Saturday, October 25, 2014

Playing Photo Doctor

You may or may not have noticed by now that I'm a bit of a writer. I always have been, since I was able to hold a pencil and form letters on paper. I wanted to be a journalist and/or a novelist way back in childhood, and even when theology and music got mixed up in my ambitions, writing was always for me an important part of those pursuits. And now, after eight years working at a magazine where I emphatically was not wanted for my writing ability, I am finally taking home a paycheck for work that primarily consists of writing stuff.

OK, so it's not Pulitzer Prize stuff. Nobody outside a 30-mile radius of Stover, Missouri, is going to care about it. But it's a step.

I would like to say it's ironic that the more I have to write for a living, the less I end up writing for my blog, but it's really not ironic. For the moment it's just a matter of time management. I don't expect the dry spell of the last few months to continue. The blog will, I think, continue to fulfill a need that even the writing I do for a living does not meet.

Meanwhile, my newspaper job has me trying to do something at a professional level that I have only previously done now and then as a rank amateur. I speak of photography. And though my previous magazine job required me to learn a lot about Photoshop, I've never had to use it the way I do now. Which is to say, making the photos I shoot for publication look like they don't suck.

During my first week or so on this job, my publisher walked me through some steps to do with nearly all photos that will go into the newspaper. Later I compared notes with a couple of co-workers and my immediate predecessor as editor of this paper. Still I struggled. Still my pictures came out looking like crap. This past week I took some entirely different pointers from another editor who has been working for my publisher for a long time, with more consistently satisfactory results. I'm going to be trying his approach, as far as I can imitate it.

Here, as an aid to my own memory, is a brief run-down on the steps I have been trying to use, and the ones my colleague taught me this past week.

The "Kill the Magenta" Approach
Here's the procedure I established based on my initial instructions from the publisher and tips from my immediate co-workers.

First, drag a copy of all the day's photos from the camera's memory card into a folder where you can store them for your records. Preview the new photos, choose the ones you want to doctor up for publication, and copy those onto the desktop. After opening them in Photoshop, choose "File/Save As" and re-name them with a brief descriptor followed by the identifying number at the end of the original file name. This way you can keep the edited photo together with the original shot.

One of the first steps after creating this new JPG file is to change the color-space (under "Image/Mode") from RGB to CMYK, because the camera shoots in RGB but the color pages in the newspaper are printed in CMYK. (For you newbies, these jumbles of letters represent different processes of mixing colors. In electronic photography, the image data is layered in three or four color channels: either Red-Green-Blue or Cyan-Magenta-Yellow-blacK.)

I would often try "Image/Adjustments/Auto Levels" before changing the mode to CMYK, just to see if it made any difference. If not, or if the image became darker or some other weird change came over it, I would undo it. If Auto Levels corrected any obvious problems, such as being too dark or having too much red or yellow, I would bless RGB for its service and then fire it and replace it with CMYK. Somehow I never found Auto Levels helpful after making the mode change.

Another thing you will do quite early in the process is crop the photo to eliminate extraneous details and allow the main subject(s) to fill more of the frame. It is advisable to crop off the sides but to leave the top and bottom of the picture as is, so there is room to pull the edges of the photo in or out depending on the space the layout editor needs to fill.

Then you can re-size the photo ("Image/Image Size") so that it snaps right into the layout without needing to be resized. The resolution needs to be 200 psi (pixels per square inch). The only other important value is the width, which depends on how many columns wide you intend the photo to be. In our newspaper, most photos are either two columns (3.558") or three columns wide (5.388"). Other possible widths in our six-column broadsheet format are 1.729" for one column, 7.217" for four columns, 9.046" for five columns (generally used once per issue, in the front page's lead photo), and 10.875" for six columns, which is purely theoretical so far in my experience. Once you set the width, the other image dimensions will adjust themselves in proportion.

This is a good point at which to pause and save the file. Then you can create the web version of the photo, to be posted online later. To do this, go back to Image Size and make it 7 inches for a portrait photo, 6 inches for a landscape, and 72 psi either way. Then do a "File/Save for Web," making sure to park the new file on the desktop and adding "web" at the beginning of the filename.

The funny thing about this save-for-web gimmick is that it creates a new file, but the file that remains open in Photoshop is still the one you were working on previously. So the next step is to undo what you just did to the image size before proceeding further.

Then come the adjustments. My original brief was to go into "Image/Adjustments/Levels," set the color channel on Magenta, and drag both the right (white) and the middle (gray) slider to the left to reduce the levels of magenta in the image. The purpose of this was not only to lighten the picture, which would otherwise tend to look too dark in print, but also to correct for a tendency of magenta ink to bleed into newsprint faster than the other colors.

Later, as I messed around and found things that seemed for a while to work better for me, I would use "Image/Adjustments/Color Balance" instead. Then I would pull the Cyan-Red slider most of the way towards Cyan, the Magenta-Green slider part of the way towards Green, and the Yellow-Blue slider part of the way towards Blue. I had to be careful, though, because it was very easy to end up with a picture that looked way too bluey-green, or people who looked like the undead.

After doing this, I would then go into Levels, select the black color channel, and pull the left (black) slider to the right as far as where the black channel started showing up on the graph. This would darken the picture a little, but more importantly it sharpened it and punched up the contrast. Another slider in the same dialog box, controlling the grayscale balance between black and white, could be adjusted to push the black end of the scale a little grayward, to correct some of the darkening tendency of the previous step.

