login info join!
Forums > Digital Art and Retouching > 8bit vs. 16bit. When logic goes out the window Search   Reply
Retoucher
Lanenga
Posts: 842
Amsterdam, Noord-Holland, Netherlands


Lately I have been testing wether working in 16bit  mode will always result in better quality images and the result is interesting to say the least.

The problem occurs at the end stage when you will have to go back to an 8bit/channel output.  Until then, the 16bit will win hands down, but after the conversion back to 8 bit, sometimes the image processed in 8bit mode looks better than the one processed in 16bit mode and converted back to 8bit mode.

This might be due to color spaces as well. But has anyone else seen this happen? Any comments, advice, solutions?
May 11 12 12:03 pm  Link  Quote 
Retoucher
Ledo retouch
Posts: 889
Lodi, California, US


I've wondered about this myself, damage to 8bit files is cumulative,
the bigger the changes, the more difference a 16bit file will make.
Most images are not destined for print and are displayed on an 8bit
monitor. As with most things, it depends on the image and use, most
of my workflow is 16bit through frequency separation then a 8bit copy
to finish.
May 11 12 12:41 pm  Link  Quote 
Photographer
Ruben Vasquez
Posts: 3,038
Puyallup, Washington, US


I might be one of the few who think this way but I don't really see an advantage working exclusively in 16 bit rather than 8 bit unless you also work with computer generated images such as painting in an eyelash, adding noise or a gradient; which is something I frequently do.

If the only thing you're doing is manipulating the image in an, I guess, more organic fashion (such as curves or b&w, etc), without the use of cgi, then I have yet to see a 16 bit image manipulated in the exact same way as an 8 bit image and come out looking significantly better. In a side by side comparison, I honestly can't tell the difference between the two without someone telling me outright which one is which.
May 11 12 12:42 pm  Link  Quote 
Photographer
Luminos
Posts: 6,055
Columbia, Maryland, US


Lanenga wrote:
Lately I have been testing wether working in 16bit  mode will always result in better quality images and the result is interesting to say the least.

The problem occurs at the end stage when you will have to go back to an 8bit/channel output.  Until then, the 16bit will win hands down, but after the conversion back to 8 bit, sometimes the image processed in 8bit mode looks better than the one processed in 16bit mode and converted back to 8bit mode.

This might be due to color spaces as well. But has anyone else seen this happen? Any comments, advice, solutions?

Working in a bigger room will always give you more opportunity to grow the effort and work more efficiently.   

Great work can be done in a small room but will present greater challenges to the worker.

And sometimes the smaller room can force on the worker conditions that turn out to be to the advantage of someone who would have committed an error given more room.

Working in 16 bits, particularly when starting with more than 8 bits in the first place, will always provide more technical room to work.   The key is to know what to do with the room you have, or know if you even need all that space.

May 11 12 12:49 pm  Link  Quote 
Photographer
ChanStudio - OtherSide
Posts: 5,309
Alpharetta, Georgia, US


May 11 12 12:57 pm  Link  Quote 
Photographer
Sergei Rodionov
Posts: 865
Dallas, Texas, US


Ledo retouch wrote:
Most images are not destined for print and are displayed on an 8bit
monitor.

Most of typical printers do work in 8bit Jpegs btw..

But if you extend that logic wit monitors - you dont need 8 either b/c your monitor is actually 6 bits, most likely smile

Point is its like rounding error.. More headroom you got - less mess and errors you will get in the end. Gradients smoother & etc.

Anyway. Part of reason why you might not see much difference is because you dealing with 12 bit files from amateur dSLR.. difference becomes more when you got into 14.. and just kicks you in crotch when you deal with 16 bit RAWs from MDFB

May 11 12 01:08 pm  Link  Quote 
Retoucher
Lanenga
Posts: 842
Amsterdam, Noord-Holland, Netherlands


ChanStudio - OtherSide wrote:
Maybe this would help:

http://www.photoshopessentials.com/essentials/16-bit/

And how would this help me?
I know what bit depth is and how to work in them.

The information is not even accurate. Photoshop 16bit mode does not give you a value range of 16bit (2^16) it only gives you a value range of 15bit+1 ((2^15)+1)

Luminos wrote:
Working in a bigger room will always give you more opportunity to grow the effort and work more efficiently.   

Great work can be done in a small room but will present greater challenges to the worker.

And sometimes the smaller room can force on the worker conditions that turn out to be to the advantage of someone who would have committed an error given more room.

Working in 16 bits, particularly when starting with more than 8 bits in the first place, will always provide more technical room to work.   The key is to know what to do with the room you have, or know if you even need all that space.

Maybe my question was not clear enough, but I am not asking if I should be working in 8bit vs 16bit mode or what the difference is between these modes. I am talking about the fact that on an 8bit output, the 16bit edit does not always seem to give the better result.

May 11 12 01:18 pm  Link  Quote 
Photographer
Photographe
Posts: 2,350
Bristol, England, United Kingdom


Lanenga wrote:
Lately I have been testing wether working in 16bit  mode will always result in better quality images and the result is interesting to say the least.

The problem occurs at the end stage when you will have to go back to an 8bit/channel output.  Until then, the 16bit will win hands down, but after the conversion back to 8 bit, sometimes the image processed in 8bit mode looks better than the one processed in 16bit mode and converted back to 8bit mode.

This might be due to color spaces as well. But has anyone else seen this happen? Any comments, advice, solutions?

I used 8 bit and indexed colour for a lot of games design and development textures.

Definitely I notice problems when switching between modes from 16 bit to 8 bit. Sometimes a texture can darken. Sometimes text is horrible.
But then PS disallows so many functions when you are in 8-bit or so it seems.

May 11 12 01:25 pm  Link  Quote 
Photographer
ChanStudio
Posts: 9,155
Alpharetta, Georgia, US


Lanenga wrote:

And how would this help me?
I know what bit depth is and how to work in them.

The information is not even accurate. Photoshop 16bit mode does not give you a value range of 16bit (2^16) it only gives you a value range of 15bit+1 ((2^15)+1)


Maybe my question was not clear enough, but I am not asking if I should be working in 8bit vs 16bit mode or what the difference is between these modes. I am talking about the fact that on an 8bit output, the 16bit edit does not always seem to give the better result.

It isn't because 16 bits doesn't give better results.  Keep in mind that your monitor is only 8 bits and 10 bits (if you have great monitor with display port connected).   With 16 bits, you have lots of rooms to move your data around.  If you have 8 bits, you only have limited space and when you move data round (anything that is outside of that 8 bits zone), you going to lose data.  Once data is lost, you won't be able to get it back.


On the monitor I use (connected using Display Port), I don't see the differences between 8 bits and 16 bits until I start moving data around.  I do see banding when I manipulate 8 bits files around (using 8 bits mode).



By the way, unsigned 16 bits is 2^16.  That would give you 65,536.

May 11 12 01:34 pm  Link  Quote 
Retoucher
Lanenga
Posts: 842
Amsterdam, Noord-Holland, Netherlands


I realize I have forgot to mention a few things.

1. I am not talking about photographs only.
2. An image of 14bit/channel converted to 8 bit/channel will of course never give the same result as a 14bit/channel image converted to 16 bit/channel

What I am talking about is an 8 or less bit/channel image.
Once edited in a 8bit/channel workflow.
And once edited in a 16bit/channel workflow.

This way I would be able to see the difference in editing in 8bit mode vs 16bit mode.

Now, while editing, the differences are as expected. The large 16bit/channel bit depth gives a visual better result.

But then when done editing and both edits are being output to an 8 bit medium the advantage of working in 16bit mode seems to disappear sometimes.

And you might think, well DUH you are throwing away a bunch of information. And while that being true, I would not expect the result of the 8bit workflow to be better than that of the 16bit workflow and then throwing away part of the information.

Hope this makes it a bit clearer as to what I am talking about and what I am trying to see if others have come across the same issues.
May 11 12 01:37 pm  Link  Quote 
Retoucher
Lanenga
Posts: 842
Amsterdam, Noord-Holland, Netherlands


ChanStudio wrote:
It isn't because 16 bits doesn't give better results.  Keep in mind that your monitor is only 8 bits and 10 bits (if you have great monitor with display port connected).   With 16 bits, you have lots of rooms to move your data around.  If you have 8 bits, you only have limited space and when you move data round (anything that is outside of that 8 bits zone), you going to lose data.  Once data is lost, you won't be able to get it back.


On the monitor I use (connected using Display Port), I don't see the differences between 8 bits and 16 bits until I start moving data around.  I do see banding when I manipulate 8 bits files around (using 8 bits mode).

Thank you for the explanation, but this is not what I was talking about.

ChanStudio wrote:
By the way, unsigned 16 bits is 2^16.  That would give you 65,536.

I know. I didn't say anything about this.
I said photoshop is not giving you a value range of 2^16 = 65536, but of 2^15  = 32678

May 11 12 01:40 pm  Link  Quote 
Photographer
Frozen Instant Imagery
Posts: 3,584
Dublin, Dublin, Ireland


Sergei Rodionov wrote:

Most of typical printers do work in 8bit Jpegs btw..

But if you extend that logic wit monitors - you dont need 8 either b/c your monitor is actually 6 bits, most likely smile

Point is its like rounding error.. More headroom you got - less mess and errors you will get in the end. Gradients smoother & etc.

Anyway. Part of reason why you might not see much difference is because you dealing with 12 bit files from amateur dSLR.. difference becomes more when you got into 14.. and just kicks you in crotch when you deal with 16 bit RAWs from MDFB

There are plenty of true 8-bit screens (some with hardware calibration), and even some 10-bit screens. What makes you assume that the OP is working on a 6-bit screen? It would be more generous to assume that someone who is serious about retouching would be using a sensible choice of monitor.

I do agree with you about rounding and other errors - using 16 bit lets you keep the accumulating errors in the low order bits, so they can disappear when you discard them to go to 8 bit. And I agree with you that 14-bit RAW gives you more latitude for manipulation, but it's unkind to suggest that the OP is using "12 bit files from amateur DSLR". There are pro and semi-pro DSLRs which produce 12 bit files, like the original Canon 5D (people still use them), and Sony's high end Alpha (like the A900).

May 11 12 02:10 pm  Link  Quote 
Photographer
Sean Baker Photo
Posts: 8,036
Silver Spring, Maryland, US


Can you show us an example of what you're talking about?  I understand what you're saying; just can't visualize the result.
May 11 12 02:47 pm  Link  Quote 
Retoucher
George Thomson
Posts: 695
Concord, California, US


I'm not sure what you mean, but PS converts 15bit to 8 by dithering. Theoretically it may "add" some noise, but otherwise it shouldn't affect the image.
May 11 12 03:24 pm  Link  Quote 
Photographer
Sergei Rodionov
Posts: 865
Dallas, Texas, US


Frozen Instant Imagery wrote:
There are pro and semi-pro DSLRs which produce 12 bit files, like the original Canon 5D (people still use them), and Sony's high end Alpha (like the A900).

1) Canon 5D is amateur camera. I am sorry, but this is true. There is no such thing as "semi-pro". Its advanced amateur.
And there are plenty of old pro dSLR that produce 12 bit files too, mind you. I might be wrong but there is very very good chance that majority of RAW files are still 12 bit.

2) With regard on what things displayed at. How many people you personally know, who do not do graphic design/retouching/photography and use IPS based monitors? How many of them using newer cheaper macs with 6 bit monitors? Orignal point was that its getting displayed on web in 8bit, which is true. But average Joe's monitor is even poorer than 8 bit still. Once all civilized world upgraded - then year.. smile We all will do with 8+ bit on monitors and do happy dances..
(want true horror? try to look at images on Accer cheap flat panels.. smile)

May 11 12 03:40 pm  Link  Quote 
Photographer
Frozen Instant Imagery
Posts: 3,584
Dublin, Dublin, Ireland


Sergei Rodionov wrote:

1) Canon 5D is amateur camera. I am sorry, but this is true. There is no such thing as "semi-pro". Its advanced amateur.
And there are plenty of old pro dSLR that produce 12 bit files too, mind you. I might be wrong but there is very very good chance that majority of RAW files are still 12 bit.

2) With regard on what things displayed at. How many people you personally know, who do not do graphic design/retouching/photography and use IPS based monitors? How many of them using newer cheaper macs with 6 bit monitors? Orignal point was that its getting displayed on web in 8bit, which is true. But average Joe's monitor is even poorer than 8 bit still. Once all civilized world upgraded - then year.. smile We all will do with 8+ bit on monitors and do happy dances..
(want true horror? try to look at images on Accer cheap flat panels.. smile)

Considering just the Canon range of DSLRs, RAW files from ALL current models are 14 bit. (The first model to offer 14 bit RAW was the 1D Mark III, in 2007, and there have been a lot of 14-bit models introduced since then) That tends to argue against your suggestion that the majority of RAW files are still 12 bit.

I don't accept your suggestion that there's no such thing as semi-pro. (I can be just as blunt as you smile ). It's arguable that every range of DSLRs has a range of "pro-ness", with some models offering few features for professional photographers, and some offering a lot of features intended for, and useful to, professionals. (Oh, picture the arguments about the amount of "pro-ness" in a new model)

I do accept that many people are using monitors with poor colour handling. Doesn't mean we shouldn't strive for the best quality we can reasonably achieve. Even if it's only one person in a thousand, or one person in a million, who can appreciate the quality of the work, it's worth it. After all, one might hope that the people to whom colour fidelity is important would be the ones using good monitors smile

Yes, I'm an optimist.

May 11 12 04:03 pm  Link  Quote 
Retoucher
George Thomson
Posts: 695
Concord, California, US


.. it's not how big it is, it's how you use it, is what makes something a "pro"

session for Elle:
http://www.youtube.com/watch?v=lbP8hKzhRKY&hd=1
May 11 12 04:26 pm  Link  Quote 
Retoucher
Robert LC
Posts: 944
Amsterdam, Noord-Holland, Netherlands


I did some in-depth tests (perhaps 8bit too technical for some, sorry)
http://www.youtube.com/watch?v=-4VbfBxsGCQ
May 11 12 04:56 pm  Link  Quote 
Retoucher
Lanenga
Posts: 842
Amsterdam, Noord-Holland, Netherlands


Figured it out!
My testing was flawed big_smile

Thanks everyone!
May 12 12 03:37 am  Link  Quote 
Photographer
rey sison photography
Posts: 1,805
Los Angeles, California, US


Lanenga wrote:
Figured it out!
My testing was flawed big_smile

Thanks everyone!

LOL Lanenga. So what was the result of the testing?

May 12 12 10:46 am  Link  Quote 
Photographer
Artifice
Posts: 30,945
Los Angeles, California, US


Don't use 8 bits for 3D CG displacement maps - I can say that much

http://www.digitalartform.com/archives/ … 16_bi.html
May 12 12 11:00 am  Link  Quote 
Retoucher
Lanenga
Posts: 842
Amsterdam, Noord-Holland, Netherlands


rey sison photography wrote:
LOL Lanenga. So what was the result of the testing?

The initial result was that working in 8bit mode would sometimes give a better result than working in 16bit mode.

Even if this were true for only a few instances, it would mean you would have to know beforehand what you were after and going to do to a shot to determine if you should be working in 8bit mode or 16bit mode.

Was so surprised by that, that I had to know if others had seen this too. (hence the "When logic goes out the winow")

But, I'd missed a little color profile conversion at the end of my test workflow that was causing the perceived "beter" results smile

In the end it was a color profile issue and not a bit depth issue...

*double face palm*

EDIT:
Some credit should go to FLEXmanta for always being their to slap some sense into me again when it comes to color profiles

May 12 12 11:36 am  Link  Quote 
Retoucher
FLEXmanta
Posts: 1,001
Madrid, Madrid, Spain


Lanenga wrote:

The initial result was that working in 8bit mode would sometimes give a better result than working in 16bit mode.

Even if this were true for only a few instances, it would mean you would have to know beforehand what you were after and going to do to a shot to determine if you should be working in 8bit mode or 16bit mode.

Was so surprised by that, that I had to know if others had seen this too. (hence the "When logic goes out the winow")

But, I'd missed a little color profile conversion at the end of my test workflow that was causing the perceived "beter" results smile

In the end it was a color profile issue and not a bit depth issue...

*double face palm*

EDIT:
Some credit should go to FLEXmanta for always being their to slap some sense into me again when it comes to color profiles

If you are confusing the visual changes of converting to another profile and resampling a file by changing its bitdepth, it means you still dont understand the visual advantages of working in 16bits.

-You will not notice the differences between 16 and 8 bits until you zoom way in.
-For noticing the differences of risky color profile conversions, you need to zoom out.

You said sometimes working straight in 8 bits is better. Define better (now i know you meant that the colors were more visually attractive, which is not a property of bitdepth).

May 12 12 02:41 pm  Link  Quote 
  Search   Reply



main | browse | casting/travel | forums | shout box | help | advertising | contests | share | join the mayhem

more modelmayhem on: | | | edu

©2006-2014 ModelMayhem.com. All Rights Reserved.
MODEL MAYHEM is a registered trademark.
Toggle Worksafe Mode: Off | On
Terms | Privacy | Careers