The Lo-Fi Workaround

Seems throughout the last decade or so of large and larger captures, scans, bit depth, layers, etc etc, there is the constant proclamation that memory is now cheap, storage is cheap, go for it! If you are on Macs like me, the ongoing OS evolution, chips sets, software has become burdensome. Suddenly I need a new feature in my bookkeeping software, but the upgrade won’t work on that chipset, meaning an entire new computer, meaning a lot of other software will need expensive upgrading too, just to remain where I was. The high tech industry depends on this of course, but in this economy, for us in the 99%, it’s become impossible. My approach to my hardware and software has become like my grandfather’s approach to his car, he learned everything he could about it’s workings and kept the same old Chrysler going for years and years. It got him from place to place, the heater worked fine, what’s not to like? However, with our larger and larger prints, big scans and captures, added with the ability to utilize a lot of great editing tools on high bit files, horsepower and storage would be good to have in spades. For those of us with particular interest in high quality black and white, it’s critical to avoid loss in single channel files, even more important if we are printing with extremely linear and continuous tone systems like K7. So, here is my workaround for keeping file size and CPU usage reasonable for my collection of old Chrysler like Macs. It’s simple, works well, and makes sense on a variety of levels, particularly if doing ongoing editing for others. My screen shots etc will be on a Mac, CS5, but I have been using this workflow for years on a variety of machines.

This is a 16 bit drum 2000ppi scan of a 5×7 B&W negative, the file size is 242.3MB, it is saved in this form.

The first step is to convert it to 8 bit, and SAVE AS, with an altered name, I usually just add “-8” to the end of the existing name.

Now all the editing begins, always using adjustment layers. Many people like to work on the image layer itself, using a variety of tools, I do not. I work for myself and others and need to make and un-make edits quickly and easily. It’s important to remember the pixel dimension must not be changed for this to work, no cropping or image size changes. I use this file to make hard proofs as I go, to see if the edits are working on paper. Sometimes a curve layer will have a new version, with the previous turned off, so I can easily go back if necessary. If I want to preview a crop, or have decided on one as I work, I used a masked levels layer as seen on this file. So, these are my tonal adjustments for this image, as adjustment layers on an 8 bit file.

Now I will transfer all of this to the 16 bit original. Select all layers by clicking on each holding down the shift key,

now go to the layers palette and scroll to “new group from layers”, you can name it anything you like.

Now all your layers are in one folder.

Make sure you save now. Leaving this open, also open your original 16 bit file, I’ve used the 2 up display to make it easy to work on both.

Holding down the shift key, so all masks will be in register, drag your group folder from the 8 bit file over your 16 bit image.

Depending on your computer, it may think for a few seconds, but you will see all the layers now on your 16 bit file when done. Close the 8 bit now, to free up computer assets to work on the larger file, flatten it, disregard any inactive layers.. for whatever reasons through the work process they are not relevant to this version (this should light some brain cells up, different groups for different versions?).

Do a “save as”, you want your original raw scan left intact. If you had a crop layer, all you have to do now is magic wand the white, invert the selection, and crop, done. Notice that now, a simple histogram refresh, will appear gapless and smooth. As a point of interest, the 16 bit scan has 31,468 levels of gray, after edits, it reports 32,769 levels of gray. The 8 bit, before edits had 247 levels of gray, after edits, 256 levels. Now you may find it odd that the edited versions report more levels than unedited. This is because of all the local adjustments, had adjustments been only global, loss would be easily reported. There is surely LESS than the original number of levels within each selected area after the curve is applied. The real point is the advantage of editing hi bit.

Now, with all the edits applied you have your high bit final, and it’s time to sharpen if you must, and then take care of any retouching, this order is important, if you had sharpened earlier, contrast adjustments may bring out artifacts, and subsequently, your sharpening will bring out specs etc, that will require retouching.

There are many benefits to this workflow, not only saving hardware assets, I use it all the time now for everything, including color. Smart Object use and Lightroom, may make some of this less relevant, but I still work with a lot of scans, or large print finals that were captures, and Photoshop is required for the most finessed work, complex masking, etc. Perhaps you will find this useful as well.