Relevant news re malfunctioning AF under artificial light

Christine Tham · 02-06-2011, 07:45 PM

Originally posted by Eric Seavey

Actually, the K5 did focus and lock in the conditions I was shooting in.

The problem is not necessarily the camera. Yes, it did focus and lock, but you have no way of knowing WHAT it focused on.

Even if you have chosen the AF point and you are absolutely certain it is focusing on the area you wanted.

The AF point is not a "point", it is actually an area around the point. It could have focused on something quite close to the point. The Canon AF system on their pro cameras make this clear by not representing the AF point as a "point" - it draws it as a red rectangle on the screen. When it locks focus, the camera is saying "I have focused on something inside the rectangle - which part of the rectangle, I can't tell you."

For example, you tried focusing on someone's eye, but it's actually focusing on the hair in front of the eye.

If you are really shooting at f1.8 at close objects, the depth of field would be extremely shallow - may even be no more than a few mm.

The best solution as I've mentioned (which I've learnt the hard way) is never shoot wide open at close objects unless I really want a super shallow depth of field.

Even flowers for example in broad daylight, anything less than f4 on the DA70mm will not capture the entire flower in focus.

You will find even at f8 the DOF is still shallow enough to give good bokeh - the main issue is exposure time which is where high ISO comes in handy.

Christine Tham · 02-06-2011, 08:04 PM

Here is a good example of just how narrow depth of field can be at close range.

This is a photo taken at f4 (nowhere near the f2 limit of the lens) - on a Canon EOS-1D Mark III - the lens is an EF135 f/2 USM:

Notice that even at f4 the depth of field is very shallow - effectively less than 1cm thick - which you can see on the left of the photo - which is the narrow strip of glass partition running down from top to bottom that's actually in focus

You will notice the only part of the Tasmanian Devil that's actually in focus is few strands of fur around the eyes and around the ears - the rest of the animal is out of focus.

I like this photo even though almost the entire photo is out of focus, because it shows just how crazy shallow DOF can get - at f4.

If you are shooting at f1.8, can you imagine how shallow your DOF is? No wonder the pictures seems out of focus - if you go back and check them now you'll probably find the camera focused on a very small part of the image - probably not the part you intended.

PS - to get that photo, I had to shoot 6-12 shots, and on most of those shots the camera focused on the wrong part of the animal - even though I was aiming at the eye for each and every shot. And even on that shot I still missed - the eye is slightly out of focus - what was in focus is a few strands of fur around the eye.

**Eric Seavey** · 02-06-2011, 08:33 PM

Christine,
I understand how narrow the DOF becomes with close subjects, since I also shoot macro (insects). I also understand that the AF sensors do not represent the dots in the viewfinder. The issues I am having relate to a consistent offset when the available light falls below a certain level. Under the same conditions the K10D hunts and takes some time to focus, but once locked it has focused correctly. The K5 seems to hunt much less, but consistently locks with a significant FF. These are conditions where the AF assist light doesn't even come on yet. Stopping down from F/1.4 to f/2.8 isn't even enough to compensate for the FF.

For taking dance photos I used to use f/1.4 30mm with my K10D, but now it seems I need to stop down to f/2.2 at least. The K10d was not known for its AF capability, but I managed with a little patience it would lock and I got a sharp photo. Now with the K5 and its improved low light AF, I end up with many more unusable shots due to the FF issue. Sure the K5 focuses fast and consistent, but FF. Like I mentioned earlier, If I could adjust the AF fine adjustment to somewhere between -20 and -30 when shooting in dim light, this would not be as much of a problem.
Thanks

tbirdas · 02-06-2011, 08:49 PM

Originally posted by Christine Tham

Here is a good example of just how narrow depth of field can be at close range.

This is a photo taken at f4 (nowhere near the f2 limit of the lens) - on a Canon EOS-1D Mark III - the lens is an EF135 f/2 USM:

Notice that even at f4 the depth of field is very shallow - effectively less than 1cm thick - which you can see on the left of the photo - which is the narrow strip of glass partition running down from top to bottom that's actually in focus

You will notice the only part of the Tasmanian Devil that's actually in focus is few strands of fur around the eyes and around the ears - the rest of the animal is out of focus.

I like this photo even though almost the entire photo is out of focus, because it shows just how crazy shallow DOF can get - at f4.

If you are shooting at f1.8, can you imagine how shallow your DOF is? No wonder the pictures seems out of focus - if you go back and check them now you'll probably find the camera focused on a very small part of the image - probably not the part you intended.

PS - to get that photo, I had to shoot 6-12 shots, and on most of those shots the camera focused on the wrong part of the animal - even though I was aiming at the eye for each and every shot. And even on that shot I still missed - the eye is slightly out of focus - what was in focus is a few strands of fur around the eye.

Isn't that affected by the size of the sensor, meaning that a full frame camera will have a narrower DOF at the same lens aperture? A crop sensor will likely not get the same result at f/4

Christine Tham · 02-06-2011, 09:00 PM

Eric,

You do realise don't you the size of crosshairs are different on K-5 compared to your K10D?

That alone would account for the differences you are reporting.

Also the K-5 changes the way it focuses, it is more consistent with Canon compared the K10D.

I have taken quite a few shots in 7fps mode on my K-5 yesterday, and it seems to me the K-5 will try it's best to focus on the closest object located in the AF area.

The K10D has a tendency to hunt around a bit more and may focus much further back.

So if you are used to the behaviour of your K10D you need to learn to shoot differently on the K-5, otherwise it would seem the K-5 is FF when in fact it's not.

I personally think the way the K-5 does it is more consistent and reliable and more like what Canon does.

I still think you are shooting at an aperture that is too large given the conditions you are describing, but at the end of the day hey it's your camera, your shot so good luck getting the picture you want!

Christine Tham · 02-06-2011, 09:01 PM

Originally posted by tbirdas

Isn't that affected by the size of the sensor, meaning that a full frame camera will have a narrower DOF at the same lens aperture? A crop sensor will likely not get the same result at f/4

EOS 1D Mark III is not full frame - it's APS.

snofox · 02-06-2011, 10:33 PM

^^^^^^^^^^^^^^^

It's a full frame CMOS sensor, isn't it?

Christine Tham · 02-06-2011, 11:04 PM

Originally posted by snofox

^^^^^^^^^^^^^^^

It's a full frame CMOS sensor, isn't it?

No. It's APS-H, as I have already said.

You are probably confusing it with the EOS-1DS Mark III, which is full frame.

Christine Tham · 02-06-2011, 11:44 PM

Here's a "real life" shot under dim indoor lighting - no I am not going to quote EV number but it is ISO3200 so that gives an indication how dark it is.

As can be seen - the face is in focus, but the hands in front aren't.

I'm glad my K-5 doesn't have an FF issue, but in reality even if it did it wouldn't have mattered in a situation like this, because the lens needed to be stopped down to bring the whole face into focus.

creampuff · 02-07-2011, 12:29 AM

Christine, thank you for the cogent explanation. What you mention about the AF point is pretty much what I've experienced.

Gimbal · 02-07-2011, 12:29 AM

Am I the only one that don't see Christines pictures?

creampuff · 02-07-2011, 12:38 AM

Originally posted by Gimbal

Am I the only one that don't see Christines pictures?

I see her photos of the Tasmanian devil and little kid fine.

Tommot1965 · 02-07-2011, 12:45 AM

Originally posted by Eric Seavey

Christine,
I understand how narrow the DOF becomes with close subjects, since I also shoot macro (insects). I also understand that the AF sensors do not represent the dots in the viewfinder. The issues I am having relate to a consistent offset when the available light falls below a certain level. Under the same conditions the K10D hunts and takes some time to focus, but once locked it has focused correctly. The K5 seems to hunt much less, but consistently locks with a significant FF. These are conditions where the AF assist light doesn't even come on yet. Stopping down from F/1.4 to f/2.8@ 50mm isn't even enough to compensate for the FF.

For taking dance photos I used to use f/1.4 30mm with my K10D, but now it seems I need to stop down to f/2.2 at least. The K10d was not known for its AF capability, but I managed with a little patience it would lock and I got a sharp photo. Now with the K5 and its improved low light AF, I end up with many more unusable shots due to the FF issue. Sure the K5 focuses fast and consistent, but FF. Like I mentioned earlier, If I could adjust the AF fine adjustment to somewhere between -20 and -30 when shooting in dim light, this would not be as much of a problem.
Thanks

Eric, at this stage, its about knowing the limitations of the K5 and working within-them to get usable images .
If you know about the K5 and FF at low EV..then why shoot at large apertures {shallow DOF] in a low EV environment ....instead take advantage of the improved high ISO over the K10 , possible three stops better, then shoot stopped down, and get a usable image even at EV3 or lower ?..this is what im doing at the moment.

