8 bits vs 32 bits? What are the advantages?

J
Posted By
jezzthisisfull
Mar 4, 2009
Views
2158
Replies
22
Status
Closed
Hi, sorry about his basic question, but, what are the advantages of working in 32 bit mode? Is a broader, deeper color range available?

Apparently some tools only work in 8 bit mode — so is 8 bit still the typical work environment?

Thanks

MacBook Pro 16” Mockups πŸ”₯

– in 4 materials (clay versions included)

– 12 scenes

– 48 MacBook Pro 16″ mockups

– 6000 x 4500 px

JD
John_Danek
Mar 4, 2009
I believe 8-bit refers to 8 bits per channel, whereas an RGB file = 24 bit color. Recently 16-bits per channel has been developed to arrive at 48 bit color. This may open up some color gamut, specifically images out of RAW or a digital camera. The additonal information then can be taken advantage of by some newer 16-bit inkjet printer RIPs and is really beneficial for photographic quality output. I would think this would be an ( ProPhoto ) RGB rendering workspace in the photographic printer. If I’m not mistaken, conventional digital press and/or offset printing would require the 16-bit data be converted to 8-bit ( 8 x 4 = 32bit CMYK ) before it is rendered via the RIP or platesetter.

So, to answer your question, yes…8-bit would still be considered your typical work environment ( depending on your work environment ). This splits the typical print workflow into two different groups: 1.) photographic ( 16-bit ), and 1.) offset printing ( 8-bit ). That’s how I understand it so far. Others will chime in, I’m sure.
AS
Ann_Shelbourne
Mar 4, 2009
Keep files that contain photographic images and gradients in 16-bit for as long as possible; and save your layered Master files as 16-bit RGB.

There are a few filters which still only work in 8-bit and you might have to convert to 8-bits if you really need to use them. But examine the possibility of using them in a separate file with a Smart Object from your original image.

32-bit is for special purposes such as when using HDR.
V
v6v6v6
Mar 4, 2009
jezzthisisfull, here’s a passage from Photoshop CS4: The Missing Manual < http://books.google.com/books?id=f38qA2yg0PYC&pg=PA43&am p;lpg=PA43&dq=how+many+colors+per+channel+32-bit&sou rce=bl&ots=yk1MHp_lAt&sig=OPINN_prhJDcUNSypzOUh1Awxb A> that shed some light on this for me. Here’s an explanation of bit depth < http://help.adobe.com/en_US/Photoshop/11.0/WSfd1234e1c4b69f3 0ea53e41001031ab64-73daa.html> on Adobe’s site.

It’s easy to get confused, especially with the mix of "bit" usage when it comes to 8-bit per channel vs. 16-bit per channel vs. a 24-bit image, etc.
AW
Allen_Wicks
Mar 4, 2009
Don’t obfuscate my ignorance… πŸ™‚
JD
John_Danek
Mar 4, 2009
Ann, could you explain "special purposes" 32-bit and HDR? I’m not familiar with it, so I’d appreciate it if you could explain it a bit further. I agree with you that staying in 16-bit as long as possible could be beneficial, especially when generated computer gradients where the transitions are much smoother. However, the concern is that when the image is converted to 32-bit ( CMYK ), or a typical 8-bit RGB, the additional bit information ( if any ) will get tossed, but will retain a smoother gradient than an original 8-bit counterpart. For instance, when someone saves their 16-bit ProPhoto RGB image ( a superwide gamut space ), then converts to 8-bit Adobe RGB, and then converts to SWOP CMYK, the additional gamut found in the ProPhoto RGB will be clipped or reduced down to the Adobe RGB gamut. But, even if you retain the 16-bit ProPhoto RGB and go straight to SWOP CMYK, the color will be clipped to fit inside the SWOP space. More color, more gamut? I’m not so sure.
CC
Chris_Cox
Mar 4, 2009
Please, keep the bit counts as per-channel to reduce confusion.

