Getting There From Here... My Image Postprocessing Workflow

DIALUP WARNING: Many large images contained within.

In the interest of sharing information, I thought I might outline the postprocessing workflow that I use. There is no one absolute method when it comes to image adjustment. I use what feels comfortable to me, and the tools and methods I use may not be for everyone. YMMV. There are probably better ways to accomplish some of the tasks that I'm unaware of. That's part of the fun of all of this... there's always room to learn.

Also, it is not a cut and dried specific process. The steps needed will vary from one coin to another due to color and reflectivity. The adjustments you make for a BU silver peace will likely be different from those you use for a VG copper piece. You try an adjustment and see if that gets you closer or farther from what you perceive the coin to look like in hand. Undo is your friend, especially when making fine adjustments by degrees. Lightening shadows by 10% may be too much for a particular coin, but 3-4% may be just right.

Some lightening and darkening adjustments also oversaturate color, so you may push the image one direction to brighten it and then desaturate to pull the color levels back to reality.

Beware of the tendency to make an image look too colorful or too lustrous. The former gives you color gradations and transitions that look phoney and the latter can blow out highlights. It's more of an art than a science... and a process of error and error (as opposed to trial and error).

Anyhoo, on to the workflow...

I use Adobe Photoshop CS5 (64-bit) for Windows. That's what the screenshots below are from. When working in Photoshop, RAM is your friend. You can never have too much. The more RAM that Photoshop has available, the more steps of undo you have, and the faster certain tasks run since the computer doesn't have to use virtual memory.

RAM is very cheap right now, comparatively speaking. I just bought 32GB to max out my desktop for $160 total. Be aware that on a PC, in order to use large amounts of RAM, you need to (1) be running 64-bit Windows, not 32-bit, and (2) if running Windows 7, you need to be running the "Professional" version (Home Premium restricts you to a maximum of 16GB of RAM... thanks Microsoft).

One feature that you should learn to use in Photoshop that will eventually save you a LOT of time, especially on tasks that you repeat over and over with different coins, is the Actions panel. Here you can record an action or series of actions, and assign them to a keypress, which you can then play back on future images. I would be dead without it.

Let's start with the before and after pics that I'll take you through.

[In order to see detail on some of the pictures in this thread, you'll likely need to click on them and view them at full size rather than the shrunk versions that display within the thread.]

BEFORE:




AFTER:




Let's open the orignal image in Photoshop.



The first thing I notice is that the image isn't exactly perfectly straight. Looking at the text, it appears to need to be rotated roughly a half a degree clockwise. Pressing Alt + i, g, a allows you to specify a numeric value to rotate an image by. You can also use Free Transform and use the mouse to eyeball it, but being a keyboard guy that doesn't have fine motor control, I try to do as much as I can via the keyboard rather than the mouse. I also try to use methods that are numerically defined rather than hand drawn.



Here is the image once it has been rotated.



The next step is to crop the image. To do it right actually involves a bit more thought and explanation than you would think (bear with me). First, select the crop tool from the tool palette (see screenshot below). We want to constrain the crop to being exactly square, so we want to fill in dimensions in the Width and Height fields in the properties bar at top. We want the crop to be nondestructive with respect to pixels, i.e., we don't want the crop to alter the pixels within the crop (I'll explain why in a bit). I use inches as a unit of measurement (using pixels would actually resize the cropped image). Since we're using a nondestructive crop, you could use any number; 4 inches x 4 inches is what I choose to use.

I crop it so that the distance between coin edge and image edge is equidestant on all 4 sides (or as close as one can get depending on how out-of-round the coin is).



Here is the image with the crop drawn.



Here is the image after double-clicking in the cropped area (or pressing Enter) and subsequently enlarging the view (Ctrl + = to zoom in, and Ctrl + - to zoom out).



The image is a bit dark for discerning detail and color. Time to tweak!

I like to start with Auto Contrast (Ctrl + Alt + Shift + l).



It's still a little on the dark side compared to the coin in hand, so now there's a couple of different ways I could go:

1. On circulated coins without lustre, I would use the Shadow/Highlights adjustment to lighten shadows (Alt + i, a, w).

2. If the coin is AU or better and I feel the lustre in the pic is a bit subdued, I will adjust both the shadow areas and boost the lustre slightly by adjusting levels (Ctrl + l).

Option #2 is what I will do in this case. When the Levels window is displayed, reducing the number at the far right of the Input Levels section will lighting highlights and overall appearance. Clicking the Preview checkbox will allow you to see before and after prior to committing the change.



Here is the image after the change. The background exterior to the coin are now blown out, but we don't care since that all will be deleted anyway.



Between the Auto-Contrast and the lightening, the color is a bit oversaturated. The copper in the center of the coin looks a bit too orange. We need to back it off just a hair. Press Ctrl + u to bring up the saturation window. You can play with the numeric adjustments and the preview window until you feel it's accurate. In this case, a saturation adjustment of -12 will do the trick.



