On Sensor Phase Detect AF Implementation

[Interceptor121]

I have two different brands of mirrorless camera with PDAF

The first set Sony always focus with the aperture number dialled in except when hitting f/11 where you can tell the camera not to stop down further

The second set panasonic focuses like a DSLR with the lens wide open unless you tell the camera to do live view and then it starts using the aparture number dialled in

Considering that the sensor is being scanned at a fixed rate and those are not pixels used for exposure I see benefits and disbenefits of both systems

In bright scenes it makes senses to close down the aperture to avoid overflow of light however in low light it seems more appropriate to focus wide open

Another parameter is the ISO value dialled in which does not seem to have an effect to the camera focus as if the camera was increasing gain in order to focus as much as possible and then applying the gain for the exposure

Finally the sensor readout can drop in low light but does not go faster in bright light the shutter speed is simulated only in live view making it not clear what it does when live view is not on but clearly cannot drop below the readout rate which again in low light can drop as well

I am trying to determine the best conditions to activate or not live view with my cameras

I am getting the impression that in bright scenes live view makes sense to avoid highlights that could create focus problems while in dark scenes not only I would avoid live view but I would also let the sensor readout droo in order to focus

I am sure Nikon and Canon may have even different implementations so this cannot be generalised but I was wondering if I am on the right track and how I can build a test procedure to validate the assumptions

 

[Horshack]

Considering that the sensor is being scanned at a fixed rate and those are not pixels used for exposure I see benefits and disbenefits of both systems

In bright scenes it makes senses to close down the aperture to avoid overflow of light however in low light it seems more appropriate to focus wide open

Another parameter is the ISO value dialled in which does not seem to have an effect to the camera focus as if the camera was increasing gain in order to focus as much as possible and then applying the gain for the exposure

Finally the sensor readout can drop in low light but does not go faster in bright light the shutter speed is simulated only in live view making it not clear what it does when live view is not on but clearly cannot drop below the readout rate which again in low light can drop as well

Outside of the aperture, the camera doesn't run the sensor at the exposure you have dialed in, ie the shutter speed and ISO, even when you have LV configured to preview exposure. It instead uses a shutter speed and ISO that keeps an exposure within a range that's usable to the metering system, then simulates the shutter speed/ISO you have dialed in. Many systems also simulate the effect of the aperture on the previewed exposure, although like you indicated some actually stop the aperture down, at least to some degree, which is done for several reasons. One is to prevent focus shift. Another is to provide a DOF preview.

Even though the exposure you see in LV is simulated, most MILCs use that simulated exposure to feed the AF system since their imaging ASIC doesn't support dual pipelines, ie one preview for LV and another feeding the AF system. This is why the AF system is sensitive to your previewed exposure - it's not because the aperture, shutter speed, and ISO is preventing the sensor from getting enough (or too much) light.

The only exception to this is on Panasonic cameras, where it does have a mode that actually uses the dialed-in exposure parameter like shutter speed. But that mode is practically-speaking unusable because it creates a stuttered LV at slow shutter speeds that make viewing and manual focusing nearly impossible, which is why other cameras don't let you do this.

 

[Interceptor121]

Considering that the sensor is being scanned at a fixed rate and those are not pixels used for exposure I see benefits and disbenefits of both systems

In bright scenes it makes senses to close down the aperture to avoid overflow of light however in low light it seems more appropriate to focus wide open

Another parameter is the ISO value dialled in which does not seem to have an effect to the camera focus as if the camera was increasing gain in order to focus as much as possible and then applying the gain for the exposure

Finally the sensor readout can drop in low light but does not go faster in bright light the shutter speed is simulated only in live view making it not clear what it does when live view is not on but clearly cannot drop below the readout rate which again in low light can drop as well

Outside of the aperture, the camera doesn't run the sensor at the exposure you have dialed in, ie the shutter speed and ISO, even when you have LV configured to preview exposure. It instead uses a shutter speed and ISO that keeps an exposure within a range that's usable to the metering system, then simulates the shutter speed/ISO you have dialed in. Many systems also simulate the effect of the aperture on the previewed exposure, although like you indicated some actually stop the aperture down, at least to some degree, which is done for several reasons. One is to prevent focus shift. Another is to provide a DOF preview.

Even though the exposure you see in LV is simulated, most MILCs use that simulated exposure to feed the AF system since their imaging ASIC doesn't support dual pipelines, ie one preview for LV and another feeding the AF system. This is why the AF system is sensitive to your previewed exposure - it's not because the aperture, shutter speed, and ISO is preventing the sensor from getting enough (or too much) light.

