Kaj E
Forum Pro
But the behavior of light is. There is nothing software can do about that.
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Lossless compressed 12-bit v.s. Lossless compressed 14-bit
- Thread starter Shangri La
- Start date
-
- Tags
- nikon d750
Manzur Fahim
Senior Member
I always shoot Uncompressed 14-bit. I like the word "lossless", but just not sure why Nikon also provides uncompressed option if it is truly lossless. And since I don't have problem with storage, I keep it uncompressed 14-bit, and have the peace of mind and not go through the life long debate of uncompressed and lossless in my head while shooting.
MisterHairy
Senior Member
As true as that may be, the behaviour of light and the behaviour of a compression algo are mutually exclusive.
michaeladawson
Forum Pro
You're free to do whatever you please. The reason Nikon provides an uncompressed option has nothing to do with whether lossless is truly lossless (which it is). Go ahead and keep on doubting though. It's clear no one will ever change your mind on this.
Kaj E
Forum Pro
As is evident from this thread there is a good reason to understand why Nikon provides three option. Each option has it's convinced audience:
The conspiracy theorists shoot uncompressed.
People with an understanding of compression shoot lossless compressed.
People with a basic understanding of physics shoot ("lossy") compressed.
gstadter
Well-known member
Shangri La
Leading Member
- Messages
- 785
- Reaction score
- 435
Thank you all guys. It has been quite a reading and learning. Now I start to realize courses like digital signal processing taken at college can be actually useful lol.
Osvaldo Cristo
Senior Member
You can recover more shadow details, if you need, from a 14-bit raw file than a 12-bit one. Otherwise everything is the same and your 12-bit raw files will deliver the same image quality you would have from its bigger cousin.
On the other hand, as you go to higher ISO settings, the shadow area will becoming more and more noisy for the original signal pushing made by camera. Actually, for higher ISO settings the "advantage" from 14-bit raw is mean less as the "recovered shadows" contains only noise.
Although I am using some JPEGs with my current D810, my practice is to use raw only at 14-bit for ISO settings up to 6,400 (included) and I switch to 12-bit raw for ISO settings above 6,400.
All the best,
On the other hand, as you go to higher ISO settings, the shadow area will becoming more and more noisy for the original signal pushing made by camera. Actually, for higher ISO settings the "advantage" from 14-bit raw is mean less as the "recovered shadows" contains only noise.
Although I am using some JPEGs with my current D810, my practice is to use raw only at 14-bit for ISO settings up to 6,400 (included) and I switch to 12-bit raw for ISO settings above 6,400.
All the best,
What do you mean ?
There are more than one answer:
- You want to record the data exactly as they come from the sensor, without even using a fully reversible compression scheme: use 14 bit uncompressed. Caveats: smaller buffer, larger files (also longer to read and transfer),. Advantage: none, really, but, if that's the option you are comfortable with, disk space is cheap and life is short
- Same as above, but want to save some space etc. Still not comfortable with the idea of round up in high lights (see above about what you're comfortable with life is short etc.) 14 bit lossless compressed.
- Same as above, want to save as much space as possible w/o any compromise in the recorded information 14 bit lossy compressed (which Nikon refers more appropriately as "visually lossless").
- Short on card free space and being able to squeeze more shots on the card is more important than a very minor (in most cases) loss of info in the shadows use 12 bit lossy compressed.
Nothing wrong with picking a choice that you feel comfortable with, but the 2 sensible choices are the last 2, depending on the situation.
MisterHairy
Senior Member
Mutually exclusive means that one is wholly unrelated to the other. In this case, this is true.
The natural dithering of the digitised signal which occurs as a result of the inherent variations in incident photon counts has implications for the impact of the lossy compression but it does not have implications for the implementation of that compression.
MisterHairy
Senior Member
The code to allow edits of posts is not working. I am trying to adjust my text to say the following:
Mutually exclusive means that one is wholly unrelated to the other. In this case, this is true.
