My last Deer shot with Panasonic GHM2 using autofocus (who needs PDAF lol)

Status
Not open for further replies.

Interceptor121

Forum Pro
Messages
12,604
Solutions
8
Reaction score
9,603
I am expecting the usual suspect to spend some time finding some fault. Well if there is one is my error. Note the bird flying in slow motion...

Just got to 500 subscribers despite zero promotion effort feel free to join

--
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
Deer Photography workshops https://interceptor121.com/2021/09/26/2021-22-deer-photography-workshops-in-woburn/
 
Last edited by a moderator:
Very nice video. Thanks for posting.

However, there is little change in distance, so almost any focus method would work. I would be more convinced I didn't need PDAF if they were running quickly toward the camera and focus was maintained.
 
Very nice video. Thanks for posting.

However, there is little change in distance, so almost any focus method would work. I would be more convinced I didn't need PDAF if they were running quickly toward the camera and focus was maintained.
If they're running quickly towards the camera you wouldn't be worried about focus systems, you'd be worried about puncture wounds.
 
Very nice video. Thanks for posting.

However, there is little change in distance, so almost any focus method would work. I would be more convinced I didn't need PDAF if they were running quickly toward the camera and focus was maintained.
 
Very nice video. Thanks for posting.

However, there is little change in distance, so almost any focus method would work. I would be more convinced I didn't need PDAF if they were running quickly toward the camera and focus was maintained.
Hahahahahahahahahahahahahaa

will see if they run me over next time

try to do a pan of a bird in video with your camera and let me know how it goes
Probably not very well since I haven't used my E-M1s for video. However, panning a bird for still photos is fairly easy with my E-M1s and impossible for my E-M10.1 with CDAF.

The GHM2 may be very good with autofocus for moving targets that change distance, I don't know, since I have not tried it.

However, the deer nor flying bird changed distance by much more than the DOF. My post was simply a response to the "who needs PDAF" and you are correct, I would not need PDAF for the deer shots, but I might if they were running at speed and changing distance.
what an absolute joke classic dpreview
 
Very nice video. Thanks for posting.

However, there is little change in distance, so almost any focus method would work. I would be more convinced I didn't need PDAF if they were running quickly toward the camera and focus was maintained.
If they're running quickly towards the camera you wouldn't be worried about focus systems, you'd be worried about puncture wounds.
Definitely, if there was a doe in estrus behind me. My White tailed deer completely ignore me, though they sometimes look irritated if I try to mow my field and get within 5 feet of them.

I do get somewhat more nervous when one of the Black bear does a bluff charge when it sees me, but then I know enough not to run.
 
They are beautiful animals.

Rather than being ‘distracted’ by the action I kept watching the body of one of the deer and I noticed that it was going slightly in and out of focus - it would suddenly get sharp and I could see more details and then that sharpness was lost and then it would come sharp again - almost as if the camera is ever so slightly hunting to find focus. Is that normal with the GHM2 ? (I have never used a Lumix camera)

As for PDAF and video . . . your camera doesn’t have a PDAF capability so I am no sure why you mention it. In any case, I thought that even the Olympus cameras that do have PDAF don’t use it during video recording.

With action such as that where the subjects are at some distance and not changing range I would rather use MF taking advantage of the DoF, and probably use focus peaking. AF would have the risk that the camera may decide to focus on something ‘unwanted’

Peter
 
Last edited:
It is hard to manual focus when you have a camera on a monopod without a focus gear because this also creates shake which is worse than the ever so slight dfd vibe

having said that while I see it on the background as when the lenses refocuses it moves due to focus breathing I really don’t see it on the animals and I use a 65” Tv. So if you tell me where you saw it I can have a look at what happened.

The point is many users go about issues that in real life do not exist. I cant think if many situations where you would want to focus on something coming at you at speed and keep focus

and you are correct in video the camera processes the whole frame not just a low resolution frame at high frame rate as it does with photos so it will have more issues as you are shooting 30/60 fps and in those there would be some slightly off

Having said that I compared my footage with many other sources and it beats the crap out of anyone out there including lumix ambassadors and bbc and the reason is the fieldcraft
 
the colors look weird
 
Maybe your monitor is off
 
Cold clear dawn, frost on the grass. Nothing unusual for a winter morning.
 
Cold clear dawn, frost on the grass. Nothing unusual for a winter morning.
Correct. There is a retouch but only on highlight shadows not on midtones so it does not rotate the hues or anything like that
 
In any case, I thought that even the Olympus cameras that do have PDAF don’t use it during video recording.
I don't think this is true. Haven't researched it extensively but if you take a look at the video below starting at the 5:47 mark, he has comparison of PDAF and CDAF in video recording. A short time later, he points out that on the E-M1 Mark II it was CDAF only in video until firmware 3.0.
 