The only exception to this is on Panasonic cameras, where it does have a mode that actually uses the dialed-in exposure parameter like shutter speed. But that mode is practically-speaking unusable because it creates a stuttered LV at slow shutter speeds that make viewing and manual focusing nearly impossible, which is why other cameras don't let you do this.

Panasonic lets you choose if you want to simulate aperture or aperture and shutter or nothing. It is very useful indeed. Sony only does that in video

the point is that the sensor is scanned anyway at fixed rate af and exposure calculations are unrelated so the camera must do the af at the readout rate it is going

to that respect having the ability to focus wide open or at dialled exposure is useful

what am thinking is how to build some test cases for bright and dark scenes to validate what option to choose on the sony system indeed it will not make a difference as the aperture is the same live view or not

 

[Horshack]

Considering that the sensor is being scanned at a fixed rate and those are not pixels used for exposure I see benefits and disbenefits of both systems

In bright scenes it makes senses to close down the aperture to avoid overflow of light however in low light it seems more appropriate to focus wide open

Another parameter is the ISO value dialled in which does not seem to have an effect to the camera focus as if the camera was increasing gain in order to focus as much as possible and then applying the gain for the exposure

Finally the sensor readout can drop in low light but does not go faster in bright light the shutter speed is simulated only in live view making it not clear what it does when live view is not on but clearly cannot drop below the readout rate which again in low light can drop as well

Outside of the aperture, the camera doesn't run the sensor at the exposure you have dialed in, ie the shutter speed and ISO, even when you have LV configured to preview exposure. It instead uses a shutter speed and ISO that keeps an exposure within a range that's usable to the metering system, then simulates the shutter speed/ISO you have dialed in. Many systems also simulate the effect of the aperture on the previewed exposure, although like you indicated some actually stop the aperture down, at least to some degree, which is done for several reasons. One is to prevent focus shift. Another is to provide a DOF preview.

Even though the exposure you see in LV is simulated, most MILCs use that simulated exposure to feed the AF system since their imaging ASIC doesn't support dual pipelines, ie one preview for LV and another feeding the AF system. This is why the AF system is sensitive to your previewed exposure - it's not because the aperture, shutter speed, and ISO is preventing the sensor from getting enough (or too much) light.

The only exception to this is on Panasonic cameras, where it does have a mode that actually uses the dialed-in exposure parameter like shutter speed. But that mode is practically-speaking unusable because it creates a stuttered LV at slow shutter speeds that make viewing and manual focusing nearly impossible, which is why other cameras don't let you do this.

Panasonic lets you choose if you want to simulate aperture or aperture and shutter or nothing. It is very useful indeed. Sony only does that in video

Yes, that what I was referring to by "it does have a mode". I own several Panasonic cameras.

the point is that the sensor is scanned anyway at fixed rate af and exposure calculations are unrelated so the camera must do the af at the readout rate it is going

to that respect having the ability to focus wide open or at dialled exposure is useful

what am thinking is how to build some test cases for bright and dark scenes to validate what option to choose on the sony system indeed it will not make a difference as the aperture is the same live view or not

Not sure what you're referring to by "readout rate". If you mean the full-sensor readout time then yes, that caps the AF sample rate. That rate is unrelated to exposure. As I indicated, outside of the control of aperture, the AF system is fed by the previewed LV exposure you see, which is indirectly affected by the exposure you've set in the camera but is not the actual exposure the sensor is run at, for the reasons I listed in the previous message.

As for test cases to measure how different simulated exposures affect AF performance, I suggest observing the behavior of the continuous AF on a cross-hatched target at magnification, a technique I demonstrate in the following video

 

[Interceptor121]

Considering that the sensor is being scanned at a fixed rate and those are not pixels used for exposure I see benefits and disbenefits of both systems

In bright scenes it makes senses to close down the aperture to avoid overflow of light however in low light it seems more appropriate to focus wide open

Another parameter is the ISO value dialled in which does not seem to have an effect to the camera focus as if the camera was increasing gain in order to focus as much as possible and then applying the gain for the exposure

Finally the sensor readout can drop in low light but does not go faster in bright light the shutter speed is simulated only in live view making it not clear what it does when live view is not on but clearly cannot drop below the readout rate which again in low light can drop as well

Outside of the aperture, the camera doesn't run the sensor at the exposure you have dialed in, ie the shutter speed and ISO, even when you have LV configured to preview exposure. It instead uses a shutter speed and ISO that keeps an exposure within a range that's usable to the metering system, then simulates the shutter speed/ISO you have dialed in. Many systems also simulate the effect of the aperture on the previewed exposure, although like you indicated some actually stop the aperture down, at least to some degree, which is done for several reasons. One is to prevent focus shift. Another is to provide a DOF preview.

