fredmiranda.com
Login

  

  Previous versions of MichaelACG's message #16213315 « A way to think about Subject Detection »

  

MichaelACG
Offline
Upload & Sell: Off
Re: A way to think about Subject Detection


morris wrote:
...
When we start focus the camera uses the traditional AF system. The Subject Detection system is looking for a subject. As soon as Subject Detection finds the subject it moves the focus selection area as if the photographer had moved the joystick to the subject. This is all that Subject Detection does, it does not focus the camera as the traditional Fuji AF system focuses the camera.
...


Thank you, Morris.

This was the explanation, helping me to understand, why and what went wrong in my bird detection attempts:
I thought that the small rectangle(s), showing that and where a bird (bird's eye) is detected by the bird detection, additionally reduce the focus area. This was a wrong assumption and Morris' explanation helped me to understand, what bird detection can do and what not.

Birds are really another challenge to focus, much more difficult than the other objects: smaller, erratically (and fast) moving, sometimes uneven background. And long lenses - making DOF and movements even more critical.

Morris' explanation is not in Fuji's manual (so - in a sense, we all are kind of doing a dis-assembling exercise ;-), but it can be easily proven. In the following examples the focus area was in the middle, and bird detection moved it onto the bird:

1. Blackbird on the ground:
You see the 3 white rectangles: 1: bird, 2: head, 3:eye. So bird detection worked perfectly.
You see the red focus zone (the circle indicates "zone"), it was a 7x7 zone.
And this 7x7 was my mistake: way too large, so rightfully the focus was somewhere, not on the bird.
https://photos.google.com/share/AF1QipOZJ3pFUV45PQanAq5trYZd4o85oJ5ExBo2Y3y_hqn_o0tCOJGXEy7TAo8dY--T8Q/photo/AF1QipMzYnBO2Wd4g9yPc79WHSPVOem8lcN14YwOb9U_?key=QjJaTXhXcXM3YVlpWF9seFg3bHNic1ZXSzVUZ2NR


2. My bird model on a fence, background looks nicely blurred.
Again bird detection worked well, this time the focus point was moved onto the eye.
But ... I (intentionally) use the largest focus point (size 6), to illustrate that this sometime is also a problem (like zone).
The X-H2S focused to the fence, slightly (2 cm) behind the bird's feet. I had 600m focal length, f/8, 5 m distance, so only around 3 cm DOF. The backfocus of 2 cm was enough to blur the bird' head.
https://photos.app.goo.gl/C3vBp5Cq1KuksaEt6

3. And here the solution for situations like this: Use the smallest focus point (size 1 in this case).
https://photos.app.goo.gl/WW58kJ7taS1eTXjKA

(First time that I try to link images here - I hope, it's working).

Finally after reading Morris' threads on bird detection, using his AF-C custom settings, many tests, a very fruitful (and nice :-) exchange of thoughts with him ... my go-to settings for birds are:



  1. Electronic shutter: very important
  2. AF-C: to follow both bird's movements and compensate my movements, when hand-holding. Remember: long lens!
  3. Try to keep the bird in the focus area. This is a tricky trade-off: Zone makes it easier than point, but you get the focus problems in busy backgrounds.
  4. As Morris always says: Try to have a nice background, do not shoot birds in busy backgrounds. But - often you don't have a choice.
  5. For (relatively) static birds, I use single-shot, and AF/MF=on with focus check and MF assist. Focus area as small as I manage to "catch" the bird, and MF override in critical situations. If the bird is really static, I even change to AF-S, because here the MF assist is a joy to use because of the automatic enlargement, when you touch the focus ring.
  6. BIF (birds in flight): multi-shot (I use 30 fps) and - important - AF/MF=off (otherwise the camera switches to Release Priority). I use focus zone, again as small as possible.
  7. In general: the smaller the focus zone the better detection works (and autofocus), but the more problems to follow the bird.
  8. And, aperture, ISO, time to properly expose and freeze the bird's movement.


Hmmm - embedding of linked images seems not to work. So you have to click the links.



Apr 08, 2023 at 12:55 AM
MichaelACG
Offline
Upload & Sell: Off
Re: A way to think about Subject Detection


morris wrote:
...
When we start focus the camera uses the traditional AF system. The Subject Detection system is looking for a subject. As soon as Subject Detection finds the subject it moves the focus selection area as if the photographer had moved the joystick to the subject. This is all that Subject Detection does, it does not focus the camera as the traditional Fuji AF system focuses the camera.
...


Thank you, Morris.

This was the explanation, helping me to understand, why and what went wrong in my bird detection attempts:
I thought that the small rectangle(s), showing that and where a bird (bird's eye) is detected by the bird detection, additionally reduce the focus area. This was a wrong assumption and Morris' explanation helped me to understand, what bird detection can do and what not.

Birds are really another challenge to focus, much more difficult than the other objects: smaller, erratically (and fast) moving, sometimes uneven background. And long lenses - making DOF and movements even more critical.

Morris' explanation is not in Fuji's manual (so - in a sense, we all are kind of doing a dis-assembling exercise ;-), but it can be easily proven. In the following examples the focus area was in the middle, and bird detection moved it onto the bird:

1. Blackbird on the ground:
You see the 3 white rectangles: 1: bird, 2: head, 3:eye. So bird detection worked perfectly.
You see the red focus zone (the circle indicates "zone"), it was a 7x7 zone.
And this 7x7 was my mistake: way too large, so rightfully the focus was somewhere, not on the bird.
https://photos.google.com/share/AF1QipOZJ3pFUV45PQanAq5trYZd4o85oJ5ExBo2Y3y_hqn_o0tCOJGXEy7TAo8dY--T8Q/photo/AF1QipMzYnBO2Wd4g9yPc79WHSPVOem8lcN14YwOb9U_?key=QjJaTXhXcXM3YVlpWF9seFg3bHNic1ZXSzVUZ2NR

2. My bird model on a fence, background looks nicely blurred.
Again bird detection worked well, this time the focus point was moved onto the eye.
But ... I (intentionally) use the largest focus point (size 6), to illustrate that this sometime is also a problem (like zone).
The X-H2S focused to the fence, slightly (2 cm) behind the bird's feet. I had 600m focal length, f/8, 5 m distance, so only around 3 cm DOF. The backfocus of 2 cm was enough to blur the bird' head.
https://photos.app.goo.gl/C3vBp5Cq1KuksaEt6

3. And here the solution for situations like this: Use the smallest focus point (size 1 in this case).
https://photos.app.goo.gl/WW58kJ7taS1eTXjKA

(First time that I try to link images here - I hope, it's working).

Finally after reading Morris' threads on bird detection, using his AF-C custom settings, many tests, a very fruitful (and nice :-) exchange of thoughts with him ... my go-to settings for birds are:



  1. Electronic shutter: very important
  2. AF-C: to follow both bird's movements and compensate my movements, when hand-holding. Remember: long lens!
  3. Try to keep the bird in the focus area. This is a tricky trade-off: Zone makes it easier than point, but you get the focus problems in busy backgrounds.
  4. As Morris always says: Try to have a nice background, do not shoot birds in busy backgrounds. But - often you don't have a choice.
  5. For (relatively) static birds, I use single-shot, and AF/MF=on with focus check and MF assist. Focus area as small as I manage to "catch" the bird, and MF override in critical situations. If the bird is really static, I even change to AF-S, because here the MF assist is a joy to use because of the automatic enlargement, when you touch the focus ring.
  6. BIF (birds in flight): multi-shot (I use 30 fps) and - important - AF/MF=off (otherwise the camera switches to Release Priority). I use focus zone, again as small as possible.
  7. In general: the smaller the focus zone the better detection works (and autofocus), but the more problems to follow the bird.
  8. And, aperture, ISO, time to properly expose and freeze the bird's movement.


Hmmm - embedding of linked images seems not to work. So you have to click the links.



Apr 08, 2023 at 12:54 AM
MichaelACG
Offline
Upload & Sell: Off
Re: A way to think about Subject Detection


morris wrote:
...
When we start focus the camera uses the traditional AF system. The Subject Detection system is looking for a subject. As soon as Subject Detection finds the subject it moves the focus selection area as if the photographer had moved the joystick to the subject. This is all that Subject Detection does, it does not focus the camera as the traditional Fuji AF system focuses the camera.
...


Thank you, Morris.

This was the explanation, helping me to understand, why and what went wrong in my bird detection attempts:
I thought that the small rectangle(s), showing that and where a bird (bird's eye) is detected by the bird detection, additionally reduce the focus area. This was a wrong assumption and Morris' explanation helped me to understand, what bird detection can do and what not.

Birds are really another challenge to focus, much more difficult than the other objects: smaller, erratically (and fast) moving, sometimes uneven background. And long lenses - making DOF and movements even more critical.

Morris' explanation is not in Fuji's manual (so - in a sense, we all are kind of doing a dis-assembling exercise ;-), but it can be easily proven. In the following examples the focus area was in the middle, and bird detection moved it onto the bird:

1. Blackbird on the ground:
You see the 3 white rectangles: 1: bird, 2: head, 3:eye. So bird detection worked perfectly.
You see the red focus zone (the circle indicates "zone"), it was a 7x7 zone.
And this 7x7 was my mistake: way too large, so rightfully the focus was somewhere, not on the bird.

2. My bird model on a fence, background looks nicely blurred.
Again bird detection worked well, this time the focus point was moved onto the eye.
But ... I (intentionally) use the largest focus point (size 6), to illustrate that this sometime is also a problem (like zone).
The X-H2S focused to the fence, slightly (2 cm) behind the bird's feet. I had 600m focal length, f/8, 5 m distance, so only around 3 cm DOF. The backfocus of 2 cm was enough to blur the bird' head.
https://photos.app.goo.gl/C3vBp5Cq1KuksaEt6

3. And here the solution for situations like this: Use the smallest focus point (size 1 in this case).
https://photos.app.goo.gl/WW58kJ7taS1eTXjKA

(First time that I try to link images here - I hope, it's working).

Finally after reading Morris' threads on bird detection, using his AF-C custom settings, many tests, a very fruitful (and nice :-) exchange of thoughts with him ... my go-to settings for birds are:



  1. Electronic shutter: very important
  2. AF-C: to follow both bird's movements and compensate my movements, when hand-holding. Remember: long lens!
  3. Try to keep the bird in the focus area. This is a tricky trade-off: Zone makes it easier than point, but you get the focus problems in busy backgrounds.
  4. As Morris always says: Try to have a nice background, do not shoot birds in busy backgrounds. But - often you don't have a choice.
  5. For (relatively) static birds, I use single-shot, and AF/MF=on with focus check and MF assist. Focus area as small as I manage to "catch" the bird, and MF override in critical situations. If the bird is really static, I even change to AF-S, because here the MF assist is a joy to use because of the automatic enlargement, when you touch the focus ring.
  6. BIF (birds in flight): multi-shot (I use 30 fps) and - important - AF/MF=off (otherwise the camera switches to Release Priority). I use focus zone, again as small as possible.
  7. In general: the smaller the focus zone the better detection works (and autofocus), but the more problems to follow the bird.
  8. And, aperture, ISO, time to properly expose and freeze the bird's movement.


Hmmm - embedding of linked images seems not to work. So you have to click the links.