The natural dithering of the digitised signal (carried through from the analogue domain and controlled by the inherent variations in incident photon counts per unit area) has implications for the impact of the lossy compression but it does not have implications for the implementation of that compression.
Mutually exclusive means that one is wholly unrelated to the other. In this case, this is true.
The natural dithering of the digitised signal (carried through from the analogue domain and controlled by the inherent variations in incident photon counts per unit area) has implications for the impact of the lossy compression but it does not have implications for the implementation of that compression.
Richard Costin
Well-known member
Hi all.
I considered this question a while back and looked into the matter online and ran my own tests.
For what it's worth I now use 12 bit lossy compressed on all my Nikons all the time as I see no difference in pushed RAWS (both upwards and downwards) in regular or extreme scenes (between 14 bit uncompressed and 12 bit lossy). This was with Lightroom 4's processing engine btw.
I think the fact these threads often go on for as long as they do is also itself testament as to how little, if any difference there is. If there was a real or even minor but visible advantage it would be clearly exampled by the full quality camp guys rather than just talked about. I do remember one 'strip' example image that showed the most tiny difference once somewhere, but that is the only occasion and such an OTT extreme example that it was (for me) of no consequence and I haven't seen anything similar since.
My main reason was for best buffer clearance with file size and raw loading speed being an added bonus.
Thanks all, Richard Costin
I considered this question a while back and looked into the matter online and ran my own tests.
For what it's worth I now use 12 bit lossy compressed on all my Nikons all the time as I see no difference in pushed RAWS (both upwards and downwards) in regular or extreme scenes (between 14 bit uncompressed and 12 bit lossy). This was with Lightroom 4's processing engine btw.
I think the fact these threads often go on for as long as they do is also itself testament as to how little, if any difference there is. If there was a real or even minor but visible advantage it would be clearly exampled by the full quality camp guys rather than just talked about. I do remember one 'strip' example image that showed the most tiny difference once somewhere, but that is the only occasion and such an OTT extreme example that it was (for me) of no consequence and I haven't seen anything similar since.
My main reason was for best buffer clearance with file size and raw loading speed being an added bonus.
Thanks all, Richard Costin
gstadter
Well-known member
Shangri La
Leading Member
- Messages
- 785
- Reaction score
- 435
Manzur Fahim
Senior Member
Hi,
I've got the D750, and currently use Lightroom 4 (doesn't support D750 raw, will upgrade once Lightroom 6 is out). As such I'm currently shooting in RAW+JPEG. I'm using the fastest 95Mb/s SD card I have seen tested, but the buffer at 6.5fps is too limiting (even with just RAW) when set to 14bit lossless. On the other hand 12B bt lossless, seems to do the trick, and the buffer full rate is noticeably faster. However when I 'need' that frame rate, chances are I won't have time to nail the exposure. So wanted to see how much the extra information meant to me.
So I took a few test shots all under exposed by 5 EV, used ViewNX to convert them into 16bit Tiffs, imported them into Lightroom, and pulled them up by 5EV to the proper exposure... then went pixel peeping (camera on a tripod, aperture constant, shutter speed varying). Half with 14 bit and half with 12 bit. ISO 100, 800, 3200, 12800. The 14bit produced marginally better, (less colour noise mainly) pictures. As the ISO increased so did the difference and the 14bit started to pull away. However this was based on looking at 50%- 400% magnification, and the pictures side by side, and a relatively long time studying them.
Further to this at the higher ISO levels, both pictures were too grainy to use (hence why the camera hasn't got a Hi5.0 setting).
I'll also try setting the iso to 3200, and doing the same thing for 0EV, -1EV, -2EV and -3EV, which is probably more realistic scenarios.
Conclusion so far (may change my mind later):
There is an advantage to shooting 14bit... however it is slight, if buffer, file transfer, lag in post processing workflow, or storage space are at all issues, use 12bit. If you are in the situation, where you need to push every last ounce out of the camera, e.g. if you are doing a night scene, with artificial lights and deep shadows, meaning you are using the highlights metering (or manual, or under exposing), and buffer and memory card space isn't a problem, then it might be worth changing from 12bit to 14bit, for those few photos (just remember to change it back).
