Z7 raw compression observations

Started 2 months ago | Discussions
Richard Costin
Richard Costin Regular Member • Posts: 162
Z7 raw compression observations
4

I always check and compare the raw compression modes when I get a new camera to make sure my current thinking of shooting 12 bit compressed still holds.

Initial results hold true for the Z7 with an interesting caveat on the banding that’s appears when unrealistically pushing the black areas above any real world levels. 12 bit mode seems to all but remove it...

https://www.richardcostin.com/post/2018/10/nikon_z7_raw_tests_001/

-- hide signature --

Richard Costin
Wildlife Photographer
www.richardcostin.com
www.facebook.com/richardcwildlife

 Richard Costin's gear list:Richard Costin's gear list
Apple iPhone X
Nikon Z7
If you believe there are incorrect tags, please send us this post using our feedback form.
Bohemien
Bohemien Contributing Member • Posts: 683
Thanks!

Thanks for doing this test, very interesting!

-- hide signature --
 Bohemien's gear list:Bohemien's gear list
Nikon Coolpix A Nikon D600
rambalmur New Member • Posts: 19
Re: Z7 raw compression observations

I just tested this myself and yes, though the banding appears to be removed, there is a significant DR penalty for shooting in 12bit mode. There is a significant green shift when lifting shadows that does not appear in the 14 bit file. This is the same behaviour I noticed on my D850 which stopped me from using any of the reduced resolution modes (which also appear to default to 12 bit).

 rambalmur's gear list:rambalmur's gear list
Nikon Z7 Nikon D750 Sony a7R II Fujifilm X-T2 Nikon AF-S Nikkor 14-24mm f/2.8G ED +15 more
Jack Hogan Veteran Member • Posts: 6,682
Re: Z7 raw compression observations
1

rambalmur wrote:

I just tested this myself and yes, though the banding appears to be removed, there is a significant DR penalty for shooting in 12bit mode. There is a significant green shift when lifting shadows that does not appear in the 14 bit file. This is the same behaviour I noticed on my D850 which stopped me from using any of the reduced resolution modes (which also appear to default to 12 bit).

If the shift is there in 12 bits and not 14 aotbe, I would guess it is probably due to an incorrect black level.

Richard Costin
OP Richard Costin Regular Member • Posts: 162
Re: Z7 raw compression observations

Interesting you are seeing a green shift, what were the conditions/settings you shot under?

As you can see on my example there was no shift.

-- hide signature --

Richard Costin
Wildlife Photographer
www.richardcostin.com
www.facebook.com/richardcwildlife

 Richard Costin's gear list:Richard Costin's gear list
Apple iPhone X
Jack Hogan Veteran Member • Posts: 6,682
Re: Z7 raw compression observations

Richard Costin wrote:

I always check and compare the raw compression modes when I get a new camera to make sure my current thinking of shooting 12 bit compressed still holds.

Initial results hold true for the Z7 with an interesting caveat on the banding that’s appears when unrealistically pushing the black areas above any real world levels. 12 bit mode seems to all but remove it...

https://www.richardcostin.com/post/2018/10/nikon_z7_raw_tests_001/

Interesting Richard, I don't know why the banding would be gone at 12 bits.  It's also difficult to know what else LR is doing differently when pushing exposure with different capture setups.

Jack

rambalmur New Member • Posts: 19
Re: Z7 raw compression observations
1

Richard Costin wrote:

Interesting you are seeing a green shift, what were the conditions/settings you shot under?

As you can see on my example there was no shift.

Here you go. 
Both images were made with identical exposures a few minutes apart - + 3.00 EV Push in exposure and +60 shadows using LR. No other changes. 
I believe the banding may be being hidden by the excess noise that is introduced by shooting in 12 bit.

 rambalmur's gear list:rambalmur's gear list
Nikon Z7 Nikon D750 Sony a7R II Fujifilm X-T2 Nikon AF-S Nikkor 14-24mm f/2.8G ED +15 more
Richard Costin
OP Richard Costin Regular Member • Posts: 162
Re: Z7 raw compression observations

Thanks for the example, I'll try and replicate here.

Seems to manifest in broad/large shadowed areas. Perhaps the compression algorithm treats these differently to shadow areas with lots of variation.

I'll flag this up with Adobe as well and see what they say.

-- hide signature --

Richard Costin
Wildlife Photographer
www.richardcostin.com
www.facebook.com/richardcwildlife

 Richard Costin's gear list:Richard Costin's gear list
Apple iPhone X
Richard Costin
OP Richard Costin Regular Member • Posts: 162
Re: Z7 raw compression observations

Interesting. If I shoot into a black closet I do indeed see some green appearing when pushed back up in post.

