Reading mode: Light Dark

Does 20D really focus more precisely than Rebel XT? Is it a "consumer" body or "pro"?

Go to last post
thanks max i thing i've made my mind up now! i didnt notice the "eye/keyhole" was much smaller on 20d eventhough i've tried both.
Max@Home said:
Sis-Q said:
The XT seems like a lot of camera for the money in
comparison to the 20d . But of all the differences between the
two models, the one that means the most to me is AF accuracy. Does
somebody out there have both cameras
Click to expand...
I gave the 350D a testrun last March...
Max@Home said:
and have you noticed a real
world difference in AF accuracy or speed of focus between the two?
Click to expand...
Yes and Yes, 20D is more accurate and faster.
Max@Home said:
If you noticed a difference was it only on the fast glass?
Click to expand...
No, on 'slower glass' the difference in AF-speed is more noticeable,
on 'faster glass' the difference is accuracy is more noticeable,
IMHO...

What I do not read in this thread, IMHO the 'eye' 'keyhole' through
which you focus is much smaller on 20D, you can aim at the pupil in
stead of the eye, if you catch my drift:

350D:



20D:



...my €0.02 worth...

Kindest regards,

Max@Home
--
Max@Home - Castricum - The Netherlands (see profile for equipment)



http://www.pbase.com/max_at_home (mind you, use underscores!)
Click to expand...
 
Whether its because the 20D has more focus points or a better pattern I'm not sure, but tends to lock onto what I want more, the 350D is no slouch but I find myself refocusing more on it.

Where the 20D beats the 350D is in low light, at a recent studio shoot, I shot with the 20D and a friend used my 350D, we used the same 50mm f1.8 mkII, we shot a silhouette with very little ambient light, I had to select center focus and focus on an edge, but the 350D just refused to focus so resorted to manual.
 
pentool said:
ok more confusions... why 50mm 1.4 is better with xt? why would any
lens be better on the xt at all?
Click to expand...
Because 20d's high precision sensors have an issue with big apertures (worse with 1.2 and 1.4 lenses mentioned).

The evidence is clear to see at http://www.canon-dslr.com , the pictures showing two shots, in a row, with completely different focus from 20d with such lenses. Varience is far less with f1.8 or f4, and very good with f2.8.

The 10d, 300/350, d60 didn't suffer this difference across lenses of different apertures (f2.8 to f1.2).

Whether it's caused by a simple programming error by canon, or a fundamental design flaw of the hp sensor I do not know. I only know the problem is their, across a number of tested bodies and lenses, and that canon has yet to acknowledge the problem or a fix for it. It may have been fixed on the quiet with a flash upgrade, but even in the last 2 months, I have had further reports from top uk pros buying 20d's that have suffered same issues with the big aperture lenses mentioned.

Regards,
Kev
 
Doug,

Someone on Fred Miranda's site has provided this link:
http://cweb.canon.jp/camera/eosd/20d/catalog/index06.html

Someone with good skill in Japanese can probably give us more information.

Notice the color coding of the sensors and the legend.

At f5.6 (blue colored on the diagram), we see the two linear pixel sensors of each array (a pair of sensors make up an array to sense the phase difference for a line of contrast). So we see two sensors arranged horizontally (the array that senses vertical lines) and two sensors arranged vertically (the array that senses horizontal lines).

At f2.8 (both green and red on the diagram) we see the same two vertically aligned sensors active, but a different set of horiztonally aligned sensors have been activated. These are farther apart, making the baseline wider, I'd suspect to provide the greater precision Canon claims with f2.8 lenses.

This is different from the way I understood it--it's not an "additional" horizontal array, but a different array with higher precision. It does correspond to Rob Galbraith's description in his review of the 20D.

This diagram and the accompanying photo of the sensor IC do raise a bunch of other questions, though. I'm having a hard time believing that the 20D is utilizing all the sensors that seem to be on the IC.

--
RDKirk
'TANSTAAFL: The only unbreakable rule in photography.'
 
I don't think the manual is very clear at all, at least not the English version. I suspect something has been lost in either the translation from technical to lay or from Japanese to English

Notice that the description of the AF capability at f5.6 is simply lifted from the 10D manual, and states that it is a cross-type "sensor" (actually, at least four pixel sensors aligned in pairs horizontally and vertically).

It unfortunately describes the sensor at f5.6 in the same terms, calling it high-precision but utterly garbling the description of the positioning of the sensors that makes it high-precision.

I didn't understand the arrangement myself, until seeing this diagram:

http://cweb.canon.jp/camera/eosd/20d/catalog/index06.html

There is an addional pair of vertical-line-sensitive sensors, but they are set farther apart, providing a wider baseline. These are activated at f2.8 INSTEAD of the pair of vertical-line-sensitive sensors activated at f5.6. This is what Rob Galbraith describes.

--
RDKirk
'TANSTAAFL: The only unbreakable rule in photography.'
 
