16-bit vs 8-bit colors in PS.... Any difference?

[David]

Ok, the subject title is kinda lame... Definitely 16-bit has a lot more colours. But is it noticeable compared to 8-bit? Not sure if this has been discussed before...

...but I notice that whether I choose 8-bit or 16-bit when I convert my RAW image into Photoshop, I can't tell the difference. Or even when I convert a 16-bit image to 8-bit in Photoshop, the change is not noticeable!

So do you guys use 16-bit or 8-bit? Why is there so much hype about 16-bit? Good in theory but may not offer so much benefit in practice?

How about printing.... Does a 16-bit image look more vibrant or punchy compared to 8-bit? Problem with 16-bit is that jpg is not available and fle size is twice as large.

Pls chip in!!

 

[Buckteeth LCL]

Give you some links so you can read up a bit on yourself

http://www.earthboundlight.com/phototips/8bit-versus-16bit-difference.html

http://www.photoshop-tutorials-plus.com/8-bit-and-16-bit.html

http://www.photoshopessentials.com/essentials/16-bit/page-2.php

Let us know if you need any explanation on the terms and concepts

 

[theRBK]

its not really more colour in the sense of a wider range of colour... its more colour in the sense that within the same amount of colour available, the colour can be more finely differentiated...

let me give an analogy: 8bit colour is like we have $1 value made up of 10x 10¢ coins, whereas 16bit colour is like $1 value made up of 100x 1¢ coins... its still $1 value but when using 1¢ coins, the $1 value is more finely differentiated

16bit colour would allow us to make drastic changes in brightness, contrast and colour shifts while minimizing colour breakup and banding... if the image does not need drastic changes to be made, then working in 8bit would certainly save time as the files are much smaller and easier to process...

as for prints, the differences would be hardly visible to the human eye, that is if the printer can even process 16bit files (printers usually just convert it to 8bit files before the data is processed for printing)

 

[fastshot]

Your entire display system is probably and typically 8 bits per color, that's why you won't notice anything different when your file is converted to 16 bits.

 

[lightbrush]

Ok, the subject title is kinda lame... Definitely 16-bit has a lot more colours. But is it noticeable compared to 8-bit? Not sure if this has been discussed before...

...but I notice that whether I choose 8-bit or 16-bit when I convert my RAW image into Photoshop, I can't tell the difference. Or even when I convert a 16-bit image to 8-bit in Photoshop, the change is not noticeable!

So do you guys use 16-bit or 8-bit? Why is there so much hype about 16-bit? Good in theory but may not offer so much benefit in practice?

How about printing.... Does a 16-bit image look more vibrant or punchy compared to 8-bit? Problem with 16-bit is that jpg is not available and fle size is twice as large.

Pls chip in!!

The difference will be noticeable when you start doing photo-editing. Every change you make will bruise the pixels. To see this, when you do editing, keep the histogram opened. In an 8-bit image, you will soon see breaks in the histogram as you edit. In the 16bit image, the breaks are much lesser. The cumulative effect of this is that your final image will suffer more from an 8-bit image than a 16-bit one. The key word here is "cumulative"....hope this helps

 

[tirza]

The difference will be noticeable when you start doing photo-editing. Every change you make will bruise the pixels. To see this, when you do editing, keep the histogram opened. In an 8-bit image, you will soon see breaks in the histogram as you edit. In the 16bit image, the breaks are much lesser. The cumulative effect of this is that your final image will suffer more from an 8-bit image than a 16-bit one. The key word here is "cumulative"....hope this helps

bro so u mean.. before edit the photos change it to 16bits first? to expand its colour range when editing?

 

[lightbrush]

bro so u mean.. before edit the photos change it to 16bits first? to expand its colour range when editing?

yes, that shld be part of your workflow. You set it down to 8 bit when you want to print. But if your printer can accept 16 bit, just leave it as it is.

 

[yqt]

bro so u mean.. before edit the photos change it to 16bits first? to expand its colour range when editing?

I'm no expert but I think if you shoot in 8 bits and than convert it to 16 bits, it will still be an 8 bits images, the original is still an 8 bit image. Not the same as editing an image shot at 16 bits.

Please correct me if I'm wrong.

 

[theRBK]

there are certain situations where changing an 8bit file to a 16bit file does seem to improve quality, like when combating banding and posterization by converting an 8bit file to 16bit mode before adding noise to smoothen it and then giving it a slight gaussian blur, but for the most part, the additional computing power needed for 16bit files, and thus slower response, seems not to result in any significant advantage... YMMV and personally, if a file is 8bit, I wouldn't bother unless I see an image breaking up... of course, personally, I would rather work on a 16bit workflow from RAW to final output...

 

[lightbrush]

I'm no expert but I think if you shoot in 8 bits and than convert it to 16 bits, it will still be an 8 bits images, the original is still an 8 bit image. Not the same as editing an image shot at 16 bits.

Please correct me if I'm wrong.

If you shoot in 8 bits and then convert it 16bits, it makes no difference if that is your final output.However, if you intend to edit it,then it may make a diff. The reason is as you edit(color saturation, hue, etc), the color undergo changes and if it hits a range that is out of gamut,then you "loose" color. Over time, these translates into the gaps that you see in the histogram.

If possible, try to shoot in 16bits and keep it in 16 bits all the way to preserve the image information.

 

[theRBK]

If you shoot in 8 bits and then convert it 16bits, it makes no difference if that is your final output.However, if you intend to edit it,then it may make a diff. The reason is as you edit(color saturation, hue, etc), the color undergo changes and if it hits a range that is out of gamut,then you "loose" color. Over time, these translates into the gaps that you see in the histogram.