Ill only open up if the subject is far enough away {increased DOF} to mask the FF..for me somewhere around 4-5 metres at F2.8@50mm anything closer and I stop down to F4..underexpose if I have to, then bump the exposure in PS during PP..the K5 handles that no worries even under exposures of two stops can be pulled back to a reasonable image..I only mess about like this if the EV is low enough to induce the FF

blende8 · 02-07-2011, 01:10 AM

Christine, it seems that you haven't followed this thread closely.
The main point is that the earlier cameras like the K20 focus correctly where the K5 fails.

I am repeating below a test, which I posted further above already:

FA50, f1.4, EV = 1,3
Fokused on the center folder.
The yellow folder is ca. 6 cm in front.
This is reproducible.

NB: The K20 is focusing correctly here, both times!

Christine Tham · 02-07-2011, 01:26 AM

Originally posted by blende8

Christine, it seems that you haven't followed this thread closely.
The main point is that the earlier cameras like the K20 focus correctly where the K5 fails.

I'm not sure you have followed my posts closely either.

My results are both the K-5 and my K10D have identical performance - both are focusing correctly down to around EV 1 (according to jolepp's calculation based on my data) but below EV 1 both have slight FF.

Under the same conditions, the Canon EOS-1D refuses to lock focus at below EV 1.

So basically my K-5 is no worse than 2 other cameras. Actually, better, since the K-5 is able to lock focus in Live View below EV 1.

02-06-2011, 08:04 PM	#377
Christine Tham Veteran Member Join Date: Nov 2008 Location: Sydney, Australia Photos: Gallery Posts: 2,269	Here is a good example of just how narrow depth of field can be at close range. This is a photo taken at f4 (nowhere near the f2 limit of the lens) - on a Canon EOS-1D Mark III - the lens is an EF135 f/2 USM: Notice that even at f4 the depth of field is very shallow - effectively less than 1cm thick - which you can see on the left of the photo - which is the narrow strip of glass partition running down from top to bottom that's actually in focus You will notice the only part of the Tasmanian Devil that's actually in focus is few strands of fur around the eyes and around the ears - the rest of the animal is out of focus. I like this photo even though almost the entire photo is out of focus, because it shows just how crazy shallow DOF can get - at f4. If you are shooting at f1.8, can you imagine how shallow your DOF is? No wonder the pictures seems out of focus - if you go back and check them now you'll probably find the camera focused on a very small part of the image - probably not the part you intended. PS - to get that photo, I had to shoot 6-12 shots, and on most of those shots the camera focused on the wrong part of the animal - even though I was aiming at the eye for each and every shot. And even on that shot I still missed - the eye is slightly out of focus - what was in focus is a few strands of fur around the eye. Last edited by Christine Tham; 02-06-2011 at 08:12 PM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Help! K7 malfunctioning	samski_1	Pentax DSLR Discussion	9	07-11-2010 03:24 PM
Probably not relevant or welcome here	climbertrev	Canon, Nikon, Sony, and Other Camera Brands	4	05-24-2010 05:15 AM
camera based sensorcleaning how relevant is it.	hein	Pentax DSLR Discussion	13	09-13-2009 12:26 PM
Light fluff but news.. sort of	jeffkrol	Pentax News and Rumors	8	12-30-2008 06:30 PM

02-06-2011, 07:45 PM	#376
Christine Tham Veteran Member Join Date: Nov 2008 Location: Sydney, Australia Photos: Gallery Posts: 2,269	Originally posted by Eric Seavey Actually, the K5 did focus and lock in the conditions I was shooting in. The problem is not necessarily the camera. Yes, it did focus and lock, but you have no way of knowing WHAT it focused on. Even if you have chosen the AF point and you are absolutely certain it is focusing on the area you wanted. The AF point is not a "point", it is actually an area around the point. It could have focused on something quite close to the point. The Canon AF system on their pro cameras make this clear by not representing the AF point as a "point" - it draws it as a red rectangle on the screen. When it locks focus, the camera is saying "I have focused on something inside the rectangle - which part of the rectangle, I can't tell you." For example, you tried focusing on someone's eye, but it's actually focusing on the hair in front of the eye. If you are really shooting at f1.8 at close objects, the depth of field would be extremely shallow - may even be no more than a few mm. The best solution as I've mentioned (which I've learnt the hard way) is never shoot wide open at close objects unless I really want a super shallow depth of field. Even flowers for example in broad daylight, anything less than f4 on the DA70mm will not capture the entire flower in focus. You will find even at f8 the DOF is still shallow enough to give good bokeh - the main issue is exposure time which is where high ISO comes in handy.

02-06-2011, 08:33 PM	#378
Eric Seavey Site Supporter Join Date: Feb 2008 Location: Irvine, CA Photos: Gallery \| Albums Posts: 307	Christine, I understand how narrow the DOF becomes with close subjects, since I also shoot macro (insects). I also understand that the AF sensors do not represent the dots in the viewfinder. The issues I am having relate to a consistent offset when the available light falls below a certain level. Under the same conditions the K10D hunts and takes some time to focus, but once locked it has focused correctly. The K5 seems to hunt much less, but consistently locks with a significant FF. These are conditions where the AF assist light doesn't even come on yet. Stopping down from F/1.4 to f/2.8 isn't even enough to compensate for the FF. For taking dance photos I used to use f/1.4 30mm with my K10D, but now it seems I need to stop down to f/2.2 at least. The K10d was not known for its AF capability, but I managed with a little patience it would lock and I got a sharp photo. Now with the K5 and its improved low light AF, I end up with many more unusable shots due to the FF issue. Sure the K5 focuses fast and consistent, but FF. Like I mentioned earlier, If I could adjust the AF fine adjustment to somewhere between -20 and -30 when shooting in dim light, this would not be as much of a problem. Thanks

02-06-2011, 08:49 PM	#379
tbirdas Forum Member Join Date: Jun 2010 Location: Indianapolis Posts: 83	Originally posted by Christine Tham Here is a good example of just how narrow depth of field can be at close range. This is a photo taken at f4 (nowhere near the f2 limit of the lens) - on a Canon EOS-1D Mark III - the lens is an EF135 f/2 USM: Notice that even at f4 the depth of field is very shallow - effectively less than 1cm thick - which you can see on the left of the photo - which is the narrow strip of glass partition running down from top to bottom that's actually in focus You will notice the only part of the Tasmanian Devil that's actually in focus is few strands of fur around the eyes and around the ears - the rest of the animal is out of focus. I like this photo even though almost the entire photo is out of focus, because it shows just how crazy shallow DOF can get - at f4. If you are shooting at f1.8, can you imagine how shallow your DOF is? No wonder the pictures seems out of focus - if you go back and check them now you'll probably find the camera focused on a very small part of the image - probably not the part you intended. PS - to get that photo, I had to shoot 6-12 shots, and on most of those shots the camera focused on the wrong part of the animal - even though I was aiming at the eye for each and every shot. And even on that shot I still missed - the eye is slightly out of focus - what was in focus is a few strands of fur around the eye. Isn't that affected by the size of the sensor, meaning that a full frame camera will have a narrower DOF at the same lens aperture? A crop sensor will likely not get the same result at f/4

02-06-2011, 09:00 PM	#380
Christine Tham Veteran Member Join Date: Nov 2008 Location: Sydney, Australia Photos: Gallery Posts: 2,269	Eric, You do realise don't you the size of crosshairs are different on K-5 compared to your K10D? That alone would account for the differences you are reporting. Also the K-5 changes the way it focuses, it is more consistent with Canon compared the K10D. I have taken quite a few shots in 7fps mode on my K-5 yesterday, and it seems to me the K-5 will try it's best to focus on the closest object located in the AF area. The K10D has a tendency to hunt around a bit more and may focus much further back. So if you are used to the behaviour of your K10D you need to learn to shoot differently on the K-5, otherwise it would seem the K-5 is FF when in fact it's not. I personally think the way the K-5 does it is more consistent and reliable and more like what Canon does. I still think you are shooting at an aperture that is too large given the conditions you are describing, but at the end of the day hey it's your camera, your shot so good luck getting the picture you want!

02-06-2011, 09:01 PM	#381
Christine Tham Veteran Member Join Date: Nov 2008 Location: Sydney, Australia Photos: Gallery Posts: 2,269	Originally posted by tbirdas Isn't that affected by the size of the sensor, meaning that a full frame camera will have a narrower DOF at the same lens aperture? A crop sensor will likely not get the same result at f/4 EOS 1D Mark III is not full frame - it's APS.

02-06-2011, 10:33 PM	#382
snofox Veteran Member Join Date: Jan 2009 Location: Fredericton New Brunswick Canada Photos: Albums Posts: 334	^^^^^^^^^^^^^^^ It's a full frame CMOS sensor, isn't it?

02-06-2011, 11:04 PM	#383
Christine Tham Veteran Member Join Date: Nov 2008 Location: Sydney, Australia Photos: Gallery Posts: 2,269	Originally posted by snofox ^^^^^^^^^^^^^^^ It's a full frame CMOS sensor, isn't it? No. It's APS-H, as I have already said. You are probably confusing it with the EOS-1DS Mark III, which is full frame.