Karl Günter Wünsch said:
I think you mixed up sensitivity and precision, which are two
different parameters of the focusing sensor. Sensitivity determines
how strong the contrast has to be to be detectable while precision
determines the amount the focus might be off the ideal position.And
the manual clearly states that the vertical sensor for the f/2.8 AF
sensor is twice as sensitive as the horizontal.
Click to expand...
I don't think the information we have so far allows us to say what the difference between "precision" and "sensitivity" is by the uses of those terms in the Canon literature--I don't think the translator has been that precise about the distinction between the two terms.

The manual talks about "sensitivity" in a situation that must clearly be about "precision"--that is, the vertical line detection when the system is in high-precision mode. The manual says:

"... vertical line detection is twice as sensitive as horizontal line detection..." Given that the difference is the baseline of the pixel array, the difference is NOT "how strong the contrast has to be," it's "the amount the focus might be off the ideal position."

--
RDKirk
'TANSTAAFL: The only unbreakable rule in photography.'
 
Hi Cindy,

I own both, the 20D and the 350D (and the 300D before that). I shoot mainly weddings (with the 350D as my backup and 2nd camera). I found, that even with a 2.8 lens on it, the 350D has a much harder time focusing in dim light then the 20D.

While the 20D is almost always spot on as long as I can see my subject, the 350D starts hunting when it gets too dark. At least that is my observation.

As to precision: If you don't use the 350D for quick action (where it has to refocus quickly), it does very well.

Best Regards,
Marco...
 
Hi, R,

Thanks so much for this. It is very helpful
RDKirk said:
Doug,

Someone on Fred Miranda's site has provided this link:
http://cweb.canon.jp/camera/eosd/20d/catalog/index06.html

Someone with good skill in Japanese can probably give us more
information.

Notice the color coding of the sensors and the legend.
Click to expand...
Yes, It would be nice to get a tranlation of the legend.
RDKirk said:
At f5.6 (blue colored on the diagram), we see the two linear pixel
sensors of each array (a pair of sensors make up an array to sense
the phase difference for a line of contrast). So we see two sensors
arranged horizontally (the array that senses vertical lines) and
two sensors arranged vertically (the array that senses horizontal
lines).
Click to expand...
Yes, these are teh ones Ron speak of as "slightly offset", which is intimated by the graphic.
RDKirk said:
At f2.8 (both green and red on the diagram) we see the same two
vertically aligned sensors active, but a different set of
horiztonally aligned sensors have been activated. These are farther
apart, making the baseline wider, I'd suspect to provide the
greater precision Canon claims with f2.8 lenses.
Click to expand...
Yes, with regard to vertical lines only.
RDKirk said:
This is different from the way I understood it--it's not an
"additional" horizontal array, but a different array with higher
precision. It does correspond to Rob Galbraith's description in his
review of the 20D.
Click to expand...
This is wholly consistent with Rob's description, as I understood it.

It also certainly suggests that, in "high-precision" mode (invoked with an aperture of at least f/2.8), the higher precision (presumably resulting from a greater "baseline") only obtains for the "horizontal" component (that sensitive to vertical lines).

That was of course my conclusion from my analysis of Rob's review. But Karl seems quite adamant that "high precision" obtains, for apertures less than f/2.8, for both directions.

The language in the manual cannot be considered absolutely definitive in that regard, saying that [with an aperture of at least f/2.8] " high-precision cross-type AF sensitive to both vertical and horizontal lines is possible." Note that this doesn say that it is "high precision" with respect to both vertical and horizontal lines. That would be a reasonable inference, but the language doesn't reaslly say that. A system that was sensitive to bith vertcal and horizontal lines, and which was "high precision" for only one axis, could certainly be covered by that language.

It also suggests that with respect to the vertical array (for horizontal lines), the "dual array", which probably results in the "greater sensitivity" that is spoken of by Rob and in the manual (and which I mistook for "greater precision", until Karl straightened me out) probablky still comes to our benefit in "high precision" mode.

But all this also means that, in "high-precision" mode, we only hasve truly high precision with respect to vertical lines.

So my best guess at this point is:

1. For horizontal lines, we have "increased sensitivity" for all apertures greater than f/5.6.

2. For vertical lines, we have "high precision" for apertures larger than f/2.8.

This is essentially my original position except for, in clause 1, the replacement of "greater precision" with "greater sensitivity" (whatever exactly that means).

Thnaks again for your inputs here.

Best regards,

Doug

Visit The Pumpkin, a library of my technical articles on photography, optics, and other topics:

http://doug.kerr.home.att.net/pumpkin

My EOS 20D must be a point and shoot camera - I find out that if I don't point, I get erratic results; if I don't shoot, I get no results.
 
so the 350d has a harder time focusing in challenging conditions . . .but when it does it is accurate. Tough call. I've been thinking about it partly because my daughter away at college has a 10d. She has assisted a few weddings and is trying to get on longterm as a wedding photographer assistant. The lower noise in low light of the 20d would be a good thing and I offered to trade her the 10d for my 20d (gulp, snif) is she gets a good job, since I don't do weddings (as some friends just called me last night to beg my services at cousin's wedding . . . ). But I would really hate to give up the 20d, I like it so much more than the 10d. Another option might be to sell the 10d and buy the XT - it has a quieter shutter than the 20d anyway. And I understand it would be comparible to the 20d in terms of low light image quality, and of course resolution. Or just use the 10d - it is a decent camera and has already done 3 weddings afterall. Any thoughts on this?

Cindy

--
cariarer said:
cariarer said:
cariarer said:
My Photo-A-Day for a Year Gallery Here: http://www.pbase.com/cindyd/2005
Click to expand...
Click to expand...
Click to expand...
Formerly half of 1ofUs or 1ofUs!
The other half of 'Us' is now 'Lensbaby'
 
AK_2 said:
Because 20d's high precision sensors have an issue with big
apertures (worse with 1.2 and 1.4 lenses mentioned).

The evidence is clear to see at http://www.canon-dslr.com , the
pictures showing two shots, in a row, with completely different
focus from 20d with such lenses. Varience is far less with f1.8 or
f4, and very good with f2.8.

The 10d, 300/350, d60 didn't suffer this difference across lenses
of different apertures (f2.8 to f1.2).

Whether it's caused by a simple programming error by canon, or a
fundamental design flaw of the hp sensor I do not know. I only know
the problem is their, across a number of tested bodies and lenses,
and that canon has yet to acknowledge the problem or a fix for it.
It may have been fixed on the quiet with a flash upgrade, but even
in the last 2 months, I have had further reports from top uk pros
buying 20d's that have suffered same issues with the big aperture
lenses mentioned.
Click to expand...
I suspect that this effect is caused by either residual spherical aberation (50mm and 85mm) or from high order spherical aberation (a real bear to correct with very fast lenses) and the sampling size of the high accuracy sensors.

If the high accuracy sensor has a smaller pixel size than the 20D, small residual uncorrected aberations can blead across one sensing pixel to the next, giving the illusion of softness without being OoF. So, the camera decides that the lens is focused as well as the phasal information can determine.

Thanks for the lens by lens data.
--
Mitch
 
thanks so much. very helpful.
AK_2 said:
pentool said:
ok more confusions... why 50mm 1.4 is better with xt? why would any
lens be better on the xt at all?
Click to expand...
Because 20d's high precision sensors have an issue with big
apertures (worse with 1.2 and 1.4 lenses mentioned).

The evidence is clear to see at http://www.canon-dslr.com , the
pictures showing two shots, in a row, with completely different
focus from 20d with such lenses. Varience is far less with f1.8 or
f4, and very good with f2.8.

The 10d, 300/350, d60 didn't suffer this difference across lenses
of different apertures (f2.8 to f1.2).

Whether it's caused by a simple programming error by canon, or a
fundamental design flaw of the hp sensor I do not know. I only know
the problem is their, across a number of tested bodies and lenses,
and that canon has yet to acknowledge the problem or a fix for it.
It may have been fixed on the quiet with a flash upgrade, but even
in the last 2 months, I have had further reports from top uk pros
buying 20d's that have suffered same issues with the big aperture
lenses mentioned.

Regards,
Kev
Click to expand...
 
Google translation of diagram:

"9 point AF sensor loading of new development. In this class high combination 焦 precision was actualized with the new F2.8 corresponding sensor.

It has the base length of two times the F5.6 corresponding sensors in the central range point, it arranges the F2.8 lens corresponding central side sensor. At the time of the lens use which is brighter than the F2.8, compared to high accuracy AF photographing is possible.
  • EF50mm F2.5 compact macro, the EF28-80mm F2.8-4l USM is excluded
With the central vertical sensor and the F5.6 corresponding central side sensor of F5.6 corresponding 2 line range, high accuracy cross type range is actualized.

Furthermore, precision is raised with the diagonal 4 range point and the left and right range point which expanded former compared to base length. As for the range brightness range of the sensor you can obtain accurate focusing even at the place where the EV-0.5 - 18 and light are little.

In addition, at the time of built-in Strobo use, it projects the AF auxiliary light from Strobo side due to the LED at the time of intermittent radiation and EX series Strobo use of the built-in Strobo. You are proud of high AF precision.

With the applied photographing zone, three AF modes and manual focusing of the one-shot AF, the dynamic body estimate * artificial intelligence servo AF and the artificial intelligence focusing AF are selection possible.

With the simple photographing zone, the AF mode which is optimum to photographing mode is automatically set."
--
RDKirk
'TANSTAAFL: The only unbreakable rule in photography.'
 
. . . or of the piece in which it appears?

I'm sure it can't be the diagram - for one thing, the annotations there are in Kanji as an image. Besides there are things on the diagram I don't find in the translation (like "2 raiso", whatever that means).

But it's fascinating.

Thanks again.

Best regards,

Doug
 
The diagram is an image that Google can't parse.

--
RDKirk
'TANSTAAFL: The only unbreakable rule in photography.'
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top