Even though the exposure you see in LV is simulated, most MILCs use that simulated exposure to feed the AF system since their imaging ASIC doesn't support dual pipelines, ie one preview for LV and another feeding the AF system. This is why the AF system is sensitive to your previewed exposure - it's not because the aperture, shutter speed, and ISO is preventing the sensor from getting enough (or too much) light.

The only exception to this is on Panasonic cameras, where it does have a mode that actually uses the dialed-in exposure parameter like shutter speed. But that mode is practically-speaking unusable because it creates a stuttered LV at slow shutter speeds that make viewing and manual focusing nearly impossible, which is why other cameras don't let you do this.

Panasonic lets you choose if you want to simulate aperture or aperture and shutter or nothing. It is very useful indeed. Sony only does that in video

Yes, that what I was referring to by "it does have a mode". I own several Panasonic cameras.

the point is that the sensor is scanned anyway at fixed rate af and exposure calculations are unrelated so the camera must do the af at the readout rate it is going

to that respect having the ability to focus wide open or at dialled exposure is useful

what am thinking is how to build some test cases for bright and dark scenes to validate what option to choose on the sony system indeed it will not make a difference as the aperture is the same live view or not

Not sure what you're referring to by "readout rate". If you mean the full-sensor readout time then yes, that caps the AF sample rate. That rate is unrelated to exposure. As I indicated, outside of the control of aperture, the AF system is fed by the previewed LV exposure you see, which is indirectly affected by the exposure you've set in the camera but is not the actual exposure the sensor is run at, for the reasons I listed in the previous message.

As for test cases to measure how different simulated exposures affect AF performance, I suggest observing the behavior of the continuous AF on a cross-hatched target at magnification, a technique I demonstrate in the following video

Interesting to see how different brands go after af implementation nikon even stops down the lens to a set value

in the case of Panasonic with constant preview off the camera is focussing wide open and with moving subjects at f/10 i have experienced occasional focus errors but I think this is due to the settings but generally works well

No correction required

however in bright light I have had issues perhaps here i should set aperture effect on to stop down the lens as I am experiencing the too much light situation when the camera focusses wide open

 

[jiberlin]

Phase detect AF works with rays from the outer areas of the lens, essential like a split-prism screen of old MF DSLR, If the aperture is closed, it is much more difficult to get a usable signal from PDAF, and above some f-numbers PDAF will not longer work (like the old split-prisms) and CDAF must used. That is (part of) the reason why cameras does not close the aperture full to the 'exposure-aperture' while focusing.

 

[Interceptor121]

Phase detect AF works with rays from the outer areas of the lens, essential like a split-prism screen of old MF DSLR, If the aperture is closed, it is much more difficult to get a usable signal from PDAF, and above some f-numbers PDAF will not longer work (like the old split-prisms) and CDAF must used. That is (part of) the reason why cameras does not close the aperture full to the 'exposure-aperture' while focusing.

PDAF on sensor doesnr use rhe same logic

having the lens wide open can cause focus shift

sony has the best autofocus on mirrorless and nikon does better than panasonic so the generalisation definitely doesnt work

Panasonic PDAF appears to address the fact mft cameras have generally more depth of field so your minimum f number is equivalent to f/5.6 in modt cases or slower

one should test f/1.4 primes to see if they focus shift
--
If you like my image I would appreciate if you follow me on social media
instagram http://instagram.com/interceptor121
My flickr sets http://www.flickr.com/photos/interceptor121/
Youtube channel http://www.youtube.com/interceptor121
Underwater Photo and Video Blog http://interceptor121.com
If you want to get in touch don't send me a PM rather contact me directly at my website/social media

 

[jiberlin]

Phase detect AF works with rays from the outer areas of the lens, essential like a split-prism screen of old MF DSLR, If the aperture is closed, it is much more difficult to get a usable signal from PDAF, and above some f-numbers PDAF will not longer work (like the old split-prisms) and CDAF must used. That is (part of) the reason why cameras does not close the aperture full to the 'exposure-aperture' while focusing.

PDAF on sensor doesnr use rhe same logic

??? What is the exact statement of this sentence?

All PDAF work with the same principals as the split prism screens.

The principle of PDAF is independent from the position of the PDAF sensor.

having the lens wide open can cause focus shift

Focus shift from closing aperture comes from optical deficits of the lens and is, as far as i know, minimized with modern high corrected lenses.

sony has the best autofocus on mirrorless and nikon does better than panasonic so the generalisation definitely doesnt work

Differences in AF performance have many reasons where the different aperture handling is only one part.