If possible, try to shoot in 16bits and keep it in 16 bits all the way to preserve the image information.

gamut is not the same as bit depth... gamut is the range of colours available to a colour profile, whereas bit depth is how finely differentiated the range of colours may be described (like what I mentioned in the 3rd post of this thread)... if one doesn't want the colours to run out of gamut, one should work in a larger colour profile instead, like Adobe RGB, the even larger ProPhoto RGB, or even LAB...

 

[yqt]

If you shoot in 8 bits and then convert it 16bits, it makes no difference if that is your final output.However, if you intend to edit it,then it may make a diff. The reason is as you edit(color saturation, hue, etc), the color undergo changes and if it hits a range that is out of gamut,then you "loose" color. Over time, these translates into the gaps that you see in the histogram.

If possible, try to shoot in 16bits and keep it in 16 bits all the way to preserve the image information.

Hope you don't mind my asking as I'm really trying to understand this, but if a colour is "out of gamut range" in 8 bits, ie; no such colour info, I don't think it will be present even if it was converted to 18 bits, correct?

I guess the best way of editing an 18 bits image is to either shoot in 18 bits or to convert a raw image into 18 bits and go from there.

 

[RayDream]

So does it makes any sense if I download one 8bit photo into PS, change it to 16bit, change the levels, then convert it back to 8bit?

I did a comparism about adjusting levels in PS and Canon's bundled software. For some reason the bundled software can edit 8bit without losing color informations like the 16bit. Is that how they do it?

 

[indaybadiday]

The difference will be noticeable when you start doing photo-editing. Every change you make will bruise the pixels. To see this, when you do editing, keep the histogram opened. In an 8-bit image, you will soon see breaks in the histogram as you edit. In the 16bit image, the breaks are much lesser. The cumulative effect of this is that your final image will suffer more from an 8-bit image than a 16-bit one. The key word here is "cumulative"....hope this helps

correct

 

[lightbrush]

Hope you don't mind my asking as I'm really trying to understand this, but if a colour is "out of gamut range" in 8 bits, ie; no such colour info, I don't think it will be present even if it was converted to 18 bits, correct?

I guess the best way of editing an 18 bits image is to either shoot in 18 bits or to convert a raw image into 18 bits and go from there.

No prob. I'm probably quite bad in explaining. Let me try again.

At the point when you change from 8 bit to 16 bit, there is no change in the colour. That's the reason why if that is your final output,then it makes no diff. However,once you start editing it, such as when you change the hue or saturation, that 8 bit x 8 bit x 8 bit (R,G,B) may not be enough to represent the final color that should be achieved. Because of this, there is a color shift and you may get "funny" colours if you pushed it too hard.

(From the computer point of view,colours are just a series of numbers. So an 8-bit Red has 2 to the pwer of 8 ie 256 shades of Red. A 16-bit Red will have 2 to the pwer 16 ie 65,536 shades of red. Combine R,G,B and you can have either 256x256x256 colors for 8bit or 65,536 x 65,536 x 65,536 colors for 16 bit. Of course, this is further constrained by the range of colors that the physical device can take. This is the color space or color gamut but that's another story altogether).

 

[theRBK]

I think the main problematic issue is banding of colours, where the differentiation of colour is not present such that when an adjustment is made, colours sort of chunk together because the differences between colours cannot be described in 8bit... you may get concentric strips with the same colour instead of a smooth graduation of tones... whether this would appear depends on what one does to an image and how drastic any adjustments made may be...

 

[yqt]

No prob. I'm probably quite bad in explaining. Let me try again.

At the point when you change from 8 bit to 16 bit, there is no change in the colour. That's the reason why if that is your final output,then it makes no diff. However,once you start editing it, such as when you change the hue or saturation, that 8 bit x 8 bit x 8 bit (R,G,B) may not be enough to represent the final color that should be achieved. Because of this, there is a color shift and you may get "funny" colours if you pushed it too hard.

(From the computer point of view,colours are just a series of numbers. So an 8-bit Red has 2 to the pwer of 8 ie 256 shades of Red. A 16-bit Red will have 2 to the pwer 16 ie 65,536 shades of red. Combine R,G,B and you can have either 256x256x256 colors for 8bit or 65,536 x 65,536 x 65,536 colors for 16 bit. Of course, this is further constrained by the range of colors that the physical device can take. This is the color space or color gamut but that's another story altogether).

Thanks. So I guess 8 bits or 16 bits are just how the image is represented, over a range of 8 bits or a bigger range of 16 bits? Correct?

 

[shojibake]

lightbrush got it right the first time around. The reason why you get banding, is that for 8 bit images, the mathematical process of calculating the results of the operation results in some form of rounding. Too many operations, the rounding errors accumulate. With a larger word count like 16bits, there's smaller rounding errors, so you won't get as many problems.

 

[theRBK]

but an important point is, will we see any difference in the output... and if not, is there a point to working with 16 bit, especially if the original image is 8bit... don't get me wrong, personally I prefer to and do work with 16bit from start to end, but that is provided that the hardware is able to take the load of working in 16bit and doing so would not take the computer too long to process any changes, which might not be the case for everyone...

 

[theveed]

let me give an analogy: 8bit colour is like we have $1 value made up of 10x 10¢ coins, whereas 16bit colour is like $1 value made up of 100x 1¢ coins... its still $1 value but when using 1¢ coins, the $1 value is more finely differentiated

Great analogy :thumbsup: