Hacker News new | comments | show | ask | jobs | submit login
World's smallest legible font is not the smallest I've seen (angband.pl)
107 points by landhar on Nov 17, 2010 | hide | past | web | favorite | 39 comments

Mine, from around 1988, looks similar: http://www.tzs.net/small.html

I didn't do lower case, though. I created this as a debugging aid when I was writing SCSI drivers for a third-party Mac drive vendor. I wanted a debug printf, but since the disk drivers run before the OS runs, the OS facilities aren't available. I made a debug printf that printed, using my tiny font, directly to screen memory. Each character is 3x5, so that with one pixel spacing you can fit a horizontal slice of two characters in a byte, so I didn't need to write a general bitblt routine.

On modern displays it is hard to read because the dots are so small. On a Mac Plus built-in display, or a Mac II with a 640x480 or 1024x768 monitor it was quite legible.

My favorite is still Flea's Knees: http://typophile.com/node/61920

I can't find it any more, but I found another once which made characters in 2x3 pixels, relying on the positions of subpixels. Wasn't the most readable, but you could still make out what was written. That was the smallest.

Oh wow, I started reading the comments. That Stonecypher guy is a man on a mission, but he sure knows what he's talking about. Lots and lots of useful and fun info about subpixel fonts there.

I did find this guy's font much more readable than the one he linked to! I could actually read the paragraph without too much squinting!

This is definitely much more readable. My screen is 15" 19x12 resolution. So everything looks pretty small, but I could still read this, while the original one wasn't readable for me.

I can't really read anything that small on a modern high-res monitor anyway, but I found this bit at the end interesting:

"in most jurisdictions bitmap fonts are not copyrightable"

Is that really the case? A quick Google finds some articles in agreement, but I'm no lawyer. If it's true, I'm surprised I haven't seen, for example, Chicago pop up in more places.

Yes it's true.

The shape of vector fonts are also not copyrightable, but the actual file is. So you are free to recreate it if you wish.


What is the definition of "create"?

For example what if I wrote a program to open the original file, examine the shapes of the fonts, then write them to a new file?

That would essentially be copy-pasting the file, wouldn't it? But how is that different from someone using their eyes to "examine the shapes of the fonts and write them to a new file"? Just because of small inaccuracies?

So what if I wrote a program to very slightly change the shapes of the font? Did I "create" a new font?

A font file is not just a compilation of vector shapes. It is essentially a small (but potentially complex) program describing their proper rasterization under different conditions. Just copying the shapes as rendered at a certain size by a certain interpreter doesn't really give you a copy of the original font.

Here's an experiment you can try: write a program that renders a font to a high res grid, then refits vector data to that data, then publish a bunch of proprietary fonts redone using your program (documenting your approach carefully) and see how long it takes to get sued.

Interesting! Thank you for the info. I'd like to learn more about fonts.

> What is the definition of "create"?

Aha! You've voiced the problem with the very notion of having any sort of control over or ownership of information.

Some would say the greatest fonts were not created, but discovered?

To the best of my knowledge, the law has no answer either in practice or in principle. In my opinion, copyright laws are not merely abused, but actually gibberish in the modern era, but, well, I worked that out 10 years ago and we still seem to be lurching onwards, so whatever.

I don't really understand the definitions - typeface is the style of the letters, you have a font file, which is the implementation of that typeface, but what does "font" on its own refer to? Is it the same as the typeface? I guess that's how I would use it in conversation.

If that's the case then I'd say you created a new implementation of existing font?

A slightly related humorous aside: the Wingding glyph order is patented.

Do you happen to have a Patent number for that?

I can't seem to find any (non-Wikipedia) references to the patent that might give a clue.

This is a great question. I actually spent a bunch of time searching since there was no citation but all I could find were references to the wikipedia page.

So this is possibly made up.

I went back and found the original diff[1] where it was added. It is from an IP in Redmond in March of 2007.

Wikipedia needs a better interface to search through diffs. I had to manually do a half-interval search. :/

[1]: http://en.wikipedia.org/w/index.php?title=Wingdings&diff...

It's one of the major reasons why games use pre-rendered font bitmap tables.

Oh... I wonder what could we do if we had these fonts at the dawn of the personal computer.

Wait! We had them!

On the Apple II, the Magic Window word processor, could use a 70-column mode with the graphics screen and a font that was 4 pixels wide. It was 7 pixels tall, but nobody made a big deal out of that. I am quite sure I had proportional fonts that could fit in a 4x5 matrix in my developer toolbox (I did lots of courseware in the mid-80's that ran on Apple IIs). Using similar fonts, Atari and C=64 computers could do 80-column text screens with this same approach (both could do 320 pixels per scan line, while the II did only 280)

I used to have a Romanian ZX 81 clone in the early 90s, on which I could somehow boot CP/M which was really not meant to run on that machine in which there was an editor called tword that rendered characters in 4x7 pixels to have double column width (the native mode was 8x8 IIRC). A friend of mine hacked this editor to be able to display letters with acute accents so that we could write in Hungarian (he replaced some special characters). We were 11 or 12. I know, "cool story bro" :)

Cool story, bro :-)

In Brazil we did similar things with accents, but Portuguese is no match for Hungarian in that regard. Also, my grandpa taught me Rovasirás and I made a font (Apple IIs had a nice hack for graphical fonts) for it.

I would like to know more about that ZX-81-like computer. We had something like it here, one that could do 192x256 pixel graphics.

I had one of these: http://www.old-computers.com/museum/computer.asp?c=632 then later this: http://www.old-computers.com/museum/computer.asp?c=629

The second one mentions a Brasilian connection :) Yes, the resolution was the same.

Check out in Windows: - open console: Start > Run > cmd - right-click title bar, select Properties - select raster font 4x6, which is actually smaller, has more characters and appears to be even more readable


This is not as impressive as some hand written books I've seen in an italian abbey where the size of the text about the same.

"but, a font I created in 2004 for side messages in a MUD client is smaller"

No it's not, his font requires a 320x240 box to display all the text, the original font fits in a 302x220 box and that even includes ~8-10 pixels of padding. http://techhammers.com/wp-content/uploads/2010/11/Worlds-sma...

"Also, my font is fixed pitch as I needed it to be that rather than proportional, so you can shave quite a bit more."

Are you taking that into account?

If you blow up the image of the Declaration text, it's quite readable. I don't see much use for it today, but it's really a very nice design. I'm skeptical that a legible font could use any fewer pixels.

Pixel font's are still _very_ useful in icons, favicons etc.

There is a nice thread @ Typophile dealing with a similar situation: http://typophile.com/node/61920

Subpixel rendering (used in the font referenced in the article) really ought to count as cheating. It effectively increases the horizontal resolution, so it can only make a real claim to be "smallest" in the vertical dimension. Furthermore, if you blow it up (which I'm guessing plenty of people are inclined to do), it actually looks worse, whereas the true bitmap font becomes, while clearly pixelated, more legible.

It's not just when you blow it up - I found the bitmapped font considerably more legible at 1:1 size. Hinting definitely makes sense at larger sizes, but when it's that small it tends to look a bit muddy to me.

Hell yes. Normally I complain that sub-pixel rendering suffers too badly from colour fringing on any display with a low enough resolution for it to be worthwhile.

In that case I'd say that the colour fringing suffered a little from having a hint of font almost visible beneath it, whereas the bitmapped font was actually pretty decent.

> Furthermore, if you blow it up (which I'm guessing plenty of people are inclined to do), it actually looks worse

Sure, if you naively blow it up it looks worse. But what if you convert the subpixels into real pixels?

Then the font would cease to be notable for anything other than its ugliness, as its size (measured in pixels) is now greater.

This reminds me of a class I didn't study well for. We where allowed one piece of computer paper for reference. I printed my friend's entire set of notes for the class using a VERY small font.

When I was done, he wanted a copy too :)

There is a smaller font than both of these two: http://typophile.com/node/61920

Obviously a strange new