Nikon: As far as i know, Nikon Z series closes the aperture maximal to 5.6 for autofocus

Panasonic PDAF appears to address the fact mft cameras have generally more depth of field so your minimum f number is equivalent to f/5.6 in modt cases or slower

??? What is the exact statement of this sentence?

one should test f/1.4 primes to see if they focus shift

As i have written above, the amount of focus shift depends on the correction of the lens. Better correction = less focus shift.

 

[Interceptor121]

Phase detect AF works with rays from the outer areas of the lens, essential like a split-prism screen of old MF DSLR, If the aperture is closed, it is much more difficult to get a usable signal from PDAF, and above some f-numbers PDAF will not longer work (like the old split-prisms) and CDAF must used. That is (part of) the reason why cameras does not close the aperture full to the 'exposure-aperture' while focusing.

PDAF on sensor doesnr use rhe same logic

??? What is the exact statement of this sentence?

All PDAF work with the same principals as the split prism screens.

The principle of PDAF is independent from the position of the PDAF sensor.

having the lens wide open can cause focus shift

Focus shift from closing aperture comes from optical deficits of the lens and is, as far as i know, minimized with modern high corrected lenses.

sony has the best autofocus on mirrorless and nikon does better than panasonic so the generalisation definitely doesnt work

Differences in AF performance have many reasons where the different aperture handling is only one part.

Nikon: As far as i know, Nikon Z series closes the aperture maximal to 5.6 for autofocus

Panasonic PDAF appears to address the fact mft cameras have generally more depth of field so your minimum f number is equivalent to f/5.6 in modt cases or slower

??? What is the exact statement of this sentence?

one should test f/1.4 primes to see if they focus shift

As i have written above, the amount of focus shift depends on the correction of the lens. Better correction = less focus shift.

Pdaf on sensor doesn’t work like the prism of a dslr af system. You have non imaging pixels and no prism

I think Panasonic is concerned by the sensor not having enough light and therefore focuses wide open

the focus error may simply depend on the lag to move the aperture blades

the burst rate varies by lens however some lenses that have high speed seem to produce more focus errors

 

[jiberlin]

Phase detect AF works with rays from the outer areas of the lens, essential like a split-prism screen of old MF DSLR, If the aperture is closed, it is much more difficult to get a usable signal from PDAF, and above some f-numbers PDAF will not longer work (like the old split-prisms) and CDAF must used. That is (part of) the reason why cameras does not close the aperture full to the 'exposure-aperture' while focusing.

PDAF on sensor doesnr use rhe same logic

??? What is the exact statement of this sentence?

All PDAF work with the same principals as the split prism screens.

The principle of PDAF is independent from the position of the PDAF sensor.

having the lens wide open can cause focus shift

Focus shift from closing aperture comes from optical deficits of the lens and is, as far as i know, minimized with modern high corrected lenses.

sony has the best autofocus on mirrorless and nikon does better than panasonic so the generalisation definitely doesnt work

Differences in AF performance have many reasons where the different aperture handling is only one part.

Nikon: As far as i know, Nikon Z series closes the aperture maximal to 5.6 for autofocus

Panasonic PDAF appears to address the fact mft cameras have generally more depth of field so your minimum f number is equivalent to f/5.6 in modt cases or slower

??? What is the exact statement of this sentence?

one should test f/1.4 primes to see if they focus shift

As i have written above, the amount of focus shift depends on the correction of the lens. Better correction = less focus shift.

Pdaf on sensor doesn’t work like the prism of a dslr af system. You have non imaging pixels and no prism

It seems to me that you have no in-depth understanding of the PDAF..

PDAF works with the principle that rays from different areas of the lens converge in fron or behind the focus plane if the subject is not in focus. It the rays from left and right are split they hit the focus plane on different locations. With an old split prism screen the split prism is used to split the rays from different areas of the lens and the result is that the images of the two parts if arge shifted against each other. For on-sensor PDAF some pixels are partially masked (or have special micro lenses) in such a way that some pixel get only rays from part of the lens. The details differ between manufacturers, but the working principal is the same.

Simple geometry shows that the image shift on focal plane depends on the size of the aperture.

I think Panasonic is concerned by the sensor not having enough light and therefore focuses wide open

This statement is pure speculation.

the focus error may simply depend on the lag to move the aperture blades

How should this happen? In my view, it is a wrong explanation.

