A7III banding revealed

Sakisf said:

English Rich said:

DenImage said:

English Rich said:

That is not what the images look like if you open the RAW files that I've provided in Lightroom and apply maximum Dehazing. Why would you be intentionally disingenuous?

If you're processing anything other than the RAW file in Lightroom the likelihood is that you may have stripped the offending bands out of the file prior to processing.

Here's what you'll see if you do apply 100% Dehazing - note the bands in the sky. Please note - I'm not suggesting that anybody would want to process images in this manner, but the bands remain as you reduce the dehazing. They are still slightly visible at about 25%.

Click to expand...

I clearly stated I processed your RAW files using my usual workflow (DXO PhotoLab 2.1 and Lightroom 8.1). You're being intentionally disingenuous by falsely accusing me.

Nonetheless, you're "image banding issue" is clearly with your post processing workflow or versions of software, because the images look fine with my workflow.

Den

Click to expand...

All I've done is demonstrate that banding does exist in Sony RAW files - and others have confirmed this with both Lightroom and Capture One - using my RAW files.

I'm simply demonstrating it out of academic interest as much as anything.

You came along and argued that they didn't exist by applying 100% dehazing to a non-RAW file (which no doubt didn't include the offending bands). It's hardly surprising that they didn't exist in such an example.

The fact that banding can easily be avoided by not applying 100% dehazing to a RAW file is obvious - but that's not the point of this discussion. I simply wanted to demonstrate the phenomenon. I rarely use dehazing full stop, as I find it to be crude instrument.

There's nothing wrong with investigating the limitations/quirks of a camera. Disingenuously trying to persuade others that they don't exist is counter productive.

And no - I'm not a Canon user. I'm somebody who has jumped ship from Canon to Sony and who is very happy with the setup he now owns.

Click to expand...

RawTherapee, only contrast has to be scaled up for them to show. Dehaze does a lot of contrast adjustment so that's why it's bringing them up. The more contrast adjustments you apply, the more they are evident, ie micro and local will bring them up more. Dehaze does a bit of everything it seems.

Tried various combinations with demosaicing and other colour stuff and they are not going away (they are more evident in the Red channel so maybe IR related?). Using a linear ICC as import shows the effect to its worst - full of lines.

The way they look on the second image looks like electronic interference. Try to do a shot with your mobile close and another with it away (especially if you are in a place without good coverage where the mobile is trying harder to find a signal to connect).

Also disable / enable efcs with the mobile close to the camera.

Click to expand...

Paul B Jones said:

English Rich said:

I've moved Lightroom's 'Dehaze' slider all the way to the right.

Click to expand...

The dehaze slider is a powerful tool, I never bump it more than a few points, if I use it at all. Pushing it all the way would be like slamming sharpening all the way to the right. Not sure anyone does that in the real world.

Click to expand...

Magnar W said:

Such "tests" might be very helpful to learn your own camera to know, and what limits you should work within for real world use. But would be silly to claim that a sensor is bad since you can provoke artifacts.

Click to expand...

Magnar W said:

Paul B Jones said:

English Rich said:

I've moved Lightroom's 'Dehaze' slider all the way to the right.

Click to expand...

The dehaze slider is a powerful tool, I never bump it more than a few points, if I use it at all. Pushing it all the way would be like slamming sharpening all the way to the right. Not sure anyone does that in the real world.

Click to expand...

Pushing sliders and alter values a lot at post processing will show a lot of artifacts, no matter what camera/sensor was used.

Such "tests" might be very helpful to learn your own camera to know, and what limits you should work within for real world use. But would be silly to claim that a sensor is bad since you can provoke artifacts.

Click to expand...

English Rich said:

Magnar W said:

Such "tests" might be very helpful to learn your own camera to know, and what limits you should work within for real world use. But would be silly to claim that a sensor is bad since you can provoke artifacts.

Click to expand...

I agree entirely. To be clear - I'm not saying that the A7III sensor is bad in any way. In fact, I think it's the best sensor that any consumer camera currently has to offer. I simply wanted to share my findings.

Click to expand...

English Rich said:

Sakisf said:

English Rich said:

DenImage said:

English Rich said:

That is not what the images look like if you open the RAW files that I've provided in Lightroom and apply maximum Dehazing. Why would you be intentionally disingenuous?

If you're processing anything other than the RAW file in Lightroom the likelihood is that you may have stripped the offending bands out of the file prior to processing.

Here's what you'll see if you do apply 100% Dehazing - note the bands in the sky. Please note - I'm not suggesting that anybody would want to process images in this manner, but the bands remain as you reduce the dehazing. They are still slightly visible at about 25%.