Here is the image after the change.



Now we need to get rid of the background. Many people handle this in different ways. The reason I have chosen my method is that it can be semi-automated and requires a minimum of freehand work.

From here on out, all the work is pretty universal from coin to coin, so I have all of the steps recorded in a total of two actions, meaning that other than one adjustment, it can be performed in about 5 seconds once you have the actions recorded.

First action: placing guides so that any masking I do is perfectly centered on the coin. I want to end up with a crosshair dead center in the middle of the coin. We will create two guides, one vertical and one horizontal. Press Alt + v, e to bring up the Add Guide window. Type in 2in for the position.

Here is where the beauty of having done the crop using INCHES rather than pixels comes into play. As long as the image is cropped to a nonpixel dimension, the position of a guide at half of the cropped dimension will ALWAYS be perfectly centered, no matter if the image is of a Half Dime or a Morgan dollar, no matter how many pixels in the image. This is what allows the setting of guides to then be automated via an action.



Image after the guide is placed.



Repeat this step, placing a guide in the other direction.



Now we can draw a perfectly circular mask positioned equidistant from the center of the coin. This is the only remaining nonautomated manual step.

Press m to select the marquee tool. By default it is a square selector. Press Shift + m to change it to a circular selector (Pressing Shift + m will toggle the selector back to square).



Hold down the Alt and Shift keys. Click on the crosshair at the center of the image and drag outwards. This will enlarge/decrease a perfectly centered, perfectly round selection area, outward from the center of the image. Once you have the selection just inside the border of the coin, you can let go of the mouse. Here is my selection:



Note how it's just a hair off positionwise. If I crop now, I'll get some of the white area at bottom. I can use the arrow keys to move the selection area compared to the image. In this case, pressing the up arrow 3 times gets the selection area to where I want it to be.



For the final set of adjustments, we need to be viewing the Layers palette. Press F7 to show the palette.



All the remaining steps can be recorded, once you get them down, as they do not need to differ from one coin to another.

We need to separate the coin from its background. Press Ctrl + x to cut the selected area.



Press Ctrl + v to paste it right back... this creates a new layer with the coin in it, so the coin and the old backrgound are now on separate layers (see the layer palette at right). It will also reposition the coin slightly to recenter it on the overall image width.



Now let's remove that background. Click on the Background layer in the layers palette. Press the Del key to delete it.



Now in a perfect world we could resize this to our final size (800 pixels per side, 1000 pixels per side, whatever), save as 24-bit PNG, and be done (to me, PNG is the perfect image format for online display of coins: it has the color palette of JPEG and the background transaparency of GIF... that way, regardless of where you display the coin it would ALWAYS look like it is floating in space). However, there is a problem with this option, depending on what tools you use for ebay listing... Turbo Lister is not compatible with the PNG format. It absolutely munges PNG files that are uploaded with it. You lose the transparency and get all sorts of chromatic aberrations and artifacts around the perfiphery of the coin. PNG is fine for any applications where you are hosting your own images, but if you are uploading them through ebay in any way, you are toast.

So... we need to create a color background for the image. You can choose any color you like. I like a medium grey (RGB 140,140,140). When I was saving high-res files for clients for them to repurpose later, I would automate the creation of 4 layers (solid white, solid black, light grey, dark grey) and save as a PSD file, so that the client could toggle underneath the coin layer, so they could generate multiple different versions depending on their preference.

For the purposes of this tutorial, I'll simply fill the background with the grey.

First, we need to set the foreground color to grey. Select the foreground color box at the bottom of the tool palette.



In the window that comes up, set the R, G, and B values to 140 and click OK.



Select the fill tool (looks like a paint bucket) from the tool palette.



Click in the transparent area to fill it with the gray.



Now it's time to resize. Press Alt + i, i to bring up the resize window. Set the values in the Width and Height boxes and click OK.



After resizing, I like to give the image a slight sharpen. I use the Unsharp Mask filter. Go to the Filter menu, select Sharpen, and then Unsharp Mask.



The values shown below are the ones I use. Experiment using the preview, but be careful not to oversharpen.



Now we can save the file. I use the Save for Web option. Press Ctrl + Alt + Shift + s. Make sure that you choose JPEG as the image type, and that the quality is at a high setting. With storage space being virtually infinite these days, I see no reason to be use any setting other than Maximum.



And there we have it... the final image.



Hopefully you found some aspects of this tutorial useful. As I initially said, my methods are certainly not the only way to process images, and there may be better ways to go about certain tasks.

These multiple tasks may seem daunting, but once you become familiar with them and set up automated actions, it really doesn't take much time at all. I can now process an image in 2-4 minutes total.

Happy imaging!