The 'true' explanation of focus shift from aperture is that a not perfect lens has different focal length on different areas (inner and outer). If the aperture is wide open, most of the picture is produced by rays from the outer area. If the aperture is closed the picture is only produced from the central area. If the focal length of inner and outer area (spherical aberration) are different this results in focus shift from closing the aperture.

the burst rate varies by lens however some lenses that have high speed seem to produce more focus errors

I can't follow this statement. In all cameras i know, the burst rate is limited by the image processing in the camera and is Independent of the used lens.

Fast lenses (big aperture) needed more precision for perfect focus, less then perfect focus is much more visible on fast lenses.

 

[Interceptor121]

Phase detect AF works with rays from the outer areas of the lens, essential like a split-prism screen of old MF DSLR, If the aperture is closed, it is much more difficult to get a usable signal from PDAF, and above some f-numbers PDAF will not longer work (like the old split-prisms) and CDAF must used. That is (part of) the reason why cameras does not close the aperture full to the 'exposure-aperture' while focusing.

PDAF on sensor doesnr use rhe same logic

??? What is the exact statement of this sentence?

All PDAF work with the same principals as the split prism screens.

The principle of PDAF is independent from the position of the PDAF sensor.

having the lens wide open can cause focus shift

Focus shift from closing aperture comes from optical deficits of the lens and is, as far as i know, minimized with modern high corrected lenses.

sony has the best autofocus on mirrorless and nikon does better than panasonic so the generalisation definitely doesnt work

Differences in AF performance have many reasons where the different aperture handling is only one part.

Nikon: As far as i know, Nikon Z series closes the aperture maximal to 5.6 for autofocus

Panasonic PDAF appears to address the fact mft cameras have generally more depth of field so your minimum f number is equivalent to f/5.6 in modt cases or slower

??? What is the exact statement of this sentence?

one should test f/1.4 primes to see if they focus shift

As i have written above, the amount of focus shift depends on the correction of the lens. Better correction = less focus shift.

Pdaf on sensor doesn’t work like the prism of a dslr af system. You have non imaging pixels and no prism

It seems to me that you have no in-depth understanding of the PDAF..

PDAF works with the principle that rays from different areas of the lens converge in fron or behind the focus plane if the subject is not in focus. It the rays from left and right are split they hit the focus plane on different locations. With an old split prism screen the split prism is used to split the rays from different areas of the lens and the result is that the images of the two parts if arge shifted against each other. For on-sensor PDAF some pixels are partially masked (or have special micro lenses) in such a way that some pixel get only rays from part of the lens. The details differ between manufacturers, but the working principal is the same.

How do you know do you have sources or this is just a guess?

Simple geometry shows that the image shift on focal plane depends on the size of the aperture.

I think Panasonic is concerned by the sensor not having enough light and therefore focuses wide open

This statement is pure speculation.

the focus error may simply depend on the lag to move the aperture blades

How should this happen? In my view, it is a wrong explanation.

The 'true' explanation of focus shift from aperture is that a not perfect lens has different focal length on different areas (inner and outer). If the aperture is wide open, most of the picture is produced by rays from the outer area. If the aperture is closed the picture is only produced from the central area. If the focal length of inner and outer area (spherical aberration) are different this results in focus shift from closing the aperture.

This is not generally true for moving subjects if the focus is not fast enough and the subject has moved you get an out of focus shot. I am not shooting a fixed target in CAF why would I do it?

the burst rate varies by lens however some lenses that have high speed seem to produce more focus errors

I can't follow this statement. In all cameras i know, the burst rate is limited by the image processing in the camera and is Independent of the used lens.

Fast lenses (big aperture) needed more precision for perfect focus, less then perfect focus is much more visible on fast lenses.

I have done some tests it varies by lens of a factor of 30% in the same situation

The lenses am trying are f/2.8 they are not fast

 

[kenw]

Phase detect AF works with rays from the outer areas of the lens, essential like a split-prism screen of old MF DSLR, If the aperture is closed, it is much more difficult to get a usable signal from PDAF, and above some f-numbers PDAF will not longer work (like the old split-prisms) and CDAF must used. That is (part of) the reason why cameras does not close the aperture full to the 'exposure-aperture' while focusing.

PDAF on sensor doesnr use rhe same logic

??? What is the exact statement of this sentence?

All PDAF work with the same principals as the split prism screens.

The principle of PDAF is independent from the position of the PDAF sensor.

having the lens wide open can cause focus shift

Focus shift from closing aperture comes from optical deficits of the lens and is, as far as i know, minimized with modern high corrected lenses.

sony has the best autofocus on mirrorless and nikon does better than panasonic so the generalisation definitely doesnt work

Differences in AF performance have many reasons where the different aperture handling is only one part.