Click to expand...

I clearly stated I processed your RAW files using my usual workflow (DXO PhotoLab 2.1 and Lightroom 8.1). You're being intentionally disingenuous by falsely accusing me.

Nonetheless, you're "image banding issue" is clearly with your post processing workflow or versions of software, because the images look fine with my workflow.

Den

Click to expand...

All I've done is demonstrate that banding does exist in Sony RAW files - and others have confirmed this with both Lightroom and Capture One - using my RAW files.

I'm simply demonstrating it out of academic interest as much as anything.

You came along and argued that they didn't exist by applying 100% dehazing to a non-RAW file (which no doubt didn't include the offending bands). It's hardly surprising that they didn't exist in such an example.

The fact that banding can easily be avoided by not applying 100% dehazing to a RAW file is obvious - but that's not the point of this discussion. I simply wanted to demonstrate the phenomenon. I rarely use dehazing full stop, as I find it to be crude instrument.

There's nothing wrong with investigating the limitations/quirks of a camera. Disingenuously trying to persuade others that they don't exist is counter productive.

And no - I'm not a Canon user. I'm somebody who has jumped ship from Canon to Sony and who is very happy with the setup he now owns.

Click to expand...

RawTherapee, only contrast has to be scaled up for them to show. Dehaze does a lot of contrast adjustment so that's why it's bringing them up. The more contrast adjustments you apply, the more they are evident, ie micro and local will bring them up more. Dehaze does a bit of everything it seems.

Tried various combinations with demosaicing and other colour stuff and they are not going away (they are more evident in the Red channel so maybe IR related?). Using a linear ICC as import shows the effect to its worst - full of lines.

The way they look on the second image looks like electronic interference. Try to do a shot with your mobile close and another with it away (especially if you are in a place without good coverage where the mobile is trying harder to find a signal to connect).

Also disable / enable efcs with the mobile close to the camera.

Click to expand...

This has nothing to do with my mobile. You can replicate the issue without any other electrical devices anywhere near the camera - another contributor has revealed exactly the same banding. Another sensible explanation (see thread) is that the lines relate to PDAF points.

Click to expand...

mcslsk said:

Magnar W said:

Paul B Jones said:

English Rich said:

I've moved Lightroom's 'Dehaze' slider all the way to the right.

Click to expand...

The dehaze slider is a powerful tool, I never bump it more than a few points, if I use it at all. Pushing it all the way would be like slamming sharpening all the way to the right. Not sure anyone does that in the real world.

Click to expand...

Pushing sliders and alter values a lot at post processing will show a lot of artifacts, no matter what camera/sensor was used.

Such "tests" might be very helpful to learn your own camera to know, and what limits you should work within for real world use. But would be silly to claim that a sensor is bad since you can provoke artifacts.

Click to expand...

Magnar, did the OP or anyone else here claim that the Sony sensor is a bad sensor? That would be silly, wouldn't it. It is the most advanced sensor out there.

Click to expand...

Sakisf said:

English Rich said:

DenImage said:

English Rich said:

That is not what the images look like if you open the RAW files that I've provided in Lightroom and apply maximum Dehazing. Why would you be intentionally disingenuous?

If you're processing anything other than the RAW file in Lightroom the likelihood is that you may have stripped the offending bands out of the file prior to processing.

Here's what you'll see if you do apply 100% Dehazing - note the bands in the sky. Please note - I'm not suggesting that anybody would want to process images in this manner, but the bands remain as you reduce the dehazing. They are still slightly visible at about 25%.

Click to expand...

I clearly stated I processed your RAW files using my usual workflow (DXO PhotoLab 2.1 and Lightroom 8.1). You're being intentionally disingenuous by falsely accusing me.

Nonetheless, you're "image banding issue" is clearly with your post processing workflow or versions of software, because the images look fine with my workflow.

Den

Click to expand...

All I've done is demonstrate that banding does exist in Sony RAW files - and others have confirmed this with both Lightroom and Capture One - using my RAW files.

I'm simply demonstrating it out of academic interest as much as anything.

You came along and argued that they didn't exist by applying 100% dehazing to a non-RAW file (which no doubt didn't include the offending bands). It's hardly surprising that they didn't exist in such an example.

The fact that banding can easily be avoided by not applying 100% dehazing to a RAW file is obvious - but that's not the point of this discussion. I simply wanted to demonstrate the phenomenon. I rarely use dehazing full stop, as I find it to be crude instrument.

There's nothing wrong with investigating the limitations/quirks of a camera. Disingenuously trying to persuade others that they don't exist is counter productive.

