Just curious what bit depth people are using while converting their raw images? I have been going through John Shaw's book, and read that he does everything in 16 bit for his master files, etc - then backs down to 8 bit for output. This seems to make sense for the long term, giving the highest quality master files. I just need to buy a couple more hard drives now...
What do you do in your workflow?
Posted by Tony Bouchard on August 21, 2007 at 3:18 PM
16-bit for digital and 8-bit for scans
Posted by Jesse Speer on August 21, 2007 at 3:42 PM
16-bit, NO DEBATE, for raw files. Process a photo filled with open, blue Colorado sky, and you'll find out why. Printers are mostly limited to 8-bit right now, which is why I suspect he backs down (I haven't read his book).
hmmm - I have been jumping around in his book (Photoshop eBook), but now I can't seem to find him talking about backing down to 8 bit specifically for printing, etc. Maybe that was for a particular adjustment type that can't be done in 16 bit... I'm a little behind in my digital workflow I guess...
Posted by Jesse Speer on August 21, 2007 at 4:07 PM
"Output" could be printing (many labs require 8-bit), submission to stock agencies (most agencies require 8-bit), etc. Regardless, you're certainly on the right track in making 16-bit master files.
Posted by Kalin Wilson on August 21, 2007 at 4:41 PM
I agree, 16-bit for all processing - RAW conversion, PS correction, master file. I keep everything in 16-bit. Even for printing, I doubt there is a difference between sending a 16-bit file into the printer pipeline and letting the driver convert it vs converting to 8-bit in PS before printing. In fact in LR, you're in a 16-bit workflow until you export or edit in an external app - at least for RAW files. So unless you are sending a file somewhere, I'd keep it 16-bit.
My Canon iPF5000 comes with an PS export plugin for printing that actually sends 10-bits of color data to the printer. So that drives my workflow a little also.
One way you can combat the file size issue is to rely on some of the great interpolation available in various software - keep the image size smaller and interpolate for larger prints. An example of this is discussed here. Still, with the low cost of hard drives, I prefer to work on my RAW files at native resolution, 16-bit and scan at near max optical resolution of my scanner, 16-bit. Working in PS on 100-200+mb images becomes the bottleneck.
Posted by Brett Deacon on August 21, 2007 at 4:54 PM
16 bit is great for RAW files. 16 bit is not great for 4000dpi drum scanned 4x5s which approach 700MB in 8-bit mode. On my PC, simply rotating a 700MB image takes long enough for me to run upstairs and grab a snack. I shudder to think of how long it would take do edit a 1.4 GB file.
Posted by Les Barstow on August 21, 2007 at 5:07 PM
Brett, you just need a faster machine. The new Mac Pro desktops come with dual quad-core processors... Just sayin'. (Speaking as a person with a 1.8GHz P4 without enough memory.)
Michael, 16-bit, definitely. If you need to do any significant work whatsoever after you export from RAW, the extra bits will come in handy.
Posted by Kalin Wilson on August 21, 2007 at 5:46 PM
I agree, Brett. My recent 4x5 scans have been at less than 4800dpi (max optical) for that reason. 35mm scans I aim for max optical or where I start picking up grain. Still, if/when my computer can handle the file size without me pulling out more hair (that I don't have), I'd opt for keeping as much data as possible.
Same as Tony, 16-bit for raw conversion, and 8-bit for 4x5 scans. The reason is the 4x5 scans rarely need to be interpolated up so banding and other issues related to interpolation (that are mitigated by high-bit files) are not really a problem.
Thanks for the responses all - I kind of figured that 16 bit would be the resounding answer.
Posted by David A Bliss on August 21, 2007 at 7:41 PM
I worked in 16 bit because somewhere along the line, someone told me to. I then switched back to 8 bit as I looked at the difference between actual images processed in both 16 and 8 bit. While I haven't done any prints over 16x24, I haven't seen much if any difference on actual printed images.
If I do end up getting the 40D, I will probably swtich back to 16 bit, simply because at 14 bit capture, I would be giving up a lot more depth going to 8 bit.
I don't think processing in 16 bit is a bad idea, especially if you have the space, memory, and processor for it. I don't... Another of my future purchases. Man, I need a raise!
Discussion
11 repliesWhat do you do in your workflow?
My Canon iPF5000 comes with an PS export plugin for printing that actually sends 10-bits of color data to the printer. So that drives my workflow a little also.
One way you can combat the file size issue is to rely on some of the great interpolation available in various software - keep the image size smaller and interpolate for larger prints. An example of this is discussed here. Still, with the low cost of hard drives, I prefer to work on my RAW files at native resolution, 16-bit and scan at near max optical resolution of my scanner, 16-bit. Working in PS on 100-200+mb images becomes the bottleneck.
Michael, 16-bit, definitely. If you need to do any significant work whatsoever after you export from RAW, the extra bits will come in handy.
Dual Quad-core, hmmm....and LOTS of RAM.
Guy
If I do end up getting the 40D, I will probably swtich back to 16 bit, simply because at 14 bit capture, I would be giving up a lot more depth going to 8 bit.
I don't think processing in 16 bit is a bad idea, especially if you have the space, memory, and processor for it. I don't... Another of my future purchases. Man, I need a raise!