Canon improved the AF for 5d mark II

Started Nov 14, 2008 | Discussions
Nicolaj Roos Regular Member • Posts: 236
Canon improved the AF for 5d mark II

For people who speaks swedish, here's a link to an interview from Photokina 2008, with swedish productchief, talking about the 5d mark II. He's saying that although the mark II's got the 9 point AF from the original 5d, the AF algoritms has been improved.

http://www.kamerabild.se/artikel/de_okanda_nyheterna_i_canon_eos_5d_mark_ii_080924110207-823.html

darktiger Regular Member • Posts: 226
Re: Canon improved the AF for 5d mark II

I agree, I played with the 5D Mark II for about 15minutes last night. The AF is a whole lot better than my 5D..... I can't wait for my pre-order to come in

 darktiger's gear list:darktiger's gear list
Fujifilm FinePix X100 Fujifilm X-E1 Fujifilm X-T1 Fujifilm X-Pro2 Fujifilm XF 35mm F1.4 R +6 more
Andre Affleck Senior Member • Posts: 2,362
Re: Canon improved the AF for 5d mark II

I used to think that algorithm and processor changes would only affect speed and not accuracy, but it may not be the case. If you think about, once the AF sensors acquire the 2 "images" that is used to detect the phase difference, how the phase detection is done (algorithm) and how the "images" are processed before detection should affect the accuracy. This is done in the processor stage (I believe). So if there are any changes in NR and contrast techniques within the AF processing, for example, it may result in more accurate AF. We may very well have a better focusing 5DMII than its predecessor. We'll soon see.

jerryk Veteran Member • Posts: 6,625
Re: Canon improved the AF for 5d mark II

Software can only help so much. The biggest issue people are complaining about is that the AF sensor on the edges are not cross type. As a result they do not work with all subjects. So if you place a moving subject on one the non-center sensors the camera may not Autofocus correctly.

-- hide signature --

jerryk.smugmug.com

Andre Affleck Senior Member • Posts: 2,362
Re: Canon improved the AF for 5d mark II

True, but it may help explain how some early tests have indicated improved AF performance given the the same hardware. Even if the edge sensors are still unidirectional, improved accuracy through software would surely be welcomed.

jerryk wrote:

Software can only help so much. The biggest issue people are
complaining about is that the AF sensor on the edges are not cross
type. As a result they do not work with all subjects. So if you
place a moving subject on one the non-center sensors the camera may
not Autofocus correctly.

memento Contributing Member • Posts: 774
which tests? (nt)
Andre Affleck Senior Member • Posts: 2,362
Re: which tests? (nt)

Nothing formal of course. There is no such thing as any formal AF accuracy testing for any camera. I think we need a review site that will commit to doing formal AF testing scientifically. I'm talking about people who tested the camera and have reported an improvement from their own 5D experience.

Here are a few more aside from than this very thread:

http://forums.dpreview.com/forums/readflat.asp?forum=1032&thread=29997714

http://forums.dpreview.com/forums/readflat.asp?forum=1032&message=30052332&q=5d+af&qf=m

memento Contributing Member • Posts: 774
Re: which tests? (nt)

ah ok, I understand.

well, as a health professional, all I'll say is that the placebo effect is capable of amazing things.

Unless Canon has secretly added cross-type points to the outer portions, I'll keep a bag of salt with me.

Sal Baker Forum Pro • Posts: 11,490
Re: which tests? (nt)

memento wrote:

ah ok, I understand.

well, as a health professional, all I'll say is that the placebo
effect is capable of amazing things.

Unless Canon has secretly added cross-type points to the outer
portions, I'll keep a bag of salt with me.

He's talking about the people who have used the 5D and the 5DII and say that the relative speed and accuracy in low light has been improved. No one said they added cross-type points to the outer portions.

If the center AF point is fast and accurate in low light, that's all I care about for my particular needs. Your needs may differ.