Apr 08, 2023 at 12:54 AM
MichaelACG
Offline
Upload & Sell: Off
Re: A way to think about Subject Detection


morris wrote:
...
When we start focus the camera uses the traditional AF system. The Subject Detection system is looking for a subject. As soon as Subject Detection finds the subject it moves the focus selection area as if the photographer had moved the joystick to the subject. This is all that Subject Detection does, it does not focus the camera as the traditional Fuji AF system focuses the camera.
...


Thank you, Morris.

This was the explanation, helping me to understand, why and what went wrong in my bird detection attempts:
I thought that the small rectangle(s), showing that and where a bird (bird's eye) is detected by the bird detection, additionally reduce the focus area. This was a wrong assumption and Morris' explanation helped me to understand, what bird detection can do and what not.

Birds are really another challenge to focus, much more difficult than the other objects: smaller, erratically (and fast) moving, sometimes uneven background. And long lenses - making DOF and movements even more critical.

Morris' explanation is not in Fuji's manual (so - in a sense, we all are kind of doing a dis-assembling exercise ;-), but it can be easily proven. In the following examples the focus area was in the middle, and bird detection moved it onto the bird:

1. Blackbird on the ground:
You see the 3 white rectangles: 1: bird, 2: head, 3:eye. So bird detection worked perfectly.
You see the red focus zone (the circle indicates "zone"), it was a 7x7 zone.
And this 7x7 was my mistake: way too large, so rightfully the focus was somewhere, not on the bird.
https://photos.google.com/share/AF1QipOZJ3pFUV45PQanAq5trYZd4o85oJ5ExBo2Y3y_hqn_o0tCOJGXEy7TAo8dY--T8Q/photo/AF1QipMzYnBO2Wd4g9yPc79WHSPVOem8lcN14YwOb9U_?key=QjJaTXhXcXM3YVlpWF9seFg3bHNic1ZXSzVUZ2NR

2. My bird model on a fence, background looks nicely blurred.
Again bird detection worked well, this time the focus point was moved onto the eye.
But ... I (intentionally) use the largest focus point (size 6), to illustrate that this sometime is also a problem (like zone).
The X-H2S focused to the fence, slightly (2 cm) behind the bird's feet. I had 600m focal length, f/8, 5 m distance, so only around 3 cm DOF. The backfocus of 2 cm was enough to blur the bird' head.
https://photos.app.goo.gl/C3vBp5Cq1KuksaEt6

3. And here the solution for situations like this: Use the smallest focus point (size 1 in this case).
https://photos.app.goo.gl/WW58kJ7taS1eTXjKA

(First time that I try to link images here - I hope, it's working).

Finally after reading Morris' threads on bird detection, using his AF-C custom settings, many tests, a very fruitful (and nice :-) exchange of thoughts with him ... my go-to settings for birds are:



  1. Electronic shutter: very important
  2. AF-C: to follow both bird's movements and compensate my movements, when hand-holding. Remember: long lens!
  3. Try to keep the bird in the focus area. This is a tricky trade-off: Zone makes it easier than point, but you get the focus problems in busy backgrounds.
  4. As Morris always says: Try to have a nice background, do not shoot birds in busy backgrounds. But - often you don't have a choice.
  5. For (relatively) static birds, I use single-shot, and AF/MF=on with focus check and MF assist. Focus area as small as I manage to "catch" the bird, and MF override in critical situations. If the bird is really static, I even change to AF-S, because here the MF assist is a joy to use because of the automatic enlargement, when you touch the focus ring.
  6. BIF (birds in flight): multi-shot (I use 30 fps) and - important - AF/MF=off (otherwise the camera switches to Release Priority). I use focus zone, again as small as possible.
  7. In general: the smaller the focus zone the better detection works (and autofocus), but the more problems to follow the bird.
  8. And, aperture, ISO, time to properly expose and freeze the bird's movement.


Hmmm - embedding of linked images seems not to work. So you have to click the links.



Apr 08, 2023 at 12:49 AM
MichaelACG
Offline
Upload & Sell: Off
Re: A way to think about Subject Detection


morris wrote:
...
When we start focus the camera uses the traditional AF system. The Subject Detection system is looking for a subject. As soon as Subject Detection finds the subject it moves the focus selection area as if the photographer had moved the joystick to the subject. This is all that Subject Detection does, it does not focus the camera as the traditional Fuji AF system focuses the camera.
...


Thank you, Morris.

This was the explanation, helping me to understand, why and what went wrong in my bird detection attempts:
I thought that the small rectangle(s), showing that and where a bird (bird's eye) is detected by the bird detection, additionally reduce the focus area. This was a wrong assumption and Morris' explanation helped me to understand, what bird detection can do and what not.

Birds are really another challenge to focus, much more difficult than the other objects: smaller, erratically (and fast) moving, sometimes uneven background. And long lenses - making DOF and movements even more critical.

Morris' explanation is not in Fuji's manual (so - in a sense, we all are kind of doing a dis-assembling exercise ;-), but it can be easily proven. In the following examples the focus area was in the middle, and bird detection moved it onto the bird:

1. Blackbird on the ground:
You see the 3 white rectangles: 1: bird, 2: head, 3:eye. So bird detection worked perfectly.
You see the red focus zone (the circle indicates "zone"), it was a 7x7 zone.
And this 7x7 was my mistake: way too large, so rightfully the focus was somewhere, not on the bird.
https://photos.google.com/share/AF1QipOZJ3pFUV45PQanAq5trYZd4o85oJ5ExBo2Y3y_hqn_o0tCOJGXEy7TAo8dY--T8Q/photo/AF1QipMzYnBO2Wd4g9yPc79WHSPVOem8lcN14YwOb9U_?key=QjJaTXhXcXM3YVlpWF9seFg3bHNic1ZXSzVUZ2NR

2. My bird model on a fence, background looks nicely blurred.
Again bird detection worked well, this time the focus point was moved onto the eye.
But ... I (intentionally) use the largest focus point (size 6), to illustrate that this sometime is also a problem (like zone).
The X-H2S focused to the fence, slightly (2 cm) behind the bird's feet. I had 600m focal length, f/8, 5 m distance, so only around 3 cm DOF. The backfocus of 2 cm was enough to blur the bird' head.
https://photos.app.goo.gl/C3vBp5Cq1KuksaEt6

3. And here the solution for situations like this: Use the smallest focus point (size 1 in this case).
https://photos.app.goo.gl/WW58kJ7taS1eTXjKA

(First time that I try to link images here - I hope, it's working).

Finally after reading Morris' threads on bird detection, using his AF-C custom settings, many tests, a very fruitful (and nice :-) exchange of thoughts with him ... my go-to settings for birds are:



  1. Electronic shutter: very important
  2. AF-C: to follow both bird's movements and compensate my movements, when hand-holding. Remember: long lens!
  3. Try to keep the bird in the focus area. This is a tricky trade-off: Zone makes it easier than point, but you get the focus problems in busy backgrounds.
  4. As Morris always says: Try to have a nice background, do not shoot birds in busy backgrounds. But - often you don't have a choice.
  5. For (relatively) static birds, I use single-shot, and AF/MF=on with focus check and MF assist. Focus area as small as I manage to "catch" the bird, and MF override in critical situations. If the bird is really static, I even change to AF-S, because here the MF assist is a joy to use because of the automatic enlargement, when you touch the focus ring.
  6. BIF (birds in flight): multi-shot (I use 30 fps) and - important - AF/MF=off (otherwise the camera switches to Release Priority). I use focus zone, again as small as possible.
  7. In general: the smaller the focus zone the better detection works (and autofocus), but the more problems to follow the bird.
  8. And, aperture, ISO, time to properly expose and freeze the bird's movement.


Hmmm - embedding of linked images seems not to work. So you have to click the links.



Apr 07, 2023 at 03:58 PM





  Previous versions of MichaelACG's message #16213315 « A way to think about Subject Detection »