Nikon: As far as i know, Nikon Z series closes the aperture maximal to 5.6 for autofocus

Panasonic PDAF appears to address the fact mft cameras have generally more depth of field so your minimum f number is equivalent to f/5.6 in modt cases or slower

??? What is the exact statement of this sentence?

one should test f/1.4 primes to see if they focus shift

As i have written above, the amount of focus shift depends on the correction of the lens. Better correction = less focus shift.

Pdaf on sensor doesn’t work like the prism of a dslr af system. You have non imaging pixels and no prism

It seems to me that you have no in-depth understanding of the PDAF..

PDAF works with the principle that rays from different areas of the lens converge in fron or behind the focus plane if the subject is not in focus. It the rays from left and right are split they hit the focus plane on different locations. With an old split prism screen the split prism is used to split the rays from different areas of the lens and the result is that the images of the two parts if arge shifted against each other. For on-sensor PDAF some pixels are partially masked (or have special micro lenses) in such a way that some pixel get only rays from part of the lens. The details differ between manufacturers, but the working principal is the same.

How do you know do you have sources or this is just a guess?

Google it. There are endless papers that describe how on sensor PDAF functions and its direct relationship with split prism AF sensors on DSLRs as well as visual split prism screens.

 

[jiberlin]

...

PDAF works with the principle that rays from different areas of the lens converge in fron or behind the focus plane if the subject is not in focus. It the rays from left and right are split they hit the focus plane on different locations. With an old split prism screen the split prism is used to split the rays from different areas of the lens and the result is that the images of the two parts if arge shifted against each other. For on-sensor PDAF some pixels are partially masked (or have special micro lenses) in such a way that some pixel get only rays from part of the lens. The details differ between manufacturers, but the working principal is the same.

How do you know do you have sources or this is just a guess?

As user kenw has written, there are plenty information on internet about this topic. Google is/could be your friend.

...

The 'true' explanation of focus shift from aperture is that a not perfect lens has different focal length on different areas (inner and outer). If the aperture is wide open, most of the picture is produced by rays from the outer area. If the aperture is closed the picture is only produced from the central area. If the focal length of inner and outer area (spherical aberration) are different this results in focus shift from closing the aperture.

This is not generally true for moving subjects if the focus is not fast enough and the subject has moved you get an out of focus shot. I am not shooting a fixed target in CAF why would I do it?

Change of focus from moving objects is not what in general called 'focus shift of lenses'. Again, ask the internet, and you will find answers (even i there is risk to find misinformation).

Holding a moving object in-focus is a very difficult job for the camera AF system, and only the top models master it partly.

the burst rate varies by lens however some lenses that have high speed seem to produce more focus errors

I can't follow this statement. In all cameras i know, the burst rate is limited by the image processing in the camera and is Independent of the used lens.

Fast lenses (big aperture) needed more precision for perfect focus, less then perfect focus is much more visible on fast lenses.

I have done some tests it varies by lens of a factor of 30% in the same situation

The lenses am trying are f/2.8 they are not fast

It seems to me that your original statement about burst rate and high speed lenses have a very different intention then the common sense.

 

[Interceptor121]

...

PDAF works with the principle that rays from different areas of the lens converge in fron or behind the focus plane if the subject is not in focus. It the rays from left and right are split they hit the focus plane on different locations. With an old split prism screen the split prism is used to split the rays from different areas of the lens and the result is that the images of the two parts if arge shifted against each other. For on-sensor PDAF some pixels are partially masked (or have special micro lenses) in such a way that some pixel get only rays from part of the lens. The details differ between manufacturers, but the working principal is the same.

How do you know do you have sources or this is just a guess?

As user kenw has written, there are plenty information on internet about this topic. Google is/could be your friend.

...

The 'true' explanation of focus shift from aperture is that a not perfect lens has different focal length on different areas (inner and outer). If the aperture is wide open, most of the picture is produced by rays from the outer area. If the aperture is closed the picture is only produced from the central area. If the focal length of inner and outer area (spherical aberration) are different this results in focus shift from closing the aperture.

This is not generally true for moving subjects if the focus is not fast enough and the subject has moved you get an out of focus shot. I am not shooting a fixed target in CAF why would I do it?

Change of focus from moving objects is not what in general called 'focus shift of lenses'. Again, ask the internet, and you will find answers (even i there is risk to find misinformation).

Holding a moving object in-focus is a very difficult job for the camera AF system, and only the top models master it partly.

the burst rate varies by lens however some lenses that have high speed seem to produce more focus errors

I can't follow this statement. In all cameras i know, the burst rate is limited by the image processing in the camera and is Independent of the used lens.