02-06-2011, 11:44 PM	#384
Christine Tham Veteran Member Join Date: Nov 2008 Location: Sydney, Australia Photos: Gallery Posts: 2,269	Here's a "real life" shot under dim indoor lighting - no I am not going to quote EV number but it is ISO3200 so that gives an indication how dark it is. As can be seen - the face is in focus, but the hands in front aren't. I'm glad my K-5 doesn't have an FF issue, but in reality even if it did it wouldn't have mattered in a situation like this, because the lens needed to be stopped down to bring the whole face into focus.

02-07-2011, 12:29 AM	#385
creampuff Veteran Member Join Date: May 2007 Location: Singapore Photos: Gallery Posts: 3,953	Christine, thank you for the cogent explanation. What you mention about the AF point is pretty much what I've experienced.

02-07-2011, 12:29 AM	#386
Gimbal Pentaxian Join Date: Apr 2007 Location: Sweden Posts: 2,106	Am I the only one that don't see Christines pictures?

02-07-2011, 12:38 AM	#387
creampuff Veteran Member Join Date: May 2007 Location: Singapore Photos: Gallery Posts: 3,953	Originally posted by Gimbal Am I the only one that don't see Christines pictures? I see her photos of the Tasmanian devil and little kid fine.

02-07-2011, 12:45 AM	#388
Tommot1965 Veteran Member Join Date: Sep 2010 Location: Perth Western Australia Photos: Gallery \| Albums Posts: 1,026	Originally posted by Eric Seavey Christine, I understand how narrow the DOF becomes with close subjects, since I also shoot macro (insects). I also understand that the AF sensors do not represent the dots in the viewfinder. The issues I am having relate to a consistent offset when the available light falls below a certain level. Under the same conditions the K10D hunts and takes some time to focus, but once locked it has focused correctly. The K5 seems to hunt much less, but consistently locks with a significant FF. These are conditions where the AF assist light doesn't even come on yet. Stopping down from F/1.4 to f/2.8@ 50mm isn't even enough to compensate for the FF. For taking dance photos I used to use f/1.4 30mm with my K10D, but now it seems I need to stop down to f/2.2 at least. The K10d was not known for its AF capability, but I managed with a little patience it would lock and I got a sharp photo. Now with the K5 and its improved low light AF, I end up with many more unusable shots due to the FF issue. Sure the K5 focuses fast and consistent, but FF. Like I mentioned earlier, If I could adjust the AF fine adjustment to somewhere between -20 and -30 when shooting in dim light, this would not be as much of a problem. Thanks Eric, at this stage, its about knowing the limitations of the K5 and working within-them to get usable images . If you know about the K5 and FF at low EV..then why shoot at large apertures {shallow DOF] in a low EV environment ....instead take advantage of the improved high ISO over the K10 , possible three stops better, then shoot stopped down, and get a usable image even at EV3 or lower ?..this is what im doing at the moment. Ill only open up if the subject is far enough away {increased DOF} to mask the FF..for me somewhere around 4-5 metres at F2.8@50mm anything closer and I stop down to F4..underexpose if I have to, then bump the exposure in PS during PP..the K5 handles that no worries even under exposures of two stops can be pulled back to a reasonable image..I only mess about like this if the EV is low enough to induce the FF

02-07-2011, 01:10 AM	#389
blende8 Veteran Member Join Date: Feb 2007 Location: Bremen, Germany Photos: Albums Posts: 1,521	Christine, it seems that you haven't followed this thread closely. The main point is that the earlier cameras like the K20 focus correctly where the K5 fails. I am repeating below a test, which I posted further above already: FA50, f1.4, EV = 1,3 Fokused on the center folder. The yellow folder is ca. 6 cm in front. This is reproducible. NB: The K20 is focusing correctly here, both times!

02-07-2011, 01:26 AM	#390
Christine Tham Veteran Member Join Date: Nov 2008 Location: Sydney, Australia Photos: Gallery Posts: 2,269	Originally posted by blende8 Christine, it seems that you haven't followed this thread closely. The main point is that the earlier cameras like the K20 focus correctly where the K5 fails. I'm not sure you have followed my posts closely either. My results are both the K-5 and my K10D have identical performance - both are focusing correctly down to around EV 1 (according to jolepp's calculation based on my data) but below EV 1 both have slight FF. Under the same conditions, the Canon EOS-1D refuses to lock focus at below EV 1. So basically my K-5 is no worse than 2 other cameras. Actually, better, since the K-5 is able to lock focus in Live View below EV 1.