GFX100S 14-bit vs 16-bit

[Rich42]

Thanks Jim. Just checked raw digger and it is 14. So why would Adobe say it is 16? Very odd.....

Because ACR processes the raw file into either 8 bit or 16 bit file format. The 16 bit file contains the 14 bits of data that the camera outputs.

Rich

 

[JimKasson]

Thanks Jim. Just checked raw digger and it is 14. So why would Adobe say it is 16? Very odd.....

Because ACR processes the raw file into either 8 bit or 16 bit file format. The 16 bit file contains the 14 bits of data that the camera outputs.

Yes. See this:

https://www.dpreview.com/forums/post/64937901

 

[Greg7579]

Thanks Jim. Just checked raw digger and it is 14. So why would Adobe say it is 16? Very odd.....

Because ACR processes the raw file into either 8 bit or 16 bit file format. The 16 bit file contains the 14 bits of data that the camera outputs.

Rich

That is interesting but Adobe says 14 bit on the Q2. It is odd that Adobe LR, PS and Bridge would report the wrong bits on the metadata readouts of the GFX 100 set at 14 bits.

 

[Rich42]

Thanks Jim. Just checked raw digger and it is 14. So why would Adobe say it is 16? Very odd.....

Because ACR processes the raw file into either 8 bit or 16 bit file format. The 16 bit file contains the 14 bits of data that the camera outputs.

Rich

That is interesting but Adobe says 14 bit on the Q2. It is odd that Adobe LR, PS and Bridge would report the wrong bits on the metadata readouts of the GFX 100 set at 14 bits.

I don't know why it's displaying like that. I've never seen ACR/LR/PS display files as anything but 8 bit or 16 bit. The file format is what's displayed and it has nothing to do with the number of bits within the file contain any useable data.

I've never processed a Q2 file. As I understand, the file is a DNG. If you want to provide a link to one of yours I'd be glad to take a look at it and see what's up with the display of bit depth/file format.

Rich

 

[SrMi]

Thanks Jim. Just checked raw digger and it is 14. So why would Adobe say it is 16? Very odd.....

Because ACR processes the raw file into either 8 bit or 16 bit file format. The 16 bit file contains the 14 bits of data that the camera outputs.

Rich

That is interesting but Adobe says 14 bit on the Q2. It is odd that Adobe LR, PS and Bridge would report the wrong bits on the metadata readouts of the GFX 100 set at 14 bits.

I don't know why it's displaying like that. I've never seen ACR/LR/PS display files as anything but 8 bit or 16 bit. The file format is what's displayed and it has nothing to do with the number of bits within the file contain any useable data.

I've never processed a Q2 file. As I understand, the file is a DNG. If you want to provide a link to one of yours I'd be glad to take a look at it and see what's up with the display of bit depth/file format.

Rich

Greg is looking at the data in Adobe Bridge. It displays the metadata of the selected image.

 

[Rich42]

Thanks Jim. Just checked raw digger and it is 14. So why would Adobe say it is 16? Very odd.....

Because ACR processes the raw file into either 8 bit or 16 bit file format. The 16 bit file contains the 14 bits of data that the camera outputs.

Rich

That is interesting but Adobe says 14 bit on the Q2. It is odd that Adobe LR, PS and Bridge would report the wrong bits on the metadata readouts of the GFX 100 set at 14 bits.

I don't know why it's displaying like that. I've never seen ACR/LR/PS display files as anything but 8 bit or 16 bit. The file format is what's displayed and it has nothing to do with the number of bits within the file contain any useable data.

I've never processed a Q2 file. As I understand, the file is a DNG. If you want to provide a link to one of yours I'd be glad to take a look at it and see what's up with the display of bit depth/file format.

Rich

Greg is looking at the data in Adobe Bridge. It displays the metadata of the selected image.

To Greg:

Dude!?

Rich

 

[TheGrilledCheese]

There is horizontal banding here, more so in the 16-bit file. I thought folks said there was no banding with the GFX 100S. From these images it is clearly there.

Where is here? Do you have a link?

If you would care to re-read the message you were responding to, you would see that I was responding to fracer's post and link. So if you want a link, please reference the message to which I was responding. There is clear banding visible in the 16-bit image. There is some in the 14-bit as well, but not as pronounced.

 

[JimKasson]

There is horizontal banding here, more so in the 16-bit file. I thought folks said there was no banding with the GFX 100S. From these images it is clearly there.

Where is here? Do you have a link?

If you would care to re-read the message you were responding to, you would see that I was responding to fracer's post and link. So if you want a link, please reference the message to which I was responding. There is clear banding visible in the 16-bit image. There is some in the 14-bit as well, but not as pronounced.

He referred to my work with the GFX 100. I didn’t test the GFX 100S.

 

[NAwlins Contrarian]

There is horizontal banding here, more so in the 16-bit file. I thought folks said there was no banding with the GFX 100S. From these images it is clearly there.

Where is here? Do you have a link?

If you would care to re-read the message you were responding to, you would see that I was responding to fracer's post and link. So if you want a link, please reference the message to which I was responding. There is clear banding visible in the 16-bit image. There is some in the 14-bit as well, but not as pronounced.

He referred to my work with the GFX 100. I didn’t test the GFX 100S.

Well that was a more restrained response than I would have given!

 

[TheGrilledCheese]

There is horizontal banding here, more so in the 16-bit file. I thought folks said there was no banding with the GFX 100S. From these images it is clearly there.

Where is here? Do you have a link?

If you would care to re-read the message you were responding to, you would see that I was responding to fracer's post and link. So if you want a link, please reference the message to which I was responding. There is clear banding visible in the 16-bit image. There is some in the 14-bit as well, but not as pronounced.

He referred to my work with the GFX 100. I didn’t test the GFX 100S.

Thanks, I thought it was for the GFX100s. Glad to know it doesn't have the banding issues of the 100. Cheers.

 

[JimKasson]

There is horizontal banding here, more so in the 16-bit file. I thought folks said there was no banding with the GFX 100S. From these images it is clearly there.

Where is here? Do you have a link?

If you would care to re-read the message you were responding to, you would see that I was responding to fracer's post and link. So if you want a link, please reference the message to which I was responding. There is clear banding visible in the 16-bit image. There is some in the 14-bit as well, but not as pronounced.

He referred to my work with the GFX 100. I didn’t test the GFX 100S.

Thanks, I thought it was for the GFX100s. Glad to know it doesn't have the banding issues of the 100. Cheers.

Was I unclear here ?

I asked for the link because I thought you were talking about a test of the GFX 100S, which is why I asked "where is here"?

And by the way, I'm not willing to declare the GFX 100S free of PDAF banding until I test it myself (this week, I hope), although some others have asserted that.

Jim

 

[Greg7579]

I'm running out of popcorn.

My money is on Jim.

 

[Atlantis107]

Hi,

I know this is a few months old now - but when you're looking in Adobe Bridge I'm pretty sure the bit depth you're seeing has nothing to do with your raw in-camera settings but is what Adobe is using to process.

Cameras mostly shoot 12 and 14 bit raw as well as 8-bit jpegs. Photoshop and other software mostly provide 8 and 16-bit processing (also 32-bit for HDR). Most monitors are only 8 bit, with professional photo editing monitors usually being 10 bit.

Think of them as three separate things. Using 14 bit raw in-camera (compressed or not) gives you a huge amount of tonal values and editing 'headroom'. If you were to edit in 8 bit in Photoshop you'd be throwing a huge amount of data away and may see banding/pixelation etc. 16 bit in Adobe (actually '15+1') keeps all the information from your 14 bit raw image and gives you the headroom to edit. Similarily, using 32bit in Photoshop further allows you to stack images for HDR.

All you're ever seeing with your eyes though is essentially an 8-bit image, the rest is editing room. Even if you have an expensive 10-bit monitor with 100 percent adobe RGB coverage and shoot in Adobe RGB (more cyan/green). Most monitors don't even have full coverage of sRGB let alone Adobe RGB.

Once you're done editing, when you convert your file to an 8-bit jpeg for final use/delivery you won't see any difference to the image - what you will lose with that jpeg is the ability to manipulate it like a raw image, as that extra headroom data has now been discarded and your editing selections are now baked in.

If you're shooting a contrasty scene with a lot of dynamic range, shoot with the maximum raw bit, usually 14 bit. shooting a landscape in 12 bit I've run into banding in the skies when pushing/pulling highlights/shadows. In photoshop, edit in 16 bit, or 32 bit if you're doing HDR. And keep in mind what monitor you're using - what's the spectrum coverage like? Is it calibrated? If it's a cheap consumer monitor, you'll never really know what your image 'truly' looks like.

J

 

[JimKasson]

Hi,

I know this is a few months old now - but when you're looking in Adobe Bridge I'm pretty sure the bit depth you're seeing has nothing to do with your raw in-camera settings but is what Adobe is using to process.

Cameras mostly shoot 12 and 14 bit raw as well as 8-bit jpegs.

The GFX 100S, the subject of this thread, allows 14-bit and 16-bit raw precision.

Photoshop and other software mostly provide 8 and 16-bit processing

Photoshop calls their step up from 8 bit precision 16 bit, but it is actually not. It is 15 bits with one additional allowed value. But the precision of Ps and raw files is not comparable without additional calculations, because Ps in general uses a tone curve, and the GFX 100S does not use a tone curve for raw files.

(also 32-bit for HDR). Most monitors are only 8 bit, with professional photo editing monitors usually being 10 bit.

Let us not mix linear representations with those using tone curves without pointing out what's going on. That's misleading.

Think of them as three separate things. Using 14 bit raw in-camera (compressed or not) gives you a huge amount of tonal values and editing 'headroom'. If you were to edit in 8 bit in Photoshop you'd be throwing a huge amount of data away and may see banding/pixelation etc. 16 bit in Adobe (actually '15+1')

There you go. That's right. But don't forget the tonoe curves.

keeps all the information from your 14 bit raw image and gives you the headroom to edit. Similarily, using 32bit in Photoshop further allows you to stack images for HDR.

All you're ever seeing with your eyes though is essentially an 8-bit image, the rest is editing room.

No! The tone curve is important here.

Even if you have an expensive 10-bit monitor with 100 percent adobe RGB coverage and shoot in Adobe RGB (more cyan/green). Most monitors don't even have full coverage of sRGB let alone Adobe RGB.

Once you're done editing, when you convert your file to an 8-bit jpeg for final use/delivery you won't see any difference to the image - what you will lose with that jpeg is the ability to manipulate it like a raw image, as that extra headroom data has now been discarded and your editing selections are now baked in.

If you're shooting a contrasty scene with a lot of dynamic range, shoot with the maximum raw bit, usually 14 bit. shooting a landscape in 12 bit I've run into banding in the skies when pushing/pulling highlights/shadows. In photoshop, edit in 16 bit, or 32 bit if you're doing HDR. And keep in mind what monitor you're using - what's the spectrum coverage like? Is it calibrated? If it's a cheap consumer monitor, you'll never really know what your image 'truly' looks like.

The comparisons that you are doing need to take the effect of the tone curves into account to be useful.

 

[Greg7579]

Atlantis, that is a great post. Really awesome. It all goes into my notes.

Now, I'll wait and see if Jim or anyone says whether or not it is all correct.

 

[Totusss]

Hello,

On website www.photonstophotos.net when you select Fujifilm GFX 100_14 and Fujifilm GFX 100_16 , there are a big gap. Do you think, it come from the horizontal banding more present?

Read Noise in DNs versus ISO Setting

 

[highdesertmesa]

Hello,

On website www.photonstophotos.net when you select Fujifilm GFX 100_14 and Fujifilm GFX 100_16 , there are a big gap. Do you think, it come from the horizontal banding more present?

https://www.photonstophotos.net/Charts/RN_ADU.htm

I have yet to see any banding on the 100S in 16 bit when pushing properly exposed, low ISO files in post.

 

[JimKasson]

Hello,

On website www.photonstophotos.net when you select Fujifilm GFX 100_14 and Fujifilm GFX 100_16 , there are a big gap. Do you think, it come from the horizontal banding more present?

https://www.photonstophotos.net/Charts/RN_ADU.htm

That curve measures the RN in counts (aka LSBs, ADUs, or DN). One count in 16 bit precision referred to the charge on the pixel on the sensor is 1/4 the amount of the RN metric as one count in 14 bit precision.

RN is not a good metric for banding.

Jim

 

[bclaff]

Hello,

On website www.photonstophotos.net when you select Fujifilm GFX 100_14 and Fujifilm GFX 100_16 , there are a big gap. Do you think, it come from the horizontal banding more present?

https://www.photonstophotos.net/Charts/RN_ADU.htm

As Jim points out this is expected because the units of the y-axis are affected by bit depth.

You can see evidence of banding looking at the PhotonsToPhotos Sensor Heatmaps


Best viewed "original size"

--
Bill ( Your trusted source for independent sensor data at PhotonsToPhotos )

 

[draschan]

hi, thanks for your test. I remember a test I saw with a hasselblad a few years ago. the capabilty to discern red tones & hues seemed significantly enhanced. in 16bit

this just came to my mind, when I saw the closeup of the red book fabric that looks different in 14 & 16bit in your test.

maybe it would be more telling to try an object with a csubtle olour gradient that spans a wide range, and to a split image comparison.

it could also be, that the 33x44 sensor is still not there, where the larger hasselblad sensors were previously.

I 'm still trying to find this test (unforuntately I can't remember where I saw it)

enjoy your camera, best thomas