alan little’s weblog

photoshop file sizes

1st August 2003 permanent link

This post will be slightly off-topic from what I’ve been writing about in this weblog so far. I’ve had separate subject area blogs for quite a while, which were really more semi-private notebooks - I was keeping them for my own use and didn’t really care if anybody else read them or not. Now I’m experimenting with mixing everything up together in one public weblog.

Which means I am, with a fairly clear conscience, about to talk about an obscure Adobe Photoshop issue that I would previously have put in my photography notebook/blog. It could be worse - at least I’m not jumping straight from Mac browsers and the relative merits of python and java as real world programming environments, to the finer points of some yoga asana I’m having difficulty with. (You wouldn’t believe how obscure that can get. I’m an active member of a couple of yoga discussion boards where arguments about the precise angle of the pelvis in a particular backbend, or all the possible implications of the sanskrit word atha (now), can go on for days)

Meanwhile, safely back to the technical - some thoughts on the use of high bit images in photoshop. High bit images is a term photoshop types use to refer to images with 16 bits per colour per pixel instead of the usual 8. And surely more data == better?


Working with 16 bit images has a non-drawback, and a drawback. The non-drawback is that they’re twice the size of 8-bit images. But so what? Memory is cheap and disk space is almost free. A few years ago when I was working for IBM, I was involved in a project to build a terabyte data warehouse, which cost several million pounds and at the time was one of the largest in Europe. Now I have over a quarter of a terabyte in my living room. And I want a LaCie Big Disk, which would bring the total storage capacity of Alan’s Data Center to over 800 gigabytes.

Let’s say a gigabyte of disk space costs about a US dollar and a 35mm picture, scanned at 4800 dpi and 8 bits per channel, is a hundred megabytes (these are right-order-of-magnitude approximations for ease of arithmetic). A 16 bit scan, therefore, is two hundred megabytes, or ten cents worth of disk space more than the 8 bit scan. If it even crosses my mind whether to save the larger file or not, then I’ve just wasted more than ten cents worth of time. And in a few years when disks have 100 times more capacity, that ten cents will be a tenth of a cent.

Memory isn’t as cheap as disk space, but I have a gigabyte and a half on my photoshop machine and that’s more than enough to work on the biggest 16 bit files my scanner is capable of producing.

The real drawback of working with 16 bit files is that you’re severely limited as to what you can do with them in photoshop. You can’t use the smart selection tools to pick out precise areas, and you can’t use layers to make changes. Layers allow you to switch your changes on and off at will in any permutation, without doing anything irreversible to the original image. They are the overwhelming advantage to working with 8 bit images. (They also substantially increase the size of your 8 bit file, back up to what a 16 bit version would have been or more - another reason why file size is a non-issue)

But quality takes precedence over speed & convenience. And more bit depth must mean better quality - I can do subtler colour corrections, with smoother results, if I have 64k tones to play with in every channel than if only I have 256. Right?

Wrong, apparently, according to The Photoshop High-Bit Advantage: Fact or Fiction. The author Jim Rich asked pre-press professionals and photographers at a conference to distinguish between prints of 8 and 16 bit images, using images he had specifically chosen to show up the potential weaknesses of 8 bit processing, and edited far more drastically than usual. And they couldn’t. They admitted they were guessing, and nobody got more than 60% of them right. Whatever theoretical advantage 16 bit images may have just isn’t visible to the naked eye, even if it’s the naked eye of an experienced graphics professional.

So maybe I should forego the comfortable feeling I get from those lovely big 16 bit files, and accept that the practical benefits of editing with layers are more important than the theoretical benefit of having more subtle colour gradations that I can’t actually see.

related entries: Photography

all text and images © 2003–2008