Picture control and autofocus

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

I have recently been hearing from some people that setting a neutral picture control helps with mirrorless cameras autofocus as they use contrast as part of the locking on. Mainly with smaller subjects and busier background. Anyone play around with this? Results? I’m going to give it a shot this weekend.
 
Picture controls definitely impact what is seen in the EVF, I can't say that I've seen a noticeable difference in AF performance but then again I might not have had the right conditions. I'm still of the mindset that most of my problems are due to the computer 4 inches behind the EVF than the computer inside the camera.

That isn't to say that picture control settings won't help, I can easily see how they could help, just I'm not able to measure an improvement with my VERY limited testing
 
I can rather report the opposite. I had always used a very flat self-made profile and the camera struggled to acquire the focus. Folks in Forum said that camera get feed from IVF. In any case I changed profile to neutral and it works good now.
Earlier with DSLR we used self-made very flat profile to get more dynamic range but now I don't know if it is stil the case.
 
I have recently been hearing from some people that setting a neutral picture control helps with mirrorless cameras autofocus as they use contrast as part of the locking on. Mainly with smaller subjects and busier background. Anyone play around with this? Results? I’m going to give it a shot this weekend.
It does not make sense to me. A neutral picture control decreases contrast and saturation compared to Standard. That can make a better starting point for editing and for video that will be post processed, but not focus. Focus needs more contrast - not less. So Neutral is better than Flat, and Standard is usually better than Neutral. But it depends on the subject and lighting as to whether Neutral is a problem setting or the ideal setting. On a bright, sunny day I would use Neutral since you have ample contrast, but with lower contrast and overcast skies, Standard would likely be better.

I have heard that the scene recognition aspect of matrix metering is used by the camera to help with subject detection, and that color is still used as part of subject tracking. Both of those are ways to improve focus. With a small subject and/or busy background, it's helpful to have a subject that is larger in the frame or at least use a small AF area like Custom 1x1 or Wide Small on the Z9. There are different choices in different cameras.
 
I think the conversations have been specific to the z9, where there is a dual stream feeding the autofocus system and that is influenced by the brightness of the evf.

The other area where picture controls come into play is the histogram and the blinkies. If you are trying to get the raw close to the right edge or as high as possible without blowouts, the picture controls impacts that. A contrasty picture control pushes the histogram more to the right, making you think you are out of room, while the neutral picture control shows a little more headroom.

It's better to use a consistent picture control then study how much headroom you have in the raw. On my camera I know if I use Faithful then I have 2/3 stops more in raw than the highest non-blinkie exposure, so I get to know what the blinkies and the histogram looks like at that point, just climbing the right wall a little.
 
This discussion is challenging what I thought I knew about RAW files and in camera processing.

The RAW file contains the raw data from the sensor and the camera settings when the sensor was read. The settings have not been applied to the data but many image processing applications will apply the camera settings to the RAW data when imported. I know when working in LrC, I can easily change the camera profile if necessary.

I can't understand how changing the profiles would change the focusing speed or accuracy. If that were the case wouldn't the sensor data have to be processed and the camera settings be applied to the data before the focus processing occurs in camera? I would think that in order to improve focus accuracy the data would pass directly from the sensor to the focus engine.

Just puzzling to me!
 
This discussion is challenging what I thought I knew about RAW files and in camera processing.

The RAW file contains the raw data from the sensor and the camera settings when the sensor was read. The settings have not been applied to the data but many image processing applications will apply the camera settings to the RAW data when imported. I know when working in LrC, I can easily change the camera profile if necessary.

I can't understand how changing the profiles would change the focusing speed or accuracy. If that were the case wouldn't the sensor data have to be processed and the camera settings be applied to the data before the focus processing occurs in camera? I would think that in order to improve focus accuracy the data would pass directly from the sensor to the focus engine.

Just puzzling to me!
Hi Marty - Here's my understanding.

The data that is used for AF is a separate stream that is also used for the EVF. That data stream is not a raw file - it's a limited data stream with the Picture Control choice applied. The dual readout allows a separate stream of limited size to be extracted from the sensor and quickly processed for the EVF and for focus. This is similar to the reason that exposure is important or increasing ISO for a brighter scene can improve AF.

So while a bit unexpected, it does affect focus. How much difference it makes depends on the scene and the choice of picture control. For example, if you use a low contrast picture control for a low contrast scene, it can make AF less effective and subject detection could be harder.
 
Yes, this is a thing. There have been a few threads about this on this forum if you want to do some research. Here's a summary:
  • PDAF and CDAF sensors focus by contrast. All cameras focus poorly when there is little contrast available
  • It appears that some mirrorless cameras focus off "raw" sensor image, and are unaffected by this phenomena
  • Some other mirrorless cameras focus off the "final" image that you see in your EVF, and are affected by this phenomena
  • It's not so much Picture Control that affects autofocus on these cameras. It's the image in the EVF/LCD, and Picture Control just happens to be one thing that can highly affect the EVF image (try the "Carbon" Picture Control setting on a Z camera for an example!)
  • The Nikon Z6/Z7 have been demonstrated to have their AF affected by the EVF/LCD image
  • The Nikon Z9 has been claimed to have this behavior, but this has never been demonstrated
  • Personally, I've demonstrated to my own satisfaction that the Z9 does not exhibit this behavior
 
This discussion is challenging what I thought I knew about RAW files and in camera processing.