Then I would increase the brightness and contrast a bit ("Image/Adjustments/Brightness-Contrast"), and if the picture needed still more lightening, I would play with the hue, saturation, and lightness (also under "Image/Adjustments").

Now and then, however, I had to deal with a picture where all my ingenuity could not lighten the image enough without either leaving it washed out and bled dry, or doing something weird to the color. Sometimes I had to go through the sequence several times, trying to find a balance that would work without torturing the picture beyond recognition.

One breakthrough came when my predecessor gave me a tip for lightening really dark images: go into Color Balance, select the "highlights" radio button, and drag the sliders in the opposite direction to what you did with the "midtones." (I never found any use in adjusting the color balance with "shadows" selected.) My forerunner said, and experience bore him out, that setting the Cyan-Red slider to 50 and the other two to somewhere around -10 would do wonders to brighten up a shot languishing in shadow.

After this, Photoshop's role in the process was pretty much over. Then we just had to drop the photo into InDesign, set a bounding box around it, put a cutline under it, put a horizontal stroke under that, put a slug line at the top of the page, print it, and send it through editing, proofing, and layout. The InDesign document would be saved, packaged in its own folder with subfolders for fonts and links, and placed in the appropriate folder depending on whether it was "In Proofing" or ready to be published. The web version of the photo, after being posted to the newspaper's website and social media with the cutline, would be added to the "links" folder along with the original photo and the doctored version, in case they would be needed during layout. Once proofread, the file would be corrected in InDesign and parked on the server for the layout editor to use, and the printed copy would be trimmed, waxed, and stuck to the layout board until the time came to assemble the dummy pages, based on which the pages would be laid out in InDesign.

Doing all this helped for a while. But still, all in all, my photos sucked.

Don't mistake me. I didn't expect Photoshop to do it all. I also experimented with different settings on my camera, guided by the voice(s) of experience. Sometimes the problem may have been in the performance of my old, decrepit camera. But apart from fiddling with the ISO settings, the focal length, the white scale settings, and whatnot, what could I do?

I could go to the office of the biggest paper published by my publisher and pick the brain of its city editor. He redirected me somewhat. Here is the procedure I'm going to work with now:

The "Save the RGB" Approach
My colleague said, for starters, I can forget about switching the color space to CMYK. To be sure, the color portions of the paper are printed in CMYK; but the postscript machine that every page goes through converts them automatically. So changing the mode to CMYK is a waste of time, especially given that it makes the file sizes larger, and that makes some procedures take longer. Plus, many quick fixes don't work as well in CMYK as in RGB, as I had already found with Auto Levels; and besides, most of the paper ends up printing in grayscale anyway, so the color data will be discarded.

I'll call this Step 0, because it's really not so much a step as a decision not to take a step I've been taking all along, one that I think has really slowed me down.

Step 1 is to crop the picture, but don't resize it yet.

Step 2 is to try Auto Contrast and then Auto Levels, using the History palette to step backwards if any of my adjustments have weird results.

Step 3, which I insert here on my own judgment because I think it serves my work flow best, is to save the file with its new filename and then resize and save the web version. Then step backward to before you resized the image and proceed.

Step 4 is to use "Image/Adjustments/Curves," pulling from the center of the diagonal line and tugging the curve towards the upper right corner so that the image lightens up enough but not too much. You can often tell whether you've gone far enough, not far enough, or too far by converting the color-space to grayscale and stepping backward again.

If one area of the photo needs to be lightened or darkened more than the rest, my colleague suggests doing this before Step 4, for example, using the burn tool to darken a sign that will look too washed out once the faces next to it are lightened just enough. He suggested setting the dodge tool (to lighten shadowy regions) at an exposure of no more than 4 or 5 percent, and the burn tool (to darken washed out regions) at a somewhat higher exposure. You can also select a region using the polygonal or magnetic lasso tools and run curves inside that region (either to lighten or darken it) before doing the same to the whole photo.

Step 5, if necessary, is to enter "Image/Adjustments/Hue-Saturation" and select just the red channel, or the yellow channel if necessary, and both de-saturate and lighten just that color channel, just enough to take out excessive reds and yellows. My colleague emphasized that this procedure simply reduces or lightens the level of that particular color without skewing the image towards another color, as "Color Balance" does.

Step 6, and it's important to do it in this order, is to size the picture. My colleague gives the good advice of making the picture at least a column wider than you plan on using, in case it is decided at layout time to make the photo bigger than you planned; this way the resolution of the larger photo will stay above 200. I, however, like being able to snap photos right into my template without having to worry about resizing them; so I will probably only go with this advice when I judge the layout could go either way. Or I could simply plan to use the bigger size in the layout, since it's easier to shrink a too-large photo down at the layout stage.

Step 7, which my informant says must happen after Step 6, is to do an Unsharp Mask filter ("Filter/Sharpen") with the values 110% (amount), 1.0 (pixels), and 0 (radius). If no other filter has been applied since you last started your computer, you can use the keyboard shortcut for filters to apply this. If, however, you're working with a scanned photo, you should apply a Despeckle filter first ("Filter/Noise"). Surprisingly, my instructor tells me the sharpening filter works best with a photo that is already in good focus; pretty much every photo that is really worth using. The result is that minor details pop out with surprising crispness. He does not recommend using it to correct an out of focus image because that can just make a bad photo look worse.

I hope this helps, because a picture is worth a thousand words. So if a lot of my photos continue to suck, then even if I'm a good writer, I'll have to do a lot of work to catch up.

No comments: