In my last blog, I said I'd share the experience of creating an interactive infographic which needed only CSS/HTML and some .gif image files. I used mainly low cost or free PC software.
Two bits took up a lot of time. One was figuring out how to deal with nested (and curved) shapes. The other was capturing all the coordinates for the hotspots. As I mentioned to a pal this morning, "It was mind-bending stuff. A bit like cleaning the windows of a skyscraper with a sponge and a bucket while learning to abseil." Not one of my most eloquent analogies, perhaps, but it was off the cuff.
The start point was an image of the basic architecture for handling the media plus a header line. The image is quite old, but I'm pretty sure I created it using the GIMP. I typed the elements of the header line into Word, then took a screen shot, pasted it into IrfanView, cut out the bit I wanted and dropped it into the GIMP. This was the base, or background, image for all that followed.
I thought about tarting up the architecture illustration - making the lines three-dimensional, for example, but it was beyond my competence. Having said that, the interactive infographic is intended to mirror my business card and the 'greybeards' slide deck. (That's my post-hoc justification, anyway.)
So, to the task at hand:
I used CSS to recognise and display the three layers: background image, hotspots and replacement image. I'd done it before, but in different ways. (I Googled 'CSS HTML hover hotspot map' or similar and found the interesting work of Stu Nicholls.) I liked his idea of redefining the <i> element in CSS to display each replacement image.
If you want to see the actual code, just 'view source' of the web page. The CSS and the HTML are all in there together. (You'll notice some 'z-index' references: these determine the stack level of each element: cursor, hotspot, replacement image and background image. The higher the number, the nearer the user.)
I used the GIMP to create separate .gif files for the background and each of the replacement images. I felt it was cheating (compared with other image-switching work I've done before where I kept all the elements in a single file and shuffled them around, such as the grid here and the photo here.) Doing it with separate files was fairly quick and straightforward and the files are tiny - less than 50k each.
UPDATE 27 JULY: They're over five times tinier now. Only the changes to the background image are stored in transparent .gif files.
I created the text boxes and drop shadows in PowerPoint for speed. I scaled the presentation pages to match the iPad-sized 'architecture' graphic, then took screen shots of eight boxes at a time and pasted them into IrfanView.
From there I cut and pasted each box in turn into the appropriate GIMP image. (Each lives in its own layer and they can be selectively merged, so they all start off transparent, apart from the background which shows through the whole lot.) Because the cut images were rectangular and the text boxes had curved corners, I had to use the GIMP's tools to effect a repair when the underlying graphics were obliterated.
To create each .gif, I saved the visible image (the highlighted area and text box on a transparent background with the base, or background, image showing through).
The HTML simply bolted together the hotspots for each image. I used the TextPad text editor for the HTML and the CSS (which I embedded in the HTML header) so it's all just one file.
The hotspots could only be determined in rectangles, which was okay for the triangles side but a nightmare for the circles side. (I admit I did try to figure out if I could nest HTML's 'circle', 'rectangle' and 'polygon' image maps to achieve the desired effects - a long story, but I gave up.) The hotspots also needed to be 'finger aware' for iPad users, so I made some of them bigger than the underlying image element.
I worked out the hotspot coordinates by cursoring over the GIMP image and taking the top/left height/width readings. To keep myself on track, I temporarily gave the hotspots a light grey background colour and used TextPad's 'View/In Web Browser' option to check progress.
Elegant it is not. But it works well enough.