Sal

 Sal Baker's gear list:Sal Baker's gear list
Ricoh GR Fujifilm X-E2 Fujifilm X-T100 Fujifilm XF 14mm F2.8 R Fujifilm XF 18-55mm F2.8-4 R LM OIS +3 more
pz797_mizzou Regular Member • Posts: 228
Re: Canon improved the AF for 5d mark II

When I use 70-200/f4 on my 5D. the center point becomes non-cross type. But it's still much faster than the other points.

-- hide signature --
saynomore Veteran Member • Posts: 4,254
Re: which tests? (nt)

Sal Baker wrote:

memento wrote:

ah ok, I understand.

well, as a health professional, all I'll say is that the placebo
effect is capable of amazing things.

Unless Canon has secretly added cross-type points to the outer
portions, I'll keep a bag of salt with me.

He's talking about the people who have used the 5D and the 5DII and
say that the relative speed and accuracy in low light has been
improved. No one said they added cross-type points to the outer
portions.

And that's exactly the kind of anecdotal evidence that makes it ripe for placebo effect. I think that's what memento was saying. "People" saying they feel or even "in their experience" the AF is faster or more accurate is not good enough, not without providing good, statistically valid evidence.

Although I wonder what memento thinks of acupuncture

-- hide signature --

Andy

Andre Affleck Senior Member • Posts: 2,362
Re: which tests? (nt)

Yes, like I said on my previous post, I think it is possible for the overall accuracy to be improved using software alone. If you think about it, there are 2 aspects to good AF performance. 1) You need good AF hardware to deliver high fidelity data, 2) you need to process that data to come up with an accurate focal distance for the lens command.

I can see using better software to take pixel strip image data from the existing hardware, and processing it with improved phase difference algorithms. This may include better contrast enhancement and NR in which to improve S/N and better detect the phase shifts.

On the other hand, this may be all just wishful thinking!

Sal Baker wrote:

He's talking about the people who have used the 5D and the 5DII and
say that the relative speed and accuracy in low light has been
improved. No one said they added cross-type points to the outer
portions.

If the center AF point is fast and accurate in low light, that's all
I care about for my particular needs. Your needs may differ.

Sal

Andre Affleck Senior Member • Posts: 2,362
Re: which tests? (nt)

I completely sympathize with what memento is saying. No doubt, the overall subjectivity of AF performance in general makes it very susceptible to the placebo affect.

On the other hand, you have to ask yourself, given the performance that you are accustom to getting with your 5D for the last 2-3 years, would you be able to detect a change in that performance given different hardware? I think I would. I know exactly how my copy behaves, and what to expect as far as keepers are concerned. I detected a change going form the 20D to the 5D and I was also afraid of that placebo affect. I proved it to myself using past images and controlled tests, and I was right.

Having said that, I still think memento is wise to remain skeptical.

saynomore wrote:

Sal Baker wrote:

memento wrote:

ah ok, I understand.

well, as a health professional, all I'll say is that the placebo
effect is capable of amazing things.

Unless Canon has secretly added cross-type points to the outer
portions, I'll keep a bag of salt with me.

He's talking about the people who have used the 5D and the 5DII and
say that the relative speed and accuracy in low light has been
improved. No one said they added cross-type points to the outer
portions.

And that's exactly the kind of anecdotal evidence that makes it ripe
for placebo effect. I think that's what memento was saying. "People"
saying they feel or even "in their experience" the AF is faster or
more accurate is not good enough, not without providing good,
statistically valid evidence.

Although I wonder what memento thinks of acupuncture

Sal Baker Forum Pro • Posts: 11,490
Re: which tests? (nt)

saynomore wrote:

And that's exactly the kind of anecdotal evidence that makes it ripe
for placebo effect. I think that's what memento was saying. "People"
saying they feel or even "in their experience" the AF is faster or
more accurate is not good enough, not without providing good,
statistically valid evidence.

Well, anecdotal evidence is all we have to go on at this point. And I would trust opinions from those who have used both cameras more than opinions from folks who haven't yet touched a 5DII, wouldn't you.

