A glimpse at automated AFMA from Nikon

[ahsanford]

Now showing on the D5 and D500, a feature we might like:

http://www.dpreview.com/news/3468248279/nikons-automated-af-fine-tune-explained

...but the implementation seems very crude and simple. Play the video in the link. Yes, it's better than having to futz around with an incrementally marked target setup, but it appears to only work at one focus distance.

Here's hoping that Canon can do better.

- A

 

[d]

Wouldn't be too difficult to better. I'm surprised there isn't the option for the camera to automatically repeat the process several times and then take the average value.

Multiple distances is a must, likewise multiple focal lengths for zooms. And since we're going to the trouble, can we also make specific calibration values for outer AF points as well, please?

d.

P.S. His tripod didn't seem particularly stable.

 

[Loswr]

I like the concept. Didn't actually see it in operation, but if it takes a single shot with CDAF and a single shot with PDAF as the basis, I doubt the accuracy and precision.

Having AFMA'd a whole bunch of lenses, I find many zooms differ across the range – for example, my 24-70/2.8 II is zero at one end and +5 at the other, that's a difference of over half the depth of focus. For me, the ability to calibrate a zoom lens at both ends is important, and trumps automation. But I'd prefer both, of course.

 

[Alex_M]

I used the Live View mode to calibrate my lenses for a while now and I found this method t be very accurate and neat.
I set the target at approx. x50 of the lens focal length to the camera on a steady tripod, switch to the Live View mode, half press the shutter button to acquire focus. Take the note of the distance indicator value on the lens. no need to take a photo. Switch to OVF mode (PDAF) and press the shutter button half way again. pay very close attention to the distance indicator readings of your lens as you do.

1. if the distance indicator readings moved towards infinity (even slightly) - your lens has BACK FOCUSED. adjust the AFMA value in camera. rinse and repeat.

2. if the distance indicator readings moved towards the Minimum Focus Distance (even slightly) - your lens has FRONT FOCUSED. adjust the AFMA value in camera. rinse and repeat.

3. If distance indicator readings has not changed at all, then the camera PDAF system is in agreement with the CDAF system that the subject was already in focus. NO need to adjust AFMA in camera settings. Rinse and repeat 2 times for consistency.

benefit of this method:

1. no need to take photos at all. saves you hundreds of shutter actuation and your valuable time.
2. critical focus image quality evaluation hard to impossible on camera LCD screen, especially outdoors.
3. no computer, LCD screen required.

this method can be used for a quick AF accuracy verification for critical applications.

I found that this method can be used with great success outdoors focusing on any static contrasty subject.

 

[kphoto99]

I used the Live View mode to calibrate my lenses for a while now and I found this method t be very accurate and neat.
I set the target at approx. x50 of the lens focal length to the camera on a steady tripod, switch to the Live View mode, half press the shutter button to acquire focus. Take the note of the distance indicator value on the lens. no need to take a photo. Switch to OVF mode (PDAF) and press the shutter button half way again. pay very close attention to the distance indicator readings of your lens as you do.

1. if the distance indicator readings moved towards infinity (even slightly) - your lens has BACK FOCUSED. adjust the AFMA value in camera. rinse and repeat.

2. if the distance indicator readings moved towards the Minimum Focus Distance (even slightly) - your lens has FRONT FOCUSED. adjust the AFMA value in camera. rinse and repeat.

3. If distance indicator readings has not changed at all, then the camera PDAF system is in agreement with the CDAF system that the subject was already in focus. NO need to adjust AFMA in camera settings. Rinse and repeat 2 times for consistency.

benefit of this method:

1. no need to take photos at all. saves you hundreds of shutter actuation and your valuable time.
2. critical focus image quality evaluation hard to impossible on camera LCD screen, especially outdoors.
3. no computer, LCD screen required.

this method can be used for a quick AF accuracy verification for critical applications.

I found that this method can be used with great success outdoors focusing on any static contrasty subject.

Doesn't work with lenses without distance scale (all STM lenses). For a better method google "dot tune".
I have the FoCal software and I have given up using it in favour of the dot tune method.

 

[AlanF]

There are several methods of doing AFMA badly. I suppose the process of "rinse" at least has the advantage of washing the camera but should be done only with weather-sealed lenses.

