A reader asked me about jpg conversions for web viewing. Specifically, how not to have them look washed out and a mess. There are lots of people who know more about this than I do, but I’ll share what I know. There are two issues to address—color and sharpness.
A mistake people frequently make is to choose the wrong color space for web display. Lots of people don’t know from color space. The simple explanation is that the color space of a photographic file describes the numerical values of the colors. If you have a file in a color space that’s really big (meaning, the space it occupies if you map out those numerical values in a 3D graph), but the color space of your output is smaller (like a computer monitor, or an offset press), you need a translator. It's called a profile.
No monitor under $6000 can display any color space other than sRGB. In a color management environment, in applications that use color management, your monitor profile converts all color spaces to look pretty much alike. You use output profiles to accurately translate what you see on a monitor to what you’ll see on a device. Step outside of a color managed environment, like the web, and it’s a different world.
Here is an example. Photo on the left was a RAW file (of my dad, by the way) that I saved in sRGB, a tiny space. On the right I saved it in ProPhoto, a huge space. That’s the first element. If you’re shooting in RAW, convert to sRGB when you open the file. If you have a tiff file from elsewhere, check to see what color space it’s in, then convert to sRGB.
The second element to good looking web photos is how you shrink them down and how you sharpen them on the way. I’m still working my way through Bruce Fraser’s sharpening book, so I don’t know the right way to do it. But I have my way. My conservative approach is the 10% method. I open the file (in RAW convertor I use the smallest file size, 1536x1024, as ACR has a good sharpening algorythm for shrinking files). Then in the Image Size dialog you pick a percentage figure instead of a pixel figure. I type in 90%. And do that again and again until I have the size I want.
Along the way, if I see the image start to fuzz up, I’ll run an Unsharp Mask filter. I use a very low radius (.9 pixel), and set the sharpening to where I can barely see an effect. Often I’m as low as 15 or 20%. I usually run Unsharp after three or four downsizings.
The top image is resized in one fell swoop. The bottom is in 10% increments.
This is, of course, tedious. If you save to the same size all the time, you can write a Photoshop action and assign a keyboard shortcut for it. And, truth be told, I don’t strictly follow this regime all the time. I often make bigger jumps in size. Or use Bicubic Sharper if I'm really in a hurry. There’s an issue too with horizontal and vertical images when you’re batching it, for which I use the File—Automate—Fit Image command. I shrink at 10% intervals to the longest dimension then run the Fit Image command for the rest and sharpen at the end.
Comments