Operation Photo Rescue's Online Community

OPR Workshop Information Exchange => General Techniques => Topic started by: G3User on June 27, 2009, 09:35:26 AM

Title: 8 bit vs 16 bit
Post by: G3User on June 27, 2009, 09:35:26 AM
I have seen conflicting advice on whether it is worth concerting 8 bit/chan images to 16bit/chan before starting restoration. Where a lot of blending is being done it may reduce the chance of pixelisation

Does anyone have any practical experience?

Athol
Title: Re: 8 bit vs 16 bit
Post by: Hannie on June 27, 2009, 11:16:22 AM
There probably is a good use for working with 16 bits images versus 8 bits but for what I'm doing I could not see any advantage.

Some time ago I did a bit of an experiment myself with an old family photo that I had to restore.
I worked on the same photo in 8 bits and 16 bits.   I restored them both the same way and the end result showed no difference, even when enlarged to 400%, no banding or degradation of the 8 bits  image version visible with the naked eye.

I decided it isn't worth it for me to work in 16 bits images.  The file size is another possible drawback.
I also remember that there were some tools/filters in Photoshop that couldn't be used when working on an 16 bits image.  (e.g. FFT)

Hannie
Title: Re: 8 bit vs 16 bit
Post by: Mhayes on June 27, 2009, 01:07:39 PM
Athol, I agree with Hannie. When Photoshop first offered the ability to work in 16 bit; I jumped at the chance, but between the slowness of working in 16 bit and the huge file size--I went back to 8 bit. You will see plenty of arguments for working in 16 bit, but I see no advantage for me personally. In Dan Margulis book, Professional Photoshop, The Classic Guide To Color Correction, he states:

Quote"As it turns out, the quality difference is nonexistent. I have tested the concept quite extensively in the intervening years. Around 20 people, to my knowledge, have run tests similar to mine. The results have always been similar to mine." (Chapter 14, pg 341)

Dan Margulis does go on to to talk about gradients should be created in 16 bit if they are to be dropped into an 8 bit file (CMYK). This should be done after the file hits CMYK. For that he later goes on to explain about some print issues of converting to CMYK and the magenta gradient and using the 16 bit to solve a banding issue. For photographs, rather than computer created graphics, 8 bit is the best way to go. Plus, we are working in the RGB color space for printing.


Margie
Title: Re: 8 bit vs 16 bit
Post by: weewood on June 27, 2009, 01:24:18 PM
I agree with Hannie and Margie, 16 bit works great for HDR photos, but other than that I would stick with 8 bit.
Title: Re: 8 bit vs 16 bit
Post by: lurch on June 27, 2009, 02:27:23 PM
In "Restoration and Retouching", second edition, Katrin Eismann has a section on the advantages of high-bit data. Eismann has a note in that section titled Don't Try to Trick Photoshop. It says
QuoteTrying to trick Photoshop into thinking it is working with a high-bit file be converting an 8-bit scan to 16-bit, applying changes, and then converting back to 8-bit doesn't work. This voodoo maneuver is a waste of time.  . . .

If I'm scanning, I always capture a 16-bit tiff, to take advantage of the smoother dynamic range. We are working with jpg photos here, however, and jpgs are 8-bit only. No sense converting them to 16-bit.
Title: Re: 8 bit vs 16 bit
Post by: G3User on June 28, 2009, 01:10:50 AM
Thanks everyone.

While browsing for tutorials I have occasionally seen the conversion recommended but if Katrin Eismann agrees with all of you, who am I to argue ;)

Athol
Title: Re: 8 bit vs 16 bit
Post by: schen on June 30, 2009, 02:19:58 PM
There are two advantages I can see using 16-bit photoshop files:
1. It gives you higher dynamic range and
2. It has little loss after repetitive file conversions.

For what we are doing here, the images were typically captured in 8-bit JPEG file.  You will not be able to regain the dynamic range once it was lost. 

I did an experiment with RGB/CMYK conversions (8-bit) on a picture with very nice skin tone a while ago.  It took more than three round trip conversions to have the image degradation visible.  In our work flow, there is no reason to do that many conversions and most of the adjustments should be done with non-destructive layers.
http://www.operationphotorescue.org/forum/index.php/topic,875.0.html (http://www.operationphotorescue.org/forum/index.php/topic,875.0.html)

In short, I don't see using 16-bit files with any advantages for what we do here.
Title: Re: 8 bit vs 16 bit
Post by: G3User on June 30, 2009, 08:36:15 PM
Thanks Schen

Working in CS3 I had noticed on occasions that repeated applications of the healing brush to blend between an area of light colour and little noise with one darker and with noise produced a very grainy effect with almost pixelated colors. Probably over doing it and working at a too high a magnification

Cheers

Athol
Title: Re: 8 bit vs 16 bit
Post by: lurch on July 27, 2011, 08:44:00 PM
Bumping this topic because of a recent very interesting discussion on RetouchPro, on the benefits (or not) of editing 8 bit images in 16 bits. It's at http://www.retouchpro.com/forums/software/33750-converting-8-bit-images-16-bit-any-benefits.html. Might have to eat my words.