The RAW file contains the raw data from the sensor and the camera settings when the sensor was read. The settings have not been applied to the data but many image processing applications will apply the camera settings to the RAW data when imported. I know when working in LrC, I can easily change the camera profile if necessary.

I can't understand how changing the profiles would change the focusing speed or accuracy. If that were the case wouldn't the sensor data have to be processed and the camera settings be applied to the data before the focus processing occurs in camera? I would think that in order to improve focus accuracy the data would pass directly from the sensor to the focus engine.

Just puzzling to me!

I think the dual stream thing only applies to the Z cameras. I know canon doesn't do it and I don't think Sony does. For non Z cameras I think it is as you said. The live histogram and the blinkies are influenced by the picture control, but not focus.
 
Yes, this is a thing. There have been a few threads about this on this forum if you want to do some research. Here's a summary:
  • PDAF and CDAF sensors focus by contrast. All cameras focus poorly when there is little contrast available
  • It appears that some mirrorless cameras focus off "raw" sensor image, and are unaffected by this phenomena
  • Some other mirrorless cameras focus off the "final" image that you see in your EVF, and are affected by this phenomena
  • It's not so much Picture Control that affects autofocus on these cameras. It's the image in the EVF/LCD, and Picture Control just happens to be one thing that can highly affect the EVF image (try the "Carbon" Picture Control setting on a Z camera for an example!)
  • The Nikon Z6/Z7 have been demonstrated to have their AF affected by the EVF/LCD image
  • The Nikon Z9 has been claimed to have this behavior, but this has never been demonstrated
  • Personally, I've demonstrated to my own satisfaction that the Z9 does not exhibit this behavior
Chris,

Thank you so much for the summary, this is what my understanding too.

The image on EVF/LCD is JPG rendering, personally I use the rear LCD screen for composition and quick check of settings.

Also, I have a question regarding the shadow / dark areas, on Z5/Z6/Z7, I always have trouble seeing the shadow & dark areas while shooting photos, even though I know the exposure is well within the tonal range. I probabaly have missed some settings.

Oliver
 
Hi Marty - Here's my understanding.

The data that is used for AF is a separate stream that is also used for the EVF. That data stream is not a raw file - it's a limited data stream with the Picture Control choice applied. The dual readout allows a separate stream of limited size to be extracted from the sensor and quickly processed for the EVF and for focus. This is similar to the reason that exposure is important or increasing ISO for a brighter scene can improve AF.

So while a bit unexpected, it does affect focus. How much difference it makes depends on the scene and the choice of picture control. For example, if you use a low contrast picture control for a low contrast scene, it can make AF less effective and subject detection could be harder.
Eric, thanks for the information I appreciate it very much!

Which brings up the question if shooting RAW, would you get better AF performance if the Picture Control is set to Vivid? Is there any consequences of shooting in one Picture Control and then changing it in post?
 
Eric, thanks for the information I appreciate it very much!

Which brings up the question if shooting RAW, would you get better AF performance if the Picture Control is set to Vivid? Is there any consequences of shooting in one Picture Control and then changing it in post?
I don't think it's going to make any real difference unless you have a problem with focus and enough contrast. Then Vivid might be useful. Normally Standard or Neutral should be sufficient. It's the times when focus is challenging and you have a lack of contrast that the settings matter more than at times when lighting is normal or good.

There is no problem changing a picture control setting in post. But as you know, picture control settings determine the histogram, blinking highlights, etc. So you want to be aware of anything you are doing that could affect your other camera choices.
 
Chris,

Thank you so much for the summary, this is what my understanding too.

The image on EVF/LCD is JPG rendering, personally I use the rear LCD screen for composition and quick check of settings.

Also, I have a question regarding the shadow / dark areas, on Z5/Z6/Z7, I always have trouble seeing the shadow & dark areas while shooting photos, even though I know the exposure is well within the tonal range. I probabaly have missed some settings.

Oliver

No problem, Oliver! If you want to test this for yourself, it's pretty easy to find a really dark place and fiddle with settings to see the effect. I find it's really only significant when the light is VERY low. Low enough that I wouldn't be shooting anything that requires AF anyway (ie: stars).

About the blacked-out shadows... the camera will generally be able to recover more detail than you can see. There's a lot of shadow detail in RAW files, and the only real cost is increased noise and sometimes some color shifts. If you're worried about losing the shadows completely, I suggest using the histogram and make sure that you don't see a ton of pixels on the very left end. That way you'll know you aren't "throwing away" shadow detail.