Fast lenses (big aperture) needed more precision for perfect focus, less then perfect focus is much more visible on fast lenses.

I have done some tests it varies by lens of a factor of 30% in the same situation

The lenses am trying are f/2.8 they are not fast

It seems to me that your original statement about burst rate and high speed lenses have a very different intention then the common sense.

High speed lenses i.e. fast not fast to focus

I found some materials on line regarding pixel size and lenses. Most modern sensor have all pixel autofocus so no micro lenses

However I still do not see how this system would differ other than being more precise as it does not need assembly or tuning of mechanical parts

Now with that in mind I see that some systems are far more complex than others Sony for example does not focus always at the dialled aperture and has a quite complex algorithm of what it does so it is not that easy

panasonic wide open is indeed quite primitive but I guess it does not need adapting. It could have issues when there is too much light and the pixels somehow are overwhelmed

 

[SKersting66]

One thing I haven't seen mentioned (directly) is that for the mirrorless camera to focus it must be able to discern the image(s) being used.

In a DSLR's PDAF system there is a separate optical path with multiple real images that have a very large depth of focus. This causes those images to almost always be (nearly) in focus, which makes them almost always possible to use for focus.
And that separate optical path also has its' own aperture limit, which causes those images to be unaffected by the lens aperture in use (unless too restrictive).

With the mirrorless that is not the case and the AF system is dependent on the (virtual) image(s) as seen in the viewfinder. So if you cannot readily discern a detail, the AF system cannot either... i.e. there is nothing for the AF system to focus on. So using a smaller aperture increases the DOFocus; increasing the usability of the image(s) and effectivity of the AF system.
I believe this is also why mirrorless will generally default to farther distances when there is nothing readily available to focus on... because DOFocus greatly increases and focus throw greatly reduces (and probability as well).

Likewise, the image(s) must be bright enough for the AF system to see (discern details); so a wider aperture may be more useful. I.e. a mirrorless camera's low light focus capability rating includes a lens aperture setting (e.g. f/1.2); that is not the case for a DSLR.

--
https://www.flickr.com/skersting

 

[mosswings]

One thing I haven't seen mentioned (directly) is that for the mirrorless camera to focus it must be able to discern the image(s) being used.

In a DSLR's PDAF system there is a separate optical path with multiple real images that have a very large depth of focus. This causes those images to almost always be (nearly) in focus, which makes them almost always possible to use for focus.
And that separate optical path also has its' own aperture limit, which causes those images to be unaffected by the lens aperture in use (unless too restrictive).

With the mirrorless that is not the case and the AF system is dependent on the (virtual) image(s) as seen in the viewfinder. So if you cannot readily discern a detail, the AF system cannot either... i.e. there is nothing for the AF system to focus on. So using a smaller aperture increases the DOFocus; increasing the usability of the image(s) and effectivity of the AF system.
I believe this is also why mirrorless will generally default to farther distances when there is nothing readily available to focus on... because DOFocus greatly increases and focus throw greatly reduces (and probability as well).

Likewise, the image(s) must be bright enough for the AF system to see (discern details); so a wider aperture may be more useful. I.e. a mirrorless camera's low light focus capability rating includes a lens aperture setting (e.g. f/1.2); that is not the case for a DSLR.

Yes, and it's also been mentioned before in other venues that OSPDAF systems don't enjoy the separation of the image pairs that DSLR PDAF provides. Thus, not only are the image pairs experiencing strong changes in sharpness as focus is changed, but also lie almost on top of each other. The result is poor focus discrimination near optimal focus. DSLR PDAF, by contrast, always has image pairs that are relatively focused and non-interfering.

 

[jiberlin]

One thing I haven't seen mentioned (directly) is that for the mirrorless camera to focus it must be able to discern the image(s) being used.

In a DSLR's PDAF system there is a separate optical path with multiple real images that have a very large depth of focus. This causes those images to almost always be (nearly) in focus, which makes them almost always possible to use for focus.
And that separate optical path also has its' own aperture limit, which causes those images to be unaffected by the lens aperture in use (unless too restrictive).

It seems to me, there is a misconception about the AF system of (D)SLR cameras. The only differences between the optical path for image, viewfinder and AF sensor are the mirrors. The main mirror used for the VF is semi transparent and direct the light to the matt screen. Behind this main mirror are (a) second mirror(s) that direct the light from the lens to the AF sensor on the bottom of the mirror housing. Only one optical path that is split in two directions.