32 bit should mean 32 bits per channel (integer or floating point), not 8 bit per channel times 4 channels, or 16 bit per channel times two channels.
There are 16 bit integers, and floating point formats — not just 8 bit per channel times two channels, or 5 bits per channel times 3 channels plus a mask bit.
There are 24 bit floating point formats — not just 8 bit per channel times 3 channels. etc.

Referring to an image as 32 bit just allows for too many possible interpretations of the bits. Say what you mean: 32 bit per channel, or 8 bit per channel CMYK, or 8 bit per channel RGBA, etc.
AS
Ann_Shelbourne
Mar 4, 2009
This a huge, complicated and confusing subject which cannot be explained in a few lines here:

Trying to take your questions one by one:

32-bit and HDR: This is a procedure where you can combine a number of digital photographs which were taken with differing exposures for the highlights and for the shadows and combine them into a single image in order to retain a wider dynamic range. (The results that I have seen from the use of this technique have mostly been somewhat bizarre and perfectly hideous!).

For instance, when someone saves their 16-bit ProPhoto RGB image ( a superwide gamut space ), then converts to 8-bit Adobe RGB, and then converts to SWOP CMYK, the additional gamut found in the ProPhoto RGB will be clipped or reduced down to the Adobe RGB gamut.

We need to break that down a bit:

Saves their 16-bit ProPhoto RGB image ( a superwide gamut space ): Correct.

Then re-opens [or duplicates], the Saved 16-bit ProPhoto RGB image and converts to 8-bit Adobe RGB:

That produces an RGB file which has been "clipped" to fit the smaller AdobeRGB space Β— but it is still a 16-bit file.

and then converts to SWOP CMYK:

Now you have CONVERTED your RGB colors to generate printing plates that will produce as close a replica as you can get to your Soft-preview of your (current) AdobeRGB image when printed on a Press which EXACTLY matches the conditions dictated by the CMYK Profile that you used.

But you still have a 16-bit per Channel file Β— UNTIL you change the Image mode to 8 Bits.

if you retain the 16-bit ProPhoto RGB and go straight to SWOP CMYK, the color will be clipped to fit inside the SWOP space.

Not exactly: Your RGB image will be mapped to individual C, M, and Y Plates, and a Black plate will be generated (using the attributes of your specified CMYK Profile concerning GCR and Total Ink) in order to give you the best reproduction on Press (when real INK hits PAPER!) of your original RGB image.

But do read more on this subject because it is supremely important that you understand it fully.

I highly recommend Dan Margulis’ "Professional Photoshop" for gaining a better understanding of this subject.
J
jezzthisisfull
Mar 4, 2009
thanks, those links in particular were very helpful

"All tools in the toolbox, except the Art History Brush tool, can be used with 16&#8209;bpc images."

actually it was the Art History Brush tool which led me to ask this question

since I am drawing 2D images and use liquify and other filters semi-frequently I think I’ll just stick to 8 bit for now…maybe switch with CS 6 or 7 in the future…

Unless someone would suggest a better workflow? These are huge 4k images with 50+ layers and many smart objects…almost all drawn with brush tools.

So, sticking with 8 bits should be fine?

Thanks so much again…all replies were very helpful
NK
Neil_Keller
Mar 4, 2009
These are huge 4k images

Are you sure you’ve got that number correct? 4k images weren’t even huge when we used floppy discs. <g>

Neil
JM
J_Maloney
Mar 4, 2009
4K is a cinema resolution, between 9 and 12 MP. Wiki. <http://en.wikipedia.org/wiki/List_of_common_resolutions>
NK
Neil_Keller
Mar 4, 2009
Thanks, J.

I didn’t see any reference to cinema from the OP, so I asked. I thought he might have meant either 4MB or 4GB still imagery.

Neil
J
jezzthisisfull
Mar 4, 2009
So if I were using basically just photoshop’s paintbrush tools and a few filters such as liquify to create 2D images, 8 bit for now should be fine?

I am not missing many colors when painting in 8 bit am I ??

….and, a 4,000 by 4,000 pixel image is considered large, right?



As a separate quick question: is there any way to pin open the color palette window so a person can move back and forth from drawing on a photoshop document to the color palette for quick color selection, rather than having to click upon the foreground/background squares in the toolbox. –And, rather than trying to pick using the smaller palette evoked by F6.