In any case, I thought that even the Olympus cameras that do have PDAF don’t use it during video recording.
I don't think this is true. Haven't researched it extensively but if you take a look at the video below starting at the 5:47 mark, he has comparison of PDAF and CDAF in video recording. A short time later, he points out that on the E-M1 Mark II it was CDAF only in video until firmware 3.0.
The only difference between CDAF or PDAF is how the camera obtains focus which may generate a little wobble if it repeats the operation too many time

All autofocus systems go through iterations and when it comes to video what matters the most is how many frames per second are you reading (the slow camera in this video has a readout of 30 fps) and how quickly you process it before you can tell autofocus what to do

Once you decide you need to refocus then this hands over to the autofocus to do or not do something about it

When you look at tracking the key part of it is as the word says tracking which is not to do with autofocus but with the ability of recognizing a pattern in a frame

This is the reason why Olympus camera in video when you put them in face detect with tracking at the end do not do that well. This is well demonstrated by dpreview of the EM1X where they say it works fine however the guy is walking very slow and on a straight line, when you slow down the frames they are all slightly off

The point is how much off until is completely off

Panasonic has made steps in detection and this works very well it has some residual issues with their DFD that produces a wobble, but this wobble is mostly visible on the background not on the subject and typically nobody looks at the background unless you have a guy that is talking for 20 minutes and you start seeing the background pulse

that pulse is non existent when the camera does not use depth from defocus so as a paradox you use a sigma or olympus lens on a panasonic camera and you dont get the pulse although it may not focus well at all
 
[No message]
 
GH5M2 somehow went missing
 
In any case, I thought that even the Olympus cameras that do have PDAF don’t use it during video recording.
I don't think this is true. Haven't researched it extensively but if you take a look at the video below starting at the 5:47 mark, he has comparison of PDAF and CDAF in video recording. A short time later, he points out that on the E-M1 Mark II it was CDAF only in video until firmware 3.0.
The only difference between CDAF or PDAF is how the camera obtains focus which may generate a little wobble if it repeats the operation too many time
The primary difference is that PDAF can detect that the subject is out of focus and can tell which direction (closer or farther) and how far the focusing is off.

CDAF can tell the subject is out of focus, but it must guess the direct and amount and then try again.

https://www.dpreview.com/interviews...-olympus-engineer-talks-om-d-e-m1-mark-iii-af

Modern mirrorless interchangeable lens cameras use one, or a combination of two main types of autofocus: contrast-detection and phase-detection. Contrast-detection autofocus works by driving focus until the contrast of a sampled area on the sensor is at its maximum - the presumed point of sharp focus. Contrast-detection is highly accurate, but can be slow, and relies on a certain amount of 'trial and error'.

Phase-detection works more like human vision, using dedicated pixels to compare light coming from your subject from two slightly different perspectives at the same time. This allows the camera to judge depth, allowing for faster focus acquisition, without the 'hunting' characteristic of many purely contrast-detection autofocus systems.

Traditional phase-detection autofocus systems rely on pixels that are sensitive to vertical lines in a scene. Some, more sophisticated systems use a 'cross-type' pixel arrangement, which can detect horizontal or vertical detail, meaning that accurate focus can be achieved even with complex, non-linear subjects.
 
In any case, I thought that even the Olympus cameras that do have PDAF don’t use it during video recording.
I don't think this is true. Haven't researched it extensively but if you take a look at the video below starting at the 5:47 mark, he has comparison of PDAF and CDAF in video recording. A short time later, he points out that on the E-M1 Mark II it was CDAF only in video until firmware 3.0.
The only difference between CDAF or PDAF is how the camera obtains focus which may generate a little wobble if it repeats the operation too many time
The primary difference is that PDAF can detect that the subject is out of focus and can tell which direction (closer or farther) and how far the focusing is off.

CDAF can tell the subject is out of focus, but it must guess the direct and amount and then try again.

https://www.dpreview.com/interviews...-olympus-engineer-talks-om-d-e-m1-mark-iii-af

Modern mirrorless interchangeable lens cameras use one, or a combination of two main types of autofocus: contrast-detection and phase-detection. Contrast-detection autofocus works by driving focus until the contrast of a sampled area on the sensor is at its maximum - the presumed point of sharp focus. Contrast-detection is highly accurate, but can be slow, and relies on a certain amount of 'trial and error'.

Phase-detection works more like human vision, using dedicated pixels to compare light coming from your subject from two slightly different perspectives at the same time. This allows the camera to judge depth, allowing for faster focus acquisition, without the 'hunting' characteristic of many purely contrast-detection autofocus systems.

Traditional phase-detection autofocus systems rely on pixels that are sensitive to vertical lines in a scene. Some, more sophisticated systems use a 'cross-type' pixel arrangement, which can detect horizontal or vertical detail, meaning that accurate focus can be achieved even with complex, non-linear subjects.
This is general theory for autofocus of a single still image, in that case the camera works on a lower resolution but higher frame rate image and uses a small subset of phase difference pixels. This method of course means it is quicker but less accurate.

The key challenge is that in video the camera is not reading a lower resolution frame at higher frame rate, is reading the whole sensor at the speed it is shooting or best case at the sensor readout speed and dropping frames.

When it is reading this information AND decided that it needs to refocus then it will perform it but the key is to understand how frequently it is reading and how much it needs to adjust.

When you activate tracking it becomes even slower as it is not using anymore phase difference but is first looking at what to track and then checking if this is still in focus assuming it did not loose the tracking

It is a whole worlds of difference when you need to write 30 or 60 frames per second to have them all in focus and the camera is working in increments so if you are on track to start with the gap between the various methods is considerably smaller. When something moves away of focus considerably it may be that phase detection can catch up quicker but your video sequence is ruined and you need to restart anyway. Is not like photos where you just drop 20 frames to use 1 or 2 you need to use all of it
 
In any case, I thought that even the Olympus cameras that do have PDAF don’t use it during video recording.
I don't think this is true. Haven't researched it extensively but if you take a look at the video below starting at the 5:47 mark, he has comparison of PDAF and CDAF in video recording. A short time later, he points out that on the E-M1 Mark II it was CDAF only in video until firmware 3.0.
The only difference between CDAF or PDAF is how the camera obtains focus which may generate a little wobble if it repeats the operation too many time
The primary difference is that PDAF can detect that the subject is out of focus and can tell which direction (closer or farther) and how far the focusing is off.

CDAF can tell the subject is out of focus, but it must guess the direct and amount and then try again.

https://www.dpreview.com/interviews...-olympus-engineer-talks-om-d-e-m1-mark-iii-af

Modern mirrorless interchangeable lens cameras use one, or a combination of two main types of autofocus: contrast-detection and phase-detection. Contrast-detection autofocus works by driving focus until the contrast of a sampled area on the sensor is at its maximum - the presumed point of sharp focus. Contrast-detection is highly accurate, but can be slow, and relies on a certain amount of 'trial and error'.

Phase-detection works more like human vision, using dedicated pixels to compare light coming from your subject from two slightly different perspectives at the same time. This allows the camera to judge depth, allowing for faster focus acquisition, without the 'hunting' characteristic of many purely contrast-detection autofocus systems.

Traditional phase-detection autofocus systems rely on pixels that are sensitive to vertical lines in a scene. Some, more sophisticated systems use a 'cross-type' pixel arrangement, which can detect horizontal or vertical detail, meaning that accurate focus can be achieved even with complex, non-linear subjects.
This is general theory for autofocus of a single still image, in that case the camera works on a lower resolution but higher frame rate image and uses a small subset of phase difference pixels. This method of course means it is quicker but less accurate.

The key challenge is that in video the camera is not reading a lower resolution frame at higher frame rate, is reading the whole sensor at the speed it is shooting or best case at the sensor readout speed and dropping frames.

When it is reading this information AND decided that it needs to refocus then it will perform it but the key is to understand how frequently it is reading and how much it needs to adjust.

When you activate tracking it becomes even slower as it is not using anymore phase difference but is first looking at what to track and then checking if this is still in focus assuming it did not loose the tracking

It is a whole worlds of difference when you need to write 30 or 60 frames per second to have them all in focus and the camera is working in increments so if you are on track to start with the gap between the various methods is considerably smaller. When something moves away of focus considerably it may be that phase detection can catch up quicker but your video sequence is ruined and you need to restart anyway. Is not like photos where you just drop 20 frames to use 1 or 2 you need to use all of it
I remain convinced that PDAF is superior for my uses for continuous AF with subjects moving towards or away from the camera. I shot youth soccer for years and could not have done if without PDAF.

If CDAF works your you in your scenarios, great. But back to your original question about PDAF - "Who needs it?" I do.
 
In any case, I thought that even the Olympus cameras that do have PDAF don’t use it during video recording.
I don't think this is true. Haven't researched it extensively but if you take a look at the video below starting at the 5:47 mark, he has comparison of PDAF and CDAF in video recording. A short time later, he points out that on the E-M1 Mark II it was CDAF only in video until firmware 3.0.
The only difference between CDAF or PDAF is how the camera obtains focus which may generate a little wobble if it repeats the operation too many time

All autofocus systems go through iterations and when it comes to video what matters the most is how many frames per second are you reading (the slow camera in this video has a readout of 30 fps) and how quickly you process it before you can tell autofocus what to do

Once you decide you need to refocus then this hands over to the autofocus to do or not do something about it

When you look at tracking the key part of it is as the word says tracking which is not to do with autofocus but with the ability of recognizing a pattern in a frame

This is the reason why Olympus camera in video when you put them in face detect with tracking at the end do not do that well. This is well demonstrated by dpreview of the EM1X where they say it works fine however the guy is walking very slow and on a straight line, when you slow down the frames they are all slightly off

The point is how much off until is completely off

Panasonic has made steps in detection and this works very well it has some residual issues with their DFD that produces a wobble, but this wobble is mostly visible on the background not on the subject and typically nobody looks at the background unless you have a guy that is talking for 20 minutes and you start seeing the background pulse

that pulse is non existent when the camera does not use depth from defocus so as a paradox you use a sigma or olympus lens on a panasonic camera and you dont get the pulse although it may not focus well at all
That's interesting, and is confirmed using my Oly 12-100/4.0 on the G9.
 
Status
Not open for further replies.

Keyboard shortcuts

Back
Top