All (D)SLR i know hold the aperture wide open until exposure. The AF measurement is done with wide open aperture. If you have a (D)SLR you can check it by yourself.

With the mirrorless that is not the case and the AF system is dependent on the (virtual) image(s) as seen in the viewfinder. So if you cannot readily discern a detail, the AF system cannot either... i.e. there is nothing for the AF system to focus on. So using a smaller aperture increases the DOFocus; increasing the usability of the image(s) and effectivity of the AF system.

As written above, the AF system in (D)SLR works with aperture wide open, and the AF sensor gets the same image as the matt screen.

I believe this is also why mirrorless will generally default to farther distances when there is nothing readily available to focus on... because DOFocus greatly increases and focus throw greatly reduces (and probability as well).

Likewise, the image(s) must be bright enough for the AF system to see (discern details); so a wider aperture may be more useful. I.e. a mirrorless camera's low light focus capability rating includes a lens aperture setting (e.g. f/1.2); that is not the case for a DSLR.

As i have written previous in this thread, the PDAF works better with aperture wide open. That is the reason why MILC close the aperture only partly until exposure.

 

[jiberlin]

...

Yes, and it's also been mentioned before in other venues that OSPDAF systems don't enjoy the separation of the image pairs that DSLR PDAF provides. Thus, not only are the image pairs experiencing strong changes in sharpness as focus is changed, but also lie almost on top of each other. The result is poor focus discrimination near optimal focus. DSLR PDAF, by contrast, always has image pairs that are relatively focused and non-interfering.

As i and 'kenw' have written, the only principle difference between PDAF in SLR and MILC is the position of the sensor. A (D)SLR has the sensor on the bottom of the mirror housing and a MILC (ab)uses the sensor for PDAF. A minor difference could be the resolution of the different sensors.

 

[jiberlin]

I found some materials on line regarding pixel size and lenses. Most modern sensor have all pixel autofocus so no micro lenses

Most (all) modern sensors have micro lenses, one for each pixel. How the PDAF work differs between manufacturers. Canon uses dual-pixels while most other use some type of masking. And one manufacturer (i do not remember the name) uses quad-pixel.

The all-pixel AF is an advertisement claim, in reality, it is full area AF. At least the PDAF systems that use masking use only a part of the sensor pixels of the PDAF system.

However I still do not see how this system would differ other than being more precise as it does not need assembly or tuning of mechanical parts

Now with that in mind I see that some systems are far more complex than others Sony for example does not focus always at the dialled aperture and has a quite complex algorithm of what it does so it is not that easy

As written before, closing the aperture only partial for PDAF is because PDAF works better with wider apertures because of the geometric relations in the optical part.

The basic PDAF is comparative simple, what makes it complex is the selection which parts of the image should used, and if the subject is moving, to predict the movement.

panasonic wide open is indeed quite primitive but I guess it does not need adapting. It could have issues when there is too much light and the pixels somehow are overwhelmed

The PDAF system gets the same data as the live view (back-LCD or EVF). If the live view is not over exposed (overhelmed), the PDAF system is also not 'overhelmed'.

 

[SKersting66]

It seems to me, there is a misconception about the AF system of (D)SLR cameras.

Maybe there is...

All (D)SLR i know hold the aperture wide open until exposure. The AF measurement is done with wide open aperture. If you have a (D)SLR you can check it by yourself.

The PDAF module has its' own separator lenses which dictate the maximum aperture for the PDAF system. And it is necessarily smaller than the attached lens' max aperture (or the AF system/points will be non-functional).

This is very similar as to how/why the viewfinder also has its' own maximum aperture (~ f/2).

Three optical paths is more accurate...

As written above, the AF system in (D)SLR works with aperture wide open, and the AF sensor gets the same image as the matt screen.

Not necessarily. The AF system can function perfectly well with a manual aperture lens stopped down (but not much farther than the focus point rating).

And the SLR's PDAF system does not get "the same image" as viewed in the viewfinder. Instead it gets multiple images (4 for a cross type point) and those are only the segments of interest from the viewfinder image.

As i have written previous in this thread, the PDAF works better with aperture wide open. That is the reason why MILC close the aperture only partly until exposure.

It is a tradeoff; if it were that simple all mirrorless cameras would default to focusing wide open, and they do not.

The primary reason SLRs focus wide open is that most use the same mechanism to open the mirror and stop down the aperture mechanically (or used to). And also because stopping down significantly impacts the viewfinder image brightness and the ability to focus manually. Mirrorless cameras do not have these limitations, so they can choose to make different tradeoffs.

--
https://www.flickr.com/skersting