Thanks again…
AS
Ann_Shelbourne
Mar 4, 2009
Consider saving your colors in the Swatches Panel; and investigate the Kuler application and panel (Window menu/Extensions.
B
Buko
Mar 4, 2009
a 4,000 by 4,000 pixel image is considered large, right?

No.
J
jezzthisisfull
Mar 4, 2009
"Painting will be much smoother in 16-bits per channel Β— particularly if you paint with Fades or use Gradients. "

Wow!! That’s good to know. Shesh, you folks are the best!!! Thanks for the help and tips, again.

"4k is not large"

Ok, let’s say a person were creating professional high-res images for magazine covers, movie posters, or even giclees the size of typical museum paintings (maybe a meter square?)…what would be the best resolution? (…at 300 dpi?)

The "giclees" could be composites of smaller high-quality magazine-cover-sized images, I guess (since my standard computer may not be able to handle 1 meter sized documents of 300 dpi???). –For really nice clear high-quality images, at a dpi resolution which would look good even with future high-res/HD prints and so on, would 16 bit, 7272 by 4494, at 300 dpi be sufficient? (I’d rather not rework these images too much in the future to take advantage of whatever advances there may be in bit depth, printing, and so on…)

Thanks so much, sorry about the basic starter questions…
AS
Ann_Shelbourne
Mar 5, 2009
Let’s take the problem the other way round.

You can make these calculations mathematically Β— or you could go Photoshop/File/New.

First decide on the largest size of paper on which you want to Print; and enter that in the boxes in the New file dialog.

Then choose a suitable resolution (240 ppi should suffice for GiclΓ©e/Inkjet prints).

Choose to create a 16-bit RGB file and click OK.

You have now set-up a document which will print well on the size of paper that you need and you can start to paint.

For output to smaller sheets, just changing the dimensional size and letting the resolution take care of itself (no resampling) will often be the only changes that you need to make.
CC
Chris_Cox
Mar 5, 2009
4k x 4k is just slightly larger than web resolution — somewhere between small and average.

When you can’t fit a single flat document on a DVD-R, then it’s large.
J
jezzthisisfull
Mar 5, 2009
Thank you thank you thank you all!!
NK
Neil_Keller
Mar 5, 2009
When you can’t fit a single flat document on a DVD-R, then it’s large.

So we’re talking about 4.38 GB… <g>

Neil
WZ
Wade_Zimmerman
Mar 5, 2009
32 bit is useful if and when the capture is also 32 bit as well as the output.

It can really be useful for video and other digital capture and 4K for video is already surpassed by 5K and 6K for video not the big thing will be to go to 32 bit as well this will mean one really important missing capability of digital capture detail in the highlight and shadow areas.

That is instead of noise even at very high ISO you will actually more then acceptable detail and little or no noise. It i not so important to understand why as much as how to and when to use it as capturing in 32 bit will reduce the number of images you can capture on your media but since there are now flash cards that are 64GB then it is now possible for still photography but impossible for video so until it is practical for video it will not be available for still photography.

At least that is how I see it.
JD
John_Danek
Mar 5, 2009
I see no reference HDR in Margulis’ glossary. Ann, are you talking about Chap. 15? Isn’t that a discussion of fake profiles where he combines 2 files in an adjustment layer? I’ve taken a look at Adobe’s 32 bit and it appears they are discussing video images. I myself thought HDR ( high dynamic range ) meant a capture of 4.0 shadow detail ( i.e., scanners and cameras ).
AS
Ann_Shelbourne
Mar 5, 2009
Sorry, if I confused you!

My reference to Margulis’ book was purely in connection with the relationship between the RGB space and the creation of CMYK separations.

HDR and 32-bit Channels is an entirely different issue and you can learn more about that in Photoshop/Help/.

Must-have mockup pack for every graphic designer πŸ”₯πŸ”₯πŸ”₯

Easy-to-use drag-n-drop Photoshop scene creator with more than 2800 items.

Related Discussion Topics

Nice and short text about related topics in discussion sections