Sal

 Sal Baker's gear list:Sal Baker's gear list
Ricoh GR Fujifilm X-E2 Fujifilm X-T100 Fujifilm XF 14mm F2.8 R Fujifilm XF 18-55mm F2.8-4 R LM OIS +3 more
saynomore Veteran Member • Posts: 4,254
Re: which tests? (nt)

Sal Baker wrote:

saynomore wrote:

And that's exactly the kind of anecdotal evidence that makes it ripe
for placebo effect. I think that's what memento was saying. "People"
saying they feel or even "in their experience" the AF is faster or
more accurate is not good enough, not without providing good,
statistically valid evidence.

Well, anecdotal evidence is all we have to go on at this point. And
I would trust opinions from those who have used both cameras more
than opinions from folks who haven't yet touched a 5DII, wouldn't
you.

Sal

Actually, with all that's been said and done here, I only trust a handful of people that have built their reputation here over years. And those people do proper tests anyway, that's how they got their reputation. If what we only got going is anecdotal evidence from people that just say so, if they don't say exactly how they tested, I'd rather not know than believing something wrong.
--
Andy

XeroJay Senior Member • Posts: 2,341
What I don't understand

Maybe I'm wrong, but everyone keep saying that Canon uses the same focus system as the original 5d in the new 5dII. This can't be true, as the old 5D uses a 9 point with x-type centre point, and the new 5DII uses a 15 POINT (9, with x-type centre, plus 6 invisible assist points, two of which are 2.8)

Is this not evidence that the AF-system is different from the old 5d, and most likely improved?

 XeroJay's gear list:XeroJay's gear list
Canon EOS M6 Sony a9 Sony a7 III Canon EF 100mm F2.8L Macro IS USM Canon EF-M 22mm f/2 STM +7 more
Tom Batinich Senior Member • Posts: 1,176
Re: What I don't understand

Where did you get this info ?

XeroJay wrote:

Maybe I'm wrong, but everyone keep saying that Canon uses the same
focus system as the original 5d in the new 5dII. This can't be true,
as the old 5D uses a 9 point with x-type centre point, and the new
5DII uses a 15 POINT (9, with x-type centre, plus 6 invisible assist
points, two of which are 2.8)

Is this not evidence that the AF-system is different from the old 5d,
and most likely improved?

desmo101
desmo101 Contributing Member • Posts: 834
Re: What I don't understand

I also read the same thing 9 + 6 invisible. it isn't the same af
--
Lead, follow or get lost

http://www.flickr.com/photos/leeian/

 desmo101's gear list:desmo101's gear list
Fujifilm X-T4 Nikon AF Micro-Nikkor 60mm f/2.8D Voigtlander 40mm F2 Ultron SL II Fujifilm XF 23mm F1.4 R Fujifilm XF 56mm F1.2 R +1 more
Tom Batinich Senior Member • Posts: 1,176
Extra Auto focus points ?

This is from the Canon Website. It cleary states its 9 points, with 1 Cross Type.
Autofocus

Or am I missing something ?

Type
TTL-CT-SIR AF-dedicated CMOS sensor

AF Points
9 AF Points (1 Cross Type) + 6 AF Assist Points

Metering Range
EV -0.5-18 (at 73°F/23°C, ISO 100)

Focusing Modes
Auto, One-Shot AF, Predictive AI Servo AF, AI Focus AF, Manual focusing (MF)

AF Point Selection
Automatic selection, manual selection

Selected AF Point Display
Superimposed in viewfinder and indicated on LCD panel

AF-assist Beam

When an external EOS-dedicated Speedlite is attached to the camera, the AF-assist beam from the Speedlite will be emitted when necessary.

XeroJay wrote:

Maybe I'm wrong, but everyone keep saying that Canon uses the same
focus system as the original 5d in the new 5dII. This can't be true,
as the old 5D uses a 9 point with x-type centre point, and the new
5DII uses a 15 POINT (9, with x-type centre, plus 6 invisible assist
points, two of which are 2.8)

Is this not evidence that the AF-system is different from the old 5d,
and most likely improved?

kenmartin New Member • Posts: 13
Re: What I don't understand

The current 5D has 9+6 invisible. I do think I heard that the 5dm2 uses a couple other inputs into the AF calculation such as white balance and overall light intensity. If true, that would imply some changes were made to the AF algorithms.

Keyboard shortcuts:
FForum MMy threads