I've had some questions about it so this is a few brief notes on how I've been processing images. I hope it will be useful for anyone who wants to caption images or do photo stories but has never tried anything of the sort before.

Firstly you really need an image editing program that supports objects. A very basic
bitmap image editor is not much use. If you're doing captions / speech bubbles you'll probably need support for a vector graphics format such as SVG (Scalable Vector Graphics). Apart from this, it doesn't really matter what package you use.

I'm currently using Corel Photo Paint 9 because I've had it around for ages and I'm used to it. It's out of date now, the upgrade would cost quite a bit and it crashes a bit so I'll be looking for an alternative soon. The GIMP (www.gimp.org) will probably be the first thing I evaluate, but I don't like apps with lots of small, fiddly, windows as they get in the way when you use the "focus follows mouse" window model under X11.

Story Frames

I got the speech bubble art from www.openclipart.org, (alternate site: https://www.clipsafari.com) which is a handy resource for graphic items with permissive licenses.

The clip art was in SVG format and my version of Photo Paint is too old to read this so I used the SVG editor Inkscape (free from www.inkscape.org) to convert them to EPS format.

I always make copies of original images before I start any editing. Disk space is cheap and sometimes you screw up the image right at the point where the editor stops letting you undo things or inadvertently save to the wrong file. The latter is a big problem with repetitve workflows, you get into a habit of overriding warnings.

You'll need a fair amount of memory for heavy image editing work, especially large images. Many packages will swap to disk but this can be painfully slow. If anyone is interested the machine I do the editing on is normally used for software development with a P4 2.8GHz CPU, 2GB of ram and 450GB of disk space. Yeah it sounds a lot but try running Netbeans, JBoss and Postgres sometime.

To do the captioning you start with the raw image, in this case a JPG file and load it into the editor. The next step is to add the bubbles and text objects and size and position them. Doing the text first makes the required bubble shape obvious. The object layering has to be right, if the bubble is behind the text then the layers need to be reorganized.

With the size of images I have, a 12 point fixed font seems about right. Much larger fonts are used on the wallpaper images. Because we're working in an object-based program the text can always be resized and scaled after its been placed. A limitation of PhotoPaint is that if you edit the text you lose any scaling or other attributes you've applied to it.

After adding captions some images need a highlight area to make them stand out. This is just a filled rectangle of a neutral colour like RGB 192,192,192. The rendering mode is then changed from 'normal' to 'overlay' to make blend it with the background.

In some cases on the wallpaper images the background is too busy to make captions show up well. In this case I apply a mask to the image of the same size as the highlight rectangle. This area can then be desaturated or have its contrast changed to make the text and highlight stand out.

At this point the image is in a proprietary object-based format and is no use for display in a browser. It needs to be flattened out into a bitmap format.

When all the images for a story are done I use a Photo Paint batch process to convert them to a raster format, PNG (Portable Network Graphics) without compression. This only takes a few seconds. I could go directly to JPG here but its more flexible to keep the uncompressed images around. I can then choose image quality and compression in the web album software and change my mind easily.

Next I use the free program JAlbum (http://jalbum.net/), which is probably the best web photo album maker I've seen. After setting up the image size and quality options it doesn't take long for it to spit out the compressed JPG images and HTML for the slideshow. The output JPGs are heavily compressed. This does affect image quality but makes them 10 times smaller than the input PNGs.

I can then preview the story, fix up the silly errors I made and a couple of cycles round the loop later hopefully end up with something that can be released.

Wallpaper Captures

I can't capture full screen images in the game, the largest size it will do is 600x450, which is the size of the photo story images. To get full screen captures I use Hypersnap DX, which I have around from when I used to write about computer games. Capture options are set to special capture for Direct X, auto save and uncompressed PNG format.

The main issue is that this capture includes any game overlays (such as icons or the pause indicator) that are suppressed by the game's own snapshot function. This is an annoying problem as its easy to forget about it and ruin a whole set of screenshots. The same issue occurs with movies and external software such as GameCam.

Some of the overlays can be removed by means of the clone tool in the editing package. Working zoomed in to 1000% usually results in being able to cover up the unwanted graphic so the edit is difficult to see at normal size. This works best on dark background areas. It can be nearly impossible to do this type of edit cleanly if the area to be modified is part of the foreground e.g. someone's face.

Scanned Sigs

To do the items with signatures etc. I start with the raw image in JPG or PNG format. Then the signature or handwritten text is scanned. This is done at a much higher resolution than the final target image to avoid artifacts. I've been scanning at 600-1200 DPI.

The scan is then given a transparent background. In Photo Paint there's a tool for this called the Transparent Colour Select tool. If the scan is good the background will be uniform and one click on the white area will do the job. Occasionally there are some blemishes that need to be added to the transparency selection.

The scan can now be overlaid onto the image and sized / positioned. The image is then flattened and saved as JPG.

This was a bit time consuming, so I've now switched to a handwriting TrueType font for the document images. I managed to find one that looks similar to the writing I gave Sara when I was scanning.

Title Pages / Collages

I start with an image from the story to use as the background. This is run through a desaturate filter and then sometimes adjusted for contrast/intensity/lightness etc to give a suitable 'faded out' effect.

The story images are then copied and cropped to the size that's needed. They're pasted onto the background as objects and sized / arranged.

Lettering / highlighting is next and the image saved in the same way as the others.