RAW single vs burst -- 14-bits vs 12-bits ?

[PJD]

On p122 of the S5iiX user manual, it says, "The color depth of RAW images recorded with this camera is 14 bits. This is limited to 12 bits during burst recording, however."

I have questions:

1.) My image editing software (Affinity Photo 2.5.3 & Pixelmator Pro 3.6.4) in their EXIF windows report "16 bpc" for all S5iiX RAW files, whether shot as single shots, burst sequences, or bracketed sequences. Do any of you have software that report RAW files as being "14 bits" or "12 bits"?

2.) If you have software that report RAW files as being "14 bits" or "12 bits", can you confirm what Panasonic states in the user manual? Are single RAW exposures in fact 14 bits, and burst RAW images 12 bits? And what about bracketed shots?

3). I'm wondering if there's an "easy" way to see in my test images the difference between 14 bits and 12 bits? For example, should I take locked-off test photos of a clear blue sky and look for signs of gradient banding artifacts, etc.?

I'm mostly just curious about this. I'm not overly concerned about the 2 bit difference. However, as I often use exposure bracketing, and always shoot RAW, it would be great to know if bracketed images are "lower quality" than single shots. Thanks for any replies.

 

[inspiredtimothy]

Also interested in hearing the answer to this.

 

[Pete_W]

There was a thread about this over on DPR (that den of vipers!) a few months ago.

Panasonic II/IIx switches to 12-bit for all burst modes: L-mount (Panasonic/Sigma/Leica) Talk Forum: Digital Photography Review

Expert news, reviews and videos of the latest digital cameras, lenses, accessories, and phones. Get answers to your questions in our photography forums.

www.dpreview.com

Where you will see the difference is noise when lifting deep shadows (regardless of shutter type). The reason is likely to provide faster frames-per-second.

I have been using burst in train photography and have noticed the issue in shadows. On my next train photo outing I am planning to do my own "manual" burst, i.e. back-button focus with half-press AF turned off and just manually fire off a few shots. I hate having to cull multiple burst shots anyway so this will hopefully be better than my current "spray an pray" approach.

 

[CharlesH]

I'm wondering if there's an "easy" way to see in my test images the difference between 14 bits and 12 bits?

RawDigger measures RAW bit depth. I used it to show that S9 RAW is indeed 12 bits. If I wasn't so lazy I'd do some burst recording with the S5IX and confirm it is 12 bits. But I don't have any reason to doubt the manual. I think it's very unlikely you will see any difference between non-burst and burst pictures.

 

[PJD]

Thanks for your replies, Pete & Charles!

I'll download the free trial version of RawDigger to see if it reports S5iiX "burst" and "bracketed" RAW image sequences are the same in this regard, i.e.: 12-bits.

 

[PJD]

... I'll download the free trial version of RawDigger to see if it reports S5iiX "burst" and "bracketed" RAW image sequences are the same in this regard, i.e.: 12-bits.

... Interesting: I installed the latest free trial version of RawDigger on my Mac, and it reports S5iiX RAW images (single, burst, and bracketed) are all 14 bits. Of course, this is the first time I've used the app, so am I doing it wrong?

 

[CharlesH]

and it reports S5iiX RAW images (single, burst, and bracketed) are all 14 bits

Interesting. I took some burst shots (SH, electronic shutter, pre-burst off) with my S5IIx and I agree, it is 14 bits. Page 122 of the manual says 12 bits for burst mode. I wonder if this was changed with the new sw upgrade that added pre-burst. I did check the S9 again, it is 12 bits.

 

[Pete_W]

There was a similar discussion about the Lumix G95 on DPR. The G95 dropped from 12-bit to 10-bit in burst and bracketing modes. The person who was posting about this said that while the RAW file looked normal, i.e. reported that it was 12-bit, not all of the information had been written to the file so it was a "12-bit container" but some of the luminance values were empty, so in effect it was 10-bit depth and that translated into lower DR.

I am assuming the same applies to the S5II but is 14-bit dropping down to 12-bit.

 

[Pete_W]

I searched for the discussion in the MFT DPR forum and found it.

Two threads with information:

Re: Of course you can compare the files: Micro Four Thirds Talk Forum: Digital Photography Review

Expert news, reviews and videos of the latest digital cameras, lenses, accessories, and phones. Get answers to your questions in our photography forums.

www.dpreview.com

G90 sensor vs GH5 or G9: Micro Four Thirds Talk Forum: Digital Photography Review

Expert news, reviews and videos of the latest digital cameras, lenses, accessories, and phones. Get answers to your questions in our photography forums.

www.dpreview.com

Hopefully this will help with regards to what to look for in the S5II raw files.

 

[Lsake]

Thread 'Lumix S5II First Impressions, from a long time S5 user'

I had a look at bit depths on my first impressions post. I mentioned the bit depth in SH burst being reported as 14 despite the manual saying it will drop to 12. In reality I did notice shadows not seeming quite as recoverable on the extreme contrast small bird pictures. But this wasn’t to the point of the files being bad, and as per the manual I was expecting a loss of bits on 30fps bursts.

 

[Lsake]

Thread 'Lumix S5II First Impressions, from a long time S5 user'
https://l-mount-forum.com/community...rst-impressions-from-a-long-time-s5-user.362/

 

[PJD]

Thanks all. But just to clarify (if anyone knows), is RawDigger (or other software) reporting firmware-updated S5ii_ RAW burst files are 14 bit containers full of 14 bit image data, or 14 bit containers with 12 bits of image data padded out to 14 bits?

And further, what's the story with bracketed RAW photos? Are they the same as burst photos, regarding 14 bits of image data vs. 12 bits padded out to 14 bits?

If the latter, I won't normally use RAW bracketing (where the camera automatically fires as fast as possible), but instead I'll take multiple single RAW shots manually, where I press the shutter a few times, manually adjusting/bracketing exposure between frames, as fast as _I_ can. I don't often need to capture fast action, so an extra split second between captures is worth it to get the full image quality I paid for from my firmware-updated S5iiX.

FWIW, I plan to pose these questions to Panasonic technical support this week. It'll be interesting to see how that goes.

 

[Pete_W]

Hi Peter,

It's all a bit too technical for me so I can't help more than the links I've provided. So I look forward to hearing what Pana tech support has to say.

My guess is that nothing will have changed in the firmware update as I suspect they would have mentioned it if they had eliminated what is a bit of a downside to image quality. I don't need the extra readout speed for what I do so I would prefer the original S5's behaviour but I can understand why they made this choice.

 

[PJD]

Thanks all. But just to clarify (if anyone knows), is RawDigger (or other software) reporting firmware-updated S5ii_ RAW burst files are 14 bit containers full of 14 bit image data, or 14 bit containers with 12 bits of image data padded out to 14 bits?

And further, what's the story with bracketed RAW photos? Are they the same as burst photos, regarding 14 bits of image data vs. 12 bits padded out to 14 bits?

If the latter, I won't normally use RAW bracketing (where the camera automatically fires as fast as possible), but instead I'll take multiple single RAW shots manually, where I press the shutter a few times, manually adjusting/bracketing exposure between frames, as fast as _I_ can. I don't often need to capture fast action, so an extra split second between captures is worth it to get the full image quality I paid for from my firmware-updated S5iiX.

FWIW, I plan to pose these questions to Panasonic technical support this week. It'll be interesting to see how that goes.

An inconclusive follow-up: Panasonic technical support sent me the following reply in response to my inquiry. I'll let you know if I hear more from them later. Cheers.

==============
"... The color bit depth in bracketed is assumed to be 14 bit. However this is not a published spec by the factory. Not sure if they will provide, but will check with them. ..."
==============