The system I envision would present the author with a series of nodes that could be text, image, or video. Text would be written instructions/links, Image would be drag and drop or upload from computer, Video would be screen capture-either a selected window or the entire screen.

The tool would allow authors to create tutorials as needed, store them, present them, password-protect them, even sell them.

The biggest issue with tutorials is that they are difficult to make and wildly inconsistent-two tutorials on the same subject can contain a completely different structure. This tool would  make it easy and establish a framework for tutorials.

The finished product:

You can download the font here:

https://github.com/field2/quickworms

There are quite a few resources on how to do this, but most of the ones I’ve read, although I’m sure worked for some, went against a few of my own design principals. So I set out to find the perfect workflow for me, and here it is.

When I design icons in Photoshop (AP),  the end goal is a png sprite. Using a split window, I can zoom in on one window and see the actual size icon in the other. I can click on an anchor point and nudge it with the arrow keys, getting sub-pixel placement just right and having absolute control over the end product.

The move to vector as the final source has been really weird and challenging. In Illustrator, vectors don’t anti-alias the same was as they do in Photoshop. That is, if you draw a rectangle with the edges half-way between the edges of a pixel, in Photoshop, you might get a different grey value than if you did that in Illustrator (AI). And pixel snapping is inconsistent. I would copy/paste a shape from AP to AI, and  my perfectly sharp edges would become fuzzy, even though the paths were exactly in the same place. If I click/dragged an anchor over a pixel, then dragged it back, it would become sharp again. Very weird.

The tutorials I read through, https://github.com/blog/1135-the-making-of-octicons and http://glyphsapp.com/blog/importing-from-illustrator/ were really helpful, but as I said earlier, there are fundamental flaws with those workflows, at least as far as I’m able to incorporate them into mine. For some reason, 16×16 is this imposing number that icon designers hold sacred. It’s a good target, but it really does overly constrain design. I decided 20×20 was a much easier canvas to work within, and as long as I left a pixel or two breathing room around the icons, they look great in a 16×16 space-not too big, but not leaving out important details in the name of absolute limits. I also got nervous when scanning through the article at all the weird numbers: 2048, 2052, -17something. Do we really need all those complex numbers? As for the glyphsapp.com article, while trying to draw directly in Glyphs may be the best way, it’s gonna be tough to put aside my AI experience to learn a new method for drawing vectors until I have lots of time on my hands.

After trying all sorts of settings, I came to the conclusion that 20x20pixel icons should be designed in a 2000×2000 upm font. The glyphs article points out that 1 ai point=1 glyphs unit. So I worked in points.

Another oddity was the Glyphs vs Glyphs Mini (GM) inconstancies. Joen Asmussen, who I can’t thank enough for all his help on this, designed the icon font you see at WordPress.com, using GM and following the octicons article pretty closely. I had a trial version of Glyphs, but decided to spring for GM for the sake of consistency. Although I started the font in Glyphs, bringing it into GM was an eye opener. The major issue was that GM doesn’t allow you to change the font’s UPM settings; so when I opened the .glyphs file I created in Glyphs, I was stuck with the 1000 upm I was originally working with. When I decided to try 2000 upm, so that it mapped more (theoretically) naturally with the 20pt x 20pt AI designs, my glyphs all got cut to 50% of their size. Resizing in glyphs was not something I wanted to learn how to do, so I returned my workflow to Glyphs.

Screenshots:

New AI file, 20pt x 20pt with points as units

Keyboard increment set to 1/8th of a point, so I can nudge vectors between pixel edges as needed.

Gridline every 10pt, with 10 subdivisions

Snap to grid as needed

Select all your glyphs in Glyphs mini, set widths to 2000 upm

Choose font info in glyphs mini

Ascender/caps height to 2000, giving you a perfect square for each glyph

Icon drawn in AI, with a 20pt by 20pt box drawn around it as a bounding box

Scale it up to 2000x 2000, note the little chain lock ensuring proportional scaling

Copy the icon, Double-click a glyph in Glyphs Mini, Paste into the glyph window. Make sure the x/y/w/h are exactly as shown

Double-click on the “bounding box” and delete it.

You can download the glyphs file, AI source file, and .otf here. Stay tuned for a video post.

Designed them in AI CS6. Preferences set like this:

Note the copy as settings; apparently you need this so you can copy/paste into fontlab…

Gridline every 10px, w/10 subdivisions. Icons are based on a 10×10 grid. I made the color black so they’re easy to see.

Make sure snap to grid and snap to point are turned on, and that you can see the grid.

Note the extra strokeless/fill-less 20x20px box around the icons. You need this to copy/paste into fontlab and not have them get scaled up.

Then I created a new font in Fontlab 5, with metrics set like this:

I copy/pasted each icon-the shape and the invisible bounding box- into a separate  glyph in fontlab. I tried to make it intuitive; “a” is appearance, “d” for dashboard. “p” was problematic; posts, pages both have p.

After I generated the otf file, I used @font-face to embed it on a test page. I set a base size to 62.5% and set the icons to 2.0em; effectively rendering them at their native 20×20 pixels. The key thing is the -webkit-font-smoothing:antialiased property in the css. Without it, they look like garbage. They pretty much look like garbage anyway in firefox/opera, and I have no clue how to get them in IE. But in webkit it’s gold.

I’m trying to get the whole thing on github since I guess that’s what you’re supposed to do, so stay tuned…

Prognosis to date: it doesn’t look good.

I drew a 16×16 pixel box in Illustrator CS6.

Copy/pasted it into the lower-case “a” glyph in Fontlab 5 (with metrics set to 800 upm).

Exported otf, css3′d it into a basic html file, typed a single “a” on the page, loaded the page in Chrome and Safari, and was disappointed.

Here’s what I wanted to see: A perfect, 16×16 pixel black square, no anti-aliasing, in all major browsers.

Here’s what I got:

Chrome rendering of a 16px by 16px square embedded into the “a” glyph

Safari rendering of a 16px by 16px square embedded into the “a” glyph

Those ain’t perfect squares. They’re close, and the main dark part of the square is 16x16px, but there’s some major anti-aliasing going on.

It’s also funny to note how Chrome tints the edges with a more reddish hue than safari.

I’ll keep playing around, but webfonts for iconography aren’t looking very promising, at least as long as low-dpi screens are around.

[UPDATE 1-26-13] I’ll post on the effects of webkit font smoothing shortly, how that impacts this experiment, and what it means for other browsers…

icon files are here

codekit-icons

PSD here

icons