This looks absolutely to be related to 12 vs 14 bit (not quite sure why). Compression seems to make no difference in either mode. I usually hate to get to into this nitty gritty as often these things absolutely don't resolve into artefacts that impact my images but this one is worth looking in to.

I will report back once I have talked with Adobe & Nikon. The fact this issue was not visible on my initial test (with lots of details filling the frame) is odd and encouraging. I could understand this with the compression which would vary per image but simply dropping from 14 to 12 bit wouldn't (well shouldn't) introduces new colours so to speak.

Will set to 14bit compressed for the moment until I get some more info.

Thanks again rambalmur for flagging up your observations.

As well, if any of you full on technical chaps that run the in depth sensor tests have any thoughts I am sure we would all appreciate it. I'd love an informed explanation as to what is going on.

Rich.

-- hide signature --

Richard Costin
Wildlife Photographer
www.richardcostin.com
www.instagram.com/richard_costin
www.facebook.com/richardwildlife

 Richard Costin's gear list:Richard Costin's gear list
Apple iPhone X
rambalmur New Member • Posts: 19
Re: Z7 raw compression observations
1

Richard Costin wrote:

Interesting. If I shoot into a black closet I do indeed see some green appearing when pushed back up in post.

This looks absolutely to be related to 12 vs 14 bit (not quite sure why). Compression seems to make no difference in either mode. I usually hate to get to into this nitty gritty as often these things absolutely don't resolve into artefacts that impact my images but this one is worth looking in to.

I will report back once I have talked with Adobe & Nikon. The fact this issue was not visible on my initial test (with lots of details filling the frame) is odd and encouraging. I could understand this with the compression which would vary per image but simply dropping from 14 to 12 bit wouldn't (well shouldn't) introduces new colours so to speak.

Will set to 14bit compressed for the moment until I get some more info.

Thanks again rambalmur for flagging up your observations.

As well, if any of you full on technical chaps that run the in depth sensor tests have any thoughts I am sure we would all appreciate it. I'd love an informed explanation as to what is going on.

Rich.

I'm glad you were able to replicate my results. I noticed this issue first when I was shooting the D850. I made the mistake of assuming 12bit images would have the same shadow recovery potential as 14 and shot a couple of important portraits at a wedding working with that assumption. Let's just say they had more black and white images from that set than normal. 
I've also run into a number of oddities with banding on the Z7 as well, where I don't see banding in situations I would expect to see it and conversely, see it pop up with a much more gentle push than expected.

Please keep us appraised about what Nikon/Adobe say about this issue and any information about mitigating banding if possible. 
Thanks!

 rambalmur's gear list:rambalmur's gear list
Nikon Z7 Nikon D750 Sony a7R II Fujifilm X-T2 Nikon AF-S Nikkor 14-24mm f/2.8G ED +15 more
Richard Costin
OP Richard Costin Regular Member • Posts: 162
Re: Z7 raw compression observations
1

Will do.

I have also observed the issue progressively reduces the higher you turn up the iso and seems gone by about 1000.

Absolutely was not seeing this on the D3, D800. D810 and D500. Less experience with the D850, if this sensor is of the same generation minus the phase sensors then it makes sense it would exhibit it too. This is why I retest with each new camera I use.

Cheers, speak soon hopefully.

Rich.

-- hide signature --

Richard Costin
Wildlife Photographer
www.richardcostin.com
www.instagram.com/richard_costin
www.facebook.com/richardwildlife

 Richard Costin's gear list:Richard Costin's gear list
Apple iPhone X
Richard Costin
OP Richard Costin Regular Member • Posts: 162
Re: Z7 raw compression observations
1

This interesting set of data may be relevant...

https://blog.kasson.com/the-last-word/nikon-12-bit-raw-mode/

-- hide signature --

Richard Costin
Wildlife Photographer
www.richardcostin.com
www.instagram.com/richard_costin
www.facebook.com/richardwildlife

 Richard Costin's gear list:Richard Costin's gear list
Apple iPhone X
JimKasson
MOD JimKasson Forum Pro • Posts: 21,738
Re: Z7 raw compression observations
2

Jack Hogan wrote:

rambalmur wrote:

I just tested this myself and yes, though the banding appears to be removed, there is a significant DR penalty for shooting in 12bit mode. There is a significant green shift when lifting shadows that does not appear in the 14 bit file. This is the same behaviour I noticed on my D850 which stopped me from using any of the reduced resolution modes (which also appear to default to 12 bit).

If the shift is there in 12 bits and not 14 aotbe, I would guess it is probably due to an incorrect black level.

Yes, like the infamous Leica M240 green shadows. Too high a black level results in a green shift near black, because of the relative sensitivities of the four planes in the Bayer matrix. Too low a black level gives a magenta shift.

-- hide signature --

Posted as a regular forum member.
https://blog.kasson.com

 JimKasson's gear list:JimKasson's gear list
Nikon D5 Fujifilm GFX 50S Sony a9 Sony a7R III Sony a7 III +5 more
JimKasson
MOD JimKasson Forum Pro • Posts: 21,738
Re: Z7 raw compression observations

Richard Costin wrote:

Will do.

I have also observed the issue progressively reduces the higher you turn up the iso and seems gone by about 1000.

Same as with the M240.

Absolutely was not seeing this on the D3, D800. D810 and D500. Less experience with the D850, if this sensor is of the same generation minus the phase sensors then it makes sense it would exhibit it too. This is why I retest with each new camera I use.

Cheers, speak soon hopefully.

Rich.

-- hide signature --

Posted as a regular forum member.
https://blog.kasson.com

 JimKasson's gear list:JimKasson's gear list
Nikon D5 Fujifilm GFX 50S Sony a9 Sony a7R III Sony a7 III +5 more
Iliah Borg Forum Pro • Posts: 24,795
lossy compression
3

14-bit lossy is internally coded as 12 bits;

12-bit lossy is internally coded as 10 bits

-- hide signature --
Iliah Borg Forum Pro • Posts: 24,795
Re: Z7 raw compression observations

rambalmur wrote:

I just tested this myself and yes, though the banding appears to be removed, there is a significant DR penalty for shooting in 12bit mode. There is a significant green shift when lifting shadows that does not appear in the 14 bit file. This is the same behaviour I noticed on my D850 which stopped me from using any of the reduced resolution modes (which also appear to default to 12 bit).

Is it noticeable on out-of-camera JPEGs?

-- hide signature --
MOD TOF guy Forum Pro • Posts: 14,991
Re: lossy compression

Iliah wrote:

14-bit lossy is internally coded as 12 bits;

Is it just a conversion from 14 to 12 bits or does it use the same clever algorithm as before, starting rounding only somewhere in the mid tones and from there always keeping the rounding error lower that shot noise ?

-- hide signature --

Thierry - posted as regular forum member

 TOF guy's gear list:TOF guy's gear list
Nikon D810 Nikon D850 Nikon AF-S Nikkor 70-200mm f/2.8G ED VR Nikon AF-S DX Nikkor 12-24mm f/4G ED-IF Nikon AF-S Nikkor 24-70mm f/2.8G ED +6 more
rambalmur New Member • Posts: 19
Re: Z7 raw compression observations

Iliah Borg wrote:

rambalmur wrote:

I just tested this myself and yes, though the banding appears to be removed, there is a significant DR penalty for shooting in 12bit mode. There is a significant green shift when lifting shadows that does not appear in the 14 bit file. This is the same behaviour I noticed on my D850 which stopped me from using any of the reduced resolution modes (which also appear to default to 12 bit).

Is it noticeable on out-of-camera JPEGs?

Yes, I see the green shift in JPEG images, though interestingly there is "banding" as well. 12 Bit lossless compressed also shows the green shift but no banding.

 rambalmur's gear list:rambalmur's gear list
Nikon Z7 Nikon D750 Sony a7R II Fujifilm X-T2 Nikon AF-S Nikkor 14-24mm f/2.8G ED +15 more
Absolutic
Absolutic Veteran Member • Posts: 5,433
Re: Z7 raw compression observations

Richard Costin wrote:

Will do.

I have also observed the issue progressively reduces the higher you turn up the iso and seems gone by about 1000.

Absolutely was not seeing this on the D3, D800. D810 and D500. Less experience with the D850, if this sensor is of the same generation minus the phase sensors then it makes sense it would exhibit it too. This is why I retest with each new camera I use.

Cheers, speak soon hopefully.

Rich.

Hey Richard a little off top, but I have D500 and I see it is on your list.  Did you investigate D500, and are you telling me that I can safely shoot 12bit lossy on D500 and pretty much recover the same amount in Lightroom as I normally shoot with 14bit lossless?   I'd like to shoot smaller files if there is no penalty.

 Absolutic's gear list:Absolutic's gear list
Sony RX100 Nikon Coolpix P7700 Sony Cyber-shot DSC-RX10 Panasonic G85 Nikon D850 +24 more
Iliah Borg Forum Pro • Posts: 24,795
Re: lossy compression
1

TOF guy wrote:

Iliah wrote:

14-bit lossy is internally coded as 12 bits;

Is it just a conversion from 14 to 12 bits or does it use the same clever algorithm as before, starting rounding only somewhere in the mid tones and from there always keeping the rounding error lower that shot noise ?

The raw data contained in the NEF file, obviously before linearization (application of NEFLinearizationTable, makernotes tag 0x0096) is 10- and 12-bit, respectively.

-- hide signature --
Keyboard shortcuts:
FForum MMy threads