 

[Alex_M]

Sir, I can confirm that the method I described above is quite accurate as I was able to double and triple check the results on numerous occasions with Reikan FoCal software and also using traditional Canon recommended technique. there is also 3-Target DoF based technique that I use for the most critical AFMA.

quote author=AlanF link=topic=29663.msg592205#msg592205 date=1461302410]
There are several methods of doing AFMA badly. I suppose the process of "rinse" at least has the advantage of washing the camera but should be done only with weather-sealed lenses.
[/quote]

 

[AlanF]

Sir, I can confirm that the method I described above is quite accurate as I was able to double and triple check the results on numerous occasions with Reikan FoCal software and also using traditional Canon recommended technique. there is also 3-Target DoF based technique that I use for the most critical AFMA.

quote author=AlanF link=topic=29663.msg592205#msg592205 date=1461302410]
There are several methods of doing AFMA badly. I suppose the process of "rinse" at least has the advantage of washing the camera but should be done only with weather-sealed lenses.

[/quote]

I don't understand what "rinse" is?

 

[rfdesigner]

I used the Live View mode to calibrate my lenses for a while now and I found this method t be very accurate and neat.
I set the target at approx. x50 of the lens focal length to the camera on a steady tripod, switch to the Live View mode, half press the shutter button to acquire focus. Take the note of the distance indicator value on the lens. no need to take a photo. Switch to OVF mode (PDAF) and press the shutter button half way again. pay very close attention to the distance indicator readings of your lens as you do.

1. if the distance indicator readings moved towards infinity (even slightly) - your lens has BACK FOCUSED. adjust the AFMA value in camera. rinse and repeat.

2. if the distance indicator readings moved towards the Minimum Focus Distance (even slightly) - your lens has FRONT FOCUSED. adjust the AFMA value in camera. rinse and repeat.

3. If distance indicator readings has not changed at all, then the camera PDAF system is in agreement with the CDAF system that the subject was already in focus. NO need to adjust AFMA in camera settings. Rinse and repeat 2 times for consistency.

benefit of this method:

1. no need to take photos at all. saves you hundreds of shutter actuation and your valuable time.
2. critical focus image quality evaluation hard to impossible on camera LCD screen, especially outdoors.
3. no computer, LCD screen required.

this method can be used for a quick AF accuracy verification for critical applications.

I found that this method can be used with great success outdoors focusing on any static contrasty subject.

Doesn't work with lenses without distance scale (all STM lenses). For a better method google "dot tune".
I have the FoCal software and I have given up using it in favour of the dot tune method.

I was aware of the DotTune method prior to getting my 6D.. having got it, it was trivial to do and all my AF issues (well except speed and any randomness and STM being focus by wire) are out the window.

Magic Lantern even has it automated, not that I've tried ML yet. The only difference I did compared to the official DotTune version was I used a successive approximation method to find the end point of the "in focus range", which made the process somewhat quicker.

My Test target was just a black and white test card I printed off and stuck on the wall.

 

[ichiru]

I used the Live View mode to calibrate my lenses for a while now and I found this method t be very accurate and neat.
I set the target at approx. x50 of the lens focal length to the camera on a steady tripod, switch to the Live View mode, half press the shutter button to acquire focus. Take the note of the distance indicator value on the lens. no need to take a photo. Switch to OVF mode (PDAF) and press the shutter button half way again. pay very close attention to the distance indicator readings of your lens as you do.

1. if the distance indicator readings moved towards infinity (even slightly) - your lens has BACK FOCUSED. adjust the AFMA value in camera. rinse and repeat.

2. if the distance indicator readings moved towards the Minimum Focus Distance (even slightly) - your lens has FRONT FOCUSED. adjust the AFMA value in camera. rinse and repeat.

3. If distance indicator readings has not changed at all, then the camera PDAF system is in agreement with the CDAF system that the subject was already in focus. NO need to adjust AFMA in camera settings. Rinse and repeat 2 times for consistency.

benefit of this method:

1. no need to take photos at all. saves you hundreds of shutter actuation and your valuable time.
2. critical focus image quality evaluation hard to impossible on camera LCD screen, especially outdoors.
3. no computer, LCD screen required.

this method can be used for a quick AF accuracy verification for critical applications.

I found that this method can be used with great success outdoors focusing on any static contrasty subject.

Doesn't work with lenses without distance scale (all STM lenses). For a better method google "dot tune".
I have the FoCal software and I have given up using it in favour of the dot tune method.

I was aware of the DotTune method prior to getting my 6D.. having got it, it was trivial to do and all my AF issues (well except speed and any randomness and STM being focus by wire) are out the window.

Magic Lantern even has it automated, not that I've tried ML yet. The only difference I did compared to the official DotTune version was I used a successive approximation method to find the end point of the "in focus range", which made the process somewhat quicker.