Personally I've set it up to use 14bit, apart from on U1, which I'll use for capturing birds in flight etc, where the max fps, long(ish) continual firing and quick turn around from the buffer being full do make a difference. At this point I may even slip into 1.2x or DX crop depending on the distance of the subject and likely flight path, just to up the buffer capacity further.
Side rant
The limited buffer is annoying, but I understand Nikon's choice to limit this in this way, they could probably have to charge 35-40% more for the camera, with three times the buffer space for RAW photo storage (or lower the price of their higher models, yeah right), at which point they missed their target price bracket. Also would have been nice to have U3 and U4, instead of scene and effects (hide those away in a menu somewhere and use the fn button to cycle through them in auto mode or have them assignable to a user preference, if they must have them)
I've got the D750, and currently use Lightroom 4 (doesn't support D750 raw, will upgrade once Lightroom 6 is out). As such I'm currently shooting in RAW+JPEG. I'm using the fastest 95Mb/s SD card I have seen tested, but the buffer at 6.5fps is too limiting (even with just RAW) when set to 14bit lossless. On the other hand 12B bt lossless, seems to do the trick, and the buffer full rate is noticeably faster. However when I 'need' that frame rate, chances are I won't have time to nail the exposure. So wanted to see how much the extra information meant to me.
So I took a few test shots all under exposed by 5 EV, used ViewNX to convert them into 16bit Tiffs, imported them into Lightroom, and pulled them up by 5EV to the proper exposure... then went pixel peeping (camera on a tripod, aperture constant, shutter speed varying). Half with 14 bit and half with 12 bit. ISO 100, 800, 3200, 12800. The 14bit produced marginally better, (less colour noise mainly) pictures. As the ISO increased so did the difference and the 14bit started to pull away. However this was based on looking at 50%- 400% magnification, and the pictures side by side, and a relatively long time studying them.
Further to this at the higher ISO levels, both pictures were too grainy to use (hence why the camera hasn't got a Hi5.0 setting).
I'll also try setting the iso to 3200, and doing the same thing for 0EV, -1EV, -2EV and -3EV, which is probably more realistic scenarios.
Conclusion so far (may change my mind later):
There is an advantage to shooting 14bit... however it is slight, if buffer, file transfer, lag in post processing workflow, or storage space are at all issues, use 12bit. If you are in the situation, where you need to push every last ounce out of the camera, e.g. if you are doing a night scene, with artificial lights and deep shadows, meaning you are using the highlights metering (or manual, or under exposing), and buffer and memory card space isn't a problem, then it might be worth changing from 12bit to 14bit, for those few photos (just remember to change it back).
Personally I've set it up to use 14bit, apart from on U1, which I'll use for capturing birds in flight etc, where the max fps, long(ish) continual firing and quick turn around from the buffer being full do make a difference. At this point I may even slip into 1.2x or DX crop depending on the distance of the subject and likely flight path, just to up the buffer capacity further.
Side rant
The limited buffer is annoying, but I understand Nikon's choice to limit this in this way, they could probably have to charge 35-40% more for the camera, with three times the buffer space for RAW photo storage (or lower the price of their higher models, yeah right), at which point they missed their target price bracket. Also would have been nice to have U3 and U4, instead of scene and effects (hide those away in a menu somewhere and use the fn button to cycle through them in auto mode or have them assignable to a user preference, if they must have them)
michaeladawson
Forum Pro
Interesting.
I would repeat your comparison test with 14 bit lossy compressed files. Nikon's lossy compression algorithm is very intelligent. Your file size saving are not quite as big as 12 bit lossless, but I would wager you will get better results in post processing.
I would repeat your comparison test with 14 bit lossy compressed files. Nikon's lossy compression algorithm is very intelligent. Your file size saving are not quite as big as 12 bit lossless, but I would wager you will get better results in post processing.
Keyboard shortcuts
- f
- Forum