And no - I'm not a Canon user. I'm somebody who has jumped ship from Canon to Sony and who is very happy with the setup he now owns.

Click to expand...

RawTherapee, only contrast has to be scaled up for them to show. Dehaze does a lot of contrast adjustment so that's why it's bringing them up. The more contrast adjustments you apply, the more they are evident, ie micro and local will bring them up more. Dehaze does a bit of everything it seems.

Tried various combinations with demosaicing and other colour stuff and they are not going away (they are more evident in the Red channel so maybe IR related?). Using a linear ICC as import shows the effect to its worst - full of lines.

The way they look on the second image looks like electronic interference. Try to do a shot with your mobile close and another with it away (especially if you are in a place without good coverage where the mobile is trying harder to find a signal to connect).

Also disable / enable efcs with the mobile close to the camera.

Click to expand...

dethis2 said:

No .. no relation with interference. The bands are clearly following the PDAF lines' pattern (every 6 or 12 or 18 rows)

With RT, choose none demosaic then display at 50% and choose "red/blue/green channel preview" (use the red/green/blue/L small squares at the top)

The most visible lines are in the Blue channel .. which matches with the placement of the PDAF lines on the G2-B raw rows ..

Καλή Χρονιά

Click to expand...

dethis2 said:

No .. no relation with interference. The bands are clearly following the PDAF lines' pattern (every 6 or 12 or 18 rows)

With RT, choose none demosaic then display at 50% and choose "red/blue/green channel preview" (use the red/green/blue/L small squares at the top)

The most visible lines are in the Blue channel .. which matches with the placement of the PDAF lines on the G2-B raw rows ..

Καλή Χρονιά

Click to expand...

Dirk W said:

Interesting! It's probably something like noise when you use really high ISOs.

Click to expand...

English Rich said:

Dirk W said:

Interesting! It's probably something like noise when you use really high ISOs.

Click to expand...

Hi Dirk. This is a long thread, but if you look at the contributions titled 'PDAF' striping you'll see that a couple of individuals have discovered that the lines correspond with the positioning of phase detect auto focus points.

It seems that these lines are present in properly exposed, ISO 100 RAW files, but you have to push the files a lot to reveal them.

In practice it's not a problem - nobody would push the dehaze filter beyond about 25% anyway.

Click to expand...

Dirk W said:

English Rich said:

Dirk W said:

Interesting! It's probably something like noise when you use really high ISOs.

Click to expand...

Hi Dirk. This is a long thread, but if you look at the contributions titled 'PDAF' striping you'll see that a couple of individuals have discovered that the lines correspond with the positioning of phase detect auto focus points.

It seems that these lines are present in properly exposed, ISO 100 RAW files, but you have to push the files a lot to reveal them.

In practice it's not a problem - nobody would push the dehaze filter beyond about 25% anyway.

Click to expand...

Yeah, I see. Pushing in pp is pretty much the same as using really high ISOs I think, because from a certain ISO value up that's what the camera does in body: pushing the dark picture. If the noise reveals the pdaf pixels, interesting ! But as you write : no problem for photography.

Click to expand...

mcslsk said:

Dirk W said:

English Rich said:

Dirk W said:

Interesting! It's probably something like noise when you use really high ISOs.

Click to expand...

Hi Dirk. This is a long thread, but if you look at the contributions titled 'PDAF' striping you'll see that a couple of individuals have discovered that the lines correspond with the positioning of phase detect auto focus points.

It seems that these lines are present in properly exposed, ISO 100 RAW files, but you have to push the files a lot to reveal them.

In practice it's not a problem - nobody would push the dehaze filter beyond about 25% anyway.

Click to expand...

Yeah, I see. Pushing in pp is pretty much the same as using really high ISOs I think, because from a certain ISO value up that's what the camera does in body: pushing the dark picture. If the noise reveals the pdaf pixels, interesting ! But as you write : no problem for photography.

Click to expand...

Hallo Dirk, wouldn't say "no" problem but in the vast majority of images not visible. By the way: Sony's Imaging Edge allows to push 2 stops in post, Canon's DPP 3 stops. So the manufacturer software clearly sets limits that you can (but only for good reason) exceed in programs like LR (+5 stops).

Click to expand...

mcslsk said:

Hallo Dirk, wouldn't say "no" problem but in the vast majority of images not visible. By the way: Sony's Imaging Edge allows to push 2 stops in post, Canon's DPP 3 stops. So the manufacturer software clearly sets limits that you can (but only for good reason) exceed in programs like LR (+5 stops).

Click to expand...