy day - let alone indoors. So I don't think it's a matter of available light.

You are correct. The autofocusing system measures phase differences. The phase difference available to the autofocus sensors is greater with faster lenses. It is strictly a matter of geometry, not of the availability of light or the sensitivity of the sensors.

Great thread thank you!

The 24 dollar question is if the 1 series fixes these problems? Obviously all cameras have tolerances, so the question is how much improvement can one expect?

I have a low keeper rate of birds in flight with 20Dand 500 f4. I have enough experience to know it is mostly AI servo tracking. The camera tends to grab the wrong thing, usually water, trees or clouds rather than the bird. I have used centr only and all points.

I shoot in demanding situations, usually too far away, often poor light etc. I have all the gear I need, tripods, Wimberley etc. I have lots of practice. (just trying to avoid going off topic here)

So would a 1D-mk2 or 1D-mk2n make a significant improvement in AI servo mode captures?

ben egbert wrote:
So would a 1D-mk2 or 1D-mk2n make a significant improvement in AI servo mode captures?


As for my own experience in shooting friends playing soccer with 20D and 1D MkII. MkII peforms much better in AI servo. It's much faster in any situation. The improvement will be more then significant.