You can also use Active D-Lighting on Nikon cameras. It brightens shadows (and protects highlights) in-camera. The effect is really only relevant in JPEGs. In fact, if you use it in RAW, Lightroom doesn't handle the RAW file properly... it biases the exposure wrong, so you have to correct it manually.

But, if you shoot JPEG anyway, it's a handy feature to avoid blocked shadows.
 
You can also use Active D-Lighting on Nikon cameras. It brightens shadows (and protects highlights) in-camera. The effect is really only relevant in JPEGs. In fact, if you use it in RAW, Lightroom doesn't handle the RAW file properly... it biases the exposure wrong, so you have to correct it manually.

But, if you shoot JPEG anyway, it's a handy feature to avoid blocked shadows.
The more recent versions of Lightroom use an embedded XMP file in the NEF to apply some settings (since 2019 I believe). This shows up as adjustments to Highlights and Shadows if you have the box set in Preferences to Use Camera Settings. With ADL Low it's an adjustment of -7 for Highlights and +10 for Shadows. Other settings of ADL have different adjustments, but they also include a real adjustment in the exposure if you have any kind of auto or semi-auto exposure mode. My current experience is LR Classic handles these settings correctly - but maybe not with the setting I would choose so I am free to make further adjustments. I believe ACR handles these settings in a similar manner since the NEF converter is virtually the same, but I have not checked.

ADL is useful if you are shooting JPEG's, and ADL Low causes no harm because it has no exposure adjustment. It's also useful for video. I usually reserve it for medium to high contrast scenes and may combine it with a Neutral picture control rather than Standard.

There are other settings that come from Nikon and are read from the embedded XMP file in the NEF. Sharpening and Noise Reduction settings are included, as well as the Picture Control choice.
 
Which brings up the question if shooting RAW, would you get better AF performance if the Picture Control is set to Vivid?
It's a very good question!

I photographed braun hares lighted with an evening sun on the green grass where my Z9 didn't want to focus at all. I had a self-made very flat profile (like all in munis)
The hares were hell brown, a bit yellowisch-orange and the grass was green. The camera didn't see the animals and tried to focus on the edge of the field where light changed to shadow. It was a very strong edging.

So, for me it looked like the AF-System is a colour-blind and see the differencies in luminance only. The luminance of hares was approximately the same as for the grass, I think.
I checked on the colour wheel and other hue-luminance tables. It could definitely be the same.
For me, of course, the colour contrast was obvious. But for camera not at all. For camera the contrast was where the light changed to shadow.
Somehow I've got it and changed to neutral and AF was working as it must. It got more contrast in luminance.

It could be true that AF-system works only with luminance contrast? who can confirm or disprove it?
I think so also becasue the luminance-contrast will work in the darkness better. Remember about cat-eyes which can differ thousands shades of gray in the darkness and see much better than humans. And theoretically colour in the camera is a filter in front of the sensor. If I would be an engineer I would probably build AF on the luminance-contrast becasue it gets more data but I am not an engineer :D

That all would conirm that by flat colour-scenaries or scenaries with similar-colours (which are similar in luminocity) the vivid profill will work better becasue it accentuates the contrast.
 
It's a very good question!

I photographed braun hares lighted with an evening sun on the green grass where my Z9 didn't want to focus at all. I had a self-made very flat profile (like all in munis)
The hares were hell brown, a bit yellowisch-orange and the grass was green. The camera didn't see the animals and tried to focus on the edge of the field where light changed to shadow. It was a very strong edging.

So, for me it looked like the AF-System is a colour-blind and see the differencies in luminance only. The luminance of hares was approximately the same as for the grass, I think.
I checked on the colour wheel and other hue-luminance tables. It could definitely be the same.
For me, of course, the colour contrast was obvious. But for camera not at all. For camera the contrast was where the light changed to shadow.
Somehow I've got it and changed to neutral and AF was working as it must. It got more contrast in luminance.

It could be true that AF-system works only with luminance contrast? who can confirm or disprove it?
I think so also becasue the luminance-contrast will work in the darkness better. Remember about cat-eyes which can differ thousands shades of gray in the darkness and see much better than humans. And theoretically colour in the camera is a filter in front of the sensor. If I would be an engineer I would probably build AF on the luminance-contrast becasue it gets more data but I am not an engineer :D

That all would conirm that by flat colour-scenaries or scenaries with similar-colours (which are similar in luminocity) the vivid profill will work better becasue it accentuates the contrast.
It's not completely color blind. 3D Tracking works primarily based on color. I think for the other modes you may be right.

I know it uses scene recognition and can recognize a blue sky, but that is for metering and I'm not sure how far that extends into AF performance. I suspect it's helpful for the camera to recognize a portrait rather than a landscape, and then subject recognition for portraits of people, animals, etc.
 
I'm thinking it would be easy to test experimentally. Put a light source on a dimmer and shoot a target. Lower the light until autofocus fails then try different picture control settings to see if there is any difference. Me, I'm going out to take some pics.
 
Back
Top