8/16 bit images! What's the difference !

[john_opitz]

http://www.ledet.com/margulis/PP7_Ch15_Resolution.pdf

The 8/16 bit article starts on page 310 of this pdf. file.

[Ed_L]

Good question John. Maybe I'm getting older, and can't see the difference. But I've read a lot of text supporting the 16 bit per channel option, and I've tried several scans at 8 bit, and 16 bit without being able to see a difference. So to my mind, unless there's a strong case for using 16 bits (a lack of tonal structure), I see it as a waste of space, and a loss of Photoshop features. So far, I haven't found an image where I could see the difference.

Ed

[john_opitz]

I have to agree with you Ed. On the dcs 760 camera I use. I see no difference in 8 or 16 bit, except....the 16 bit is a little bit softer(louping the print). So I could go either way. Its the editing(in P.S.). As you stated. That's the problem. Their are ways of editng 16 bit in 8 bit mode(selecting and pasting) but it's a pain in the neck. Too time consuming and nothing to gain from it. So I'm not a 16 bit die hard.

[Trimoon]

As a printer I can see the difference. Hear is how Adobe explains it
Bit depth--also called pixel depth or color depth--measures how much color information is available to display or print each pixel in an image (more bits of information per pixel) means more available colors and more accurate color representation in the digital image.
A pixel with a bit depth of 1 has two possible values white and black.
A pixel with a bit depth of 8 has 28, or 256, possible values.
A pixel with a bit depth of 24 has 224, or roughly 16 million, possible values.
Common values for bit depth range from 1 to 64 bits per pixel.
In most cases, Lab, RGB, grayscale, and CMYK images contain 8 bits of data per color channel. This translates to a 24-bit Lab bit depth (8 bits x 3 channels); a 24-bit RGB bit depth (8 bits x 3 channels); an 8-bit grayscale bit depth (8 bits x 1 channel); and a 32-bit CMYK bit depth (8 bits x 4 channels). Photoshop can also read and import Lab, RGB, CMYK, and grayscale images that contain 16 bits of data per color channel.
I reproduce art for the artist and I need all the info I can get.

[shag-man]

I can't find the difference. The other thing. If Adobe says that 16 bits is better, then why doesn't PhotoShop support adjustment layers,filters,ect..ect.. for 16 bits.

[Trimoon]

One reason is most system can’t handle the file size
“A 16-bit-per-channel image provides finer distinctions in color, but it can have twice the file size of an 8-bit-per-channel image”

[Tim_S]

A warning about Dan Margulis

Since I consider myself, at least well read on this subject, if not actualy knowlegable about it, I'd like to point something pertinent.
Dan Margulis in not unbiased about the 8-bit vs 16-bit debate.

Take a look at Bruce Lindbloom's site, go to imaging and look for the 16-bit challange article... it is an interesting read.

--tks

[Stephen M]

You have to have a lot of fortitude to catch up with this whole debate, if you were not 'lucky' enough to be there and to follow the natural progression, as I have over time.

I would suggest reading the ACT archives for the first hand info as it unfolded (membership reqd):

http://groups.yahoo.com/group/colortheory

Or perhaps these two summaries:

http://www.ledet.com/margulis/ACT_postings/ColorCorrection/ACT-8-bit-16-bit.htm

http://www.ledet.com/margulis/ACT_postings/ColorCorrection/ACT-16-bit-2002.htm

These links in addition to the previous posts link should bring everyone up to speed on Dans latest offering:

http://www.ledet.com/margulis/PP7_Ch15_Resolution.pdf

Save this for when you have a LOT of spare time, then do your own tests as I have (I think each user should come to their own conclusion for their own setup and workflow based on their own tests).

Stephen Marsh.

[Doug Nelson]

Get with it, the new debate is 16/32

http://www.idruna.com/photogenicshdr.html

http://www.cybergrain.com/tech/hdr/

[Ed_L]

Very interesting to say the least! Thanks for the link.

Ed

[gmitchel]

The important issue is not whether you can see a difference when viewing a scanned image or one just loaded from a digital camera. The differences between 8-bits and 16-bits per channel will almost certainly not be visible to the naked eye.

The important point about 8-bit images is that they are much more likely to posterize as you edit them. This is quite easy to see by eye on a monitor *AND*, as Trimoon notes, especially easy to see on a print.

Most 16-bit images are not 16-bits per se. Most digital camera images are 12-bits per channel. PS treats anything over 8 as 16-bits. Still, a Canon DSLR RAW image has 4,096 distinct tonal possibilities for each channel compared with 256 for an 8-bit JPEG image.

Generally, I find people who advocate 8-bit images are shooting JPEGs.

16-bit images give you a lot more editing options. I keep my images 16-bits until I am ready to print them *OR* I need a feature in PS CS that does not support 16-bits (some filters, for example).

Happy holidays!

Mitch

[Robt]

Happy Holidays Mitch,

In the post above, you indicate that you convert to 8 bit before print. I could see this for the tiff final would certianly be smaller but why otherwise.

I think my 2200 does its own 8 bit conversion but is it any worse or even different than PS? I also seem to remember that you use the IP rip, how does this figure in.

I guess I'm asking for an explation of your final [after output sharpening- if this is yur last common step prior to print] workflow. Now that sentence would have gotten me a fine grade back in college..

Robert Collins

[gmitchel]

Hi Robert:

My output files are sharpened for their intended output device. Files destined for my Epson 2200 are resampled to 280 dpi, because I use ImagePrint Lite and it will resample, if I don't.

The rationale you mention is why I convert from 16-bits to 8-bits for output files. Output to desktop printers are inherently 8-bits. So, why waste the extra harddrive storage.

My working files remain in 16-bits, unless I need a feature that is 8-bits only. Then I switch to 8-bits at the last possible moment to avoid posterizing my images.

Cheers,

Mitch

[cspringer]

16bit vs. 8 bit demonstration

http://forums.dpreview.com/forums/read.asp?forum=1006&message=10308596

[CJ Max]

http://www.ledet.com/margulis/PP7_Ch15_Resolution.pdf

The 8/16 bit article starts on page 310 of this pdf. file.

I did some test prints 2 months ago with my Epson R800. All images were 8"x10" as that's the size that I print mostly.

I was able to see a difference in images with TIFF 16-bit coming out the best.

Where's the difference? Highlights aren't quite so bright. Shadows have a wee bit more detail. Saturation is better. Color transitions are smoother looking.

The least desireable images were JPEG q.12.

Also, I found that I got better images when cropping and printing from Qimage. I called Epson and talked with a tech. rep. about my findings. He said that he owns and uses an R800 and gets the same results. We have different dSLR's, but we both came up with approximately the same workflow. Interesting.

The workflow for my Sigma SD-10 is: process RAW file in Sigma Photo Pro, save as TIFF 16-bit file, postprocess in Photoshop if needed, save as TIFF 16-bit file, crop and print TIFF 16-bit file in Qimage.

I mentioned to the Epson tech. rep. that I felt that the R800 printed better from Qimage than from Photoshop. He said that he had found that to be true too. He said that the R800 software handles the compression and printing better from Qimage than Photoshop.

Whatever the reasons, color me very pleased with my TIFF 16-bit/R800 prints. I don't think that they can get any better.

Cliff.