My Test target was just a black and white test card I printed off and stuck on the wall.

I tried the magic lantern automated dot tune with my 6D with both tungsten lights and natural light (sideways window light) and although I would leave the camera in the same place... I would get significantly different values each time I repeated the process... any ideas? What type of lighting did you use? Any tips are welcome

At this stage I think taking a Live View mode shot and comparing it with different MFA values back and forth is the best option as described by 'Alex_M'... FoCal is expensive and I don't know that I can even use it with my Sigma lens...

 

[rfdesigner]

I tried the magic lantern automated dot tune with my 6D with both tungsten lights and natural light (sideways window light) and although I would leave the camera in the same place... I would get significantly different values each time I repeated the process... any ideas? What type of lighting did you use? Any tips are welcome

At this stage I think taking a Live View mode shot and comparing it with different MFA values back and forth is the best option as described by 'Alex_M'... FoCal is expensive and I don't know that I can even use it with my Sigma lens...

Well when I did it I was under CFL lighting, but it's 100% indirect lighting, no spots or anything so it's niegh on impossible to find a shadow (no idea if this helped, but that's how it was). I placed the camera on a tripod (I have a big old heavy video one which is great for this) at about the same distance I like to shoot at most for each lens (different distance for each lens)

I AF in liveview on my B&W target (koren 2003 lens test chart) Then turn the lens from AF to MF and turn off liveview, and then check the AF confirmation. I then dither the AF microadjust +/-10 and check again, then +/-5 check again (but not both ways, just one way looking for the edges of AF confirm) +/-2, check again, then +/-1 homing in on the two points where the AF confrim is flickering, rather than being fully on or off.

Any automated system is going to struggle with the flickering and is likely to come up with a different number each time, but it should be similar. The main problem I had was that one lens had one AF end beyond the end of the microadjust range, so I guessed the breadth of the AF range from the other lenses and estimated it's centre point from just one end of the AF microadjust, but it seems to have worked.

I don't know if that's helpful or "too much" but it's what I did, pretty much word for word. The advantage of using something like a Koren lens test chart is you can then pop off a couple of shots before and after and check the resoltion.. but I also checked against a horizonatal tape measure too.

 

[Zeidora]

Thanks for the two alternate option to check AF-lens combos. I bought FocusTune, and found it utterly useless. Complete waste of $200. Will throw it now in the recycling, will not even try to burden anybody else.

 

[Refurb7]

At this stage I think taking a Live View mode shot and comparing it with different MFA values back and forth is the best option as described by 'Alex_M'... FoCal is expensive and I don't know that I can even use it with my Sigma lens...

FoCal works fine with Sigma lenses. I've used it with the 35 and 50 Art lenses and it worked very well.

I hope Canon introduces an automated system like Nikon's.

 

[Woody]

I used the Magic Lantern automated AF cal function on 6D and was happy with the results.

And yes, AF cal under tungsten and natural light give different results.

Will like to have the Magic Lantern function incorporated into the actual Canon firmware. Right now, Magic Lantern only works for older cameras.

 

[ichiru]

Thanks for the two alternate option to check AF-lens combos. I bought FocusTune, and found it utterly useless. Complete waste of $200. Will throw it now in the recycling, will not even try to burden anybody else.

Thanks for the feedback on FocusTune... sorry to hear you wasted your money. Did you try FoCal ?

 

[ichiru]

Well when I did it I was under CFL lighting, but it's 100% indirect lighting, no spots or anything so it's niegh on impossible to find a shadow (no idea if this helped, but that's how it was). I placed the camera on a tripod (I have a big old heavy video one which is great for this) at about the same distance I like to shoot at most for each lens (different distance for each lens)

I AF in liveview on my B&W target (koren 2003 lens test chart) Then turn the lens from AF to MF and turn off liveview, and then check the AF confirmation. I then dither the AF microadjust +/-10 and check again, then +/-5 check again (but not both ways, just one way looking for the edges of AF confirm) +/-2, check again, then +/-1 homing in on the two points where the AF confrim is flickering, rather than being fully on or off.

Any automated system is going to struggle with the flickering and is likely to come up with a different number each time, but it should be similar. The main problem I had was that one lens had one AF end beyond the end of the microadjust range, so I guessed the breadth of the AF range from the other lenses and estimated it's centre point from just one end of the AF microadjust, but it seems to have worked.

I don't know if that's helpful or "too much" but it's what I did, pretty much word for word. The advantage of using something like a Koren lens test chart is you can then pop off a couple of shots before and after and check the resoltion.. but I also checked against a horizonatal tape measure too.
[/quote]

Thanks for sharing your experience. I do the focus tune with Magic Lantern I figure it's even more 'precise' as I don't move the camera changing settings... it's all automatic. I knew it wasn't working simply because each time I was redoing it, it was giving me a different value! Lmao.

I'll try it again today. I'll go grab some CFL I guess 5500 K would be best? My printed focusing chart my be crappy too... apparently laser ink has some reflection and as such I'll try inkjet. How would you suggest going about 'indirect light'? bouching off an umbrella would be less harsh but still direct...

Thanks again!

 

[ichiru]

FoCal works fine with Sigma lenses. I've used it with the 35 and 50 Art lenses and it worked very well.

I hope Canon introduces an automated system like Nikon's.
[/quote]

I guess you translate the Canon MFA value unto the Sigma dock by multiplying by 2?

 

[rfdesigner]

Well when I did it I was under CFL lighting, but it's 100% indirect lighting, no spots or anything so it's niegh on impossible to find a shadow (no idea if this helped, but that's how it was). I placed the camera on a tripod (I have a big old heavy video one which is great for this) at about the same distance I like to shoot at most for each lens (different distance for each lens)

I AF in liveview on my B&W target (koren 2003 lens test chart) Then turn the lens from AF to MF and turn off liveview, and then check the AF confirmation. I then dither the AF microadjust +/-10 and check again, then +/-5 check again (but not both ways, just one way looking for the edges of AF confirm) +/-2, check again, then +/-1 homing in on the two points where the AF confrim is flickering, rather than being fully on or off.

Any automated system is going to struggle with the flickering and is likely to come up with a different number each time, but it should be similar. The main problem I had was that one lens had one AF end beyond the end of the microadjust range, so I guessed the breadth of the AF range from the other lenses and estimated it's centre point from just one end of the AF microadjust, but it seems to have worked.

I don't know if that's helpful or "too much" but it's what I did, pretty much word for word. The advantage of using something like a Koren lens test chart is you can then pop off a couple of shots before and after and check the resoltion.. but I also checked against a horizonatal tape measure too.

Thanks for sharing your experience. I do the focus tune with Magic Lantern I figure it's even more 'precise' as I don't move the camera changing settings... it's all automatic. I knew it wasn't working simply because each time I was redoing it, it was giving me a different value! Lmao.

I'll try it again today. I'll go grab some CFL I guess 5500 K would be best? My printed focusing chart my be crappy too... apparently laser ink has some reflection and as such I'll try inkjet. How would you suggest going about 'indirect light'? bouching off an umbrella would be less harsh but still direct...

Thanks again!

If I were going to choose a light source I'd go for DC powered tungsten (halogen), again indirectly lit. I believe CFLs will flicker at 100Hz here or 120Hz state-side (unless they have a high frequency converter, maybe some do and some don't perhaps someone can fill in this blank), they are all spectral, Halogens are much broader emission band, only emitting light due to a hot filament, the flicker is also lower.

 

[ichiru]

If I were going to choose a light source I'd go for DC powered tungsten (halogen), again indirectly lit. I believe CFLs will flicker at 100Hz here or 120Hz state-side (unless they have a high frequency converter, maybe some do and some don't perhaps someone can fill in this blank), they are all spectral, Halogens are much broader emission band, only emitting light due to a hot filament, the flicker is also lower.
[/quote]

Hmmm only issue is I would rather calibrate for daylight colour (5600K) as that is the lighting I'm most likely to focus under... and colour seems to affect autofocus (http://www.reikan.co.uk/focalweb/index.php/2014/01/colour-affect-autofocus/)

 

[rfdesigner]

If I were going to choose a light source I'd go for DC powered tungsten (halogen), again indirectly lit. I believe CFLs will flicker at 100Hz here or 120Hz state-side (unless they have a high frequency converter, maybe some do and some don't perhaps someone can fill in this blank), they are all spectral, Halogens are much broader emission band, only emitting light due to a hot filament, the flicker is also lower.

Hmmm only issue is I would rather calibrate for daylight colour (5600K) as that is the lighting I'm most likely to focus under... and colour seems to affect autofocus (http://www.reikan.co.uk/focalweb/index.php/2014/01/colour-affect-autofocus/)


I was just concerned that if you had a flickering light is might throw things off.

How about daylight?.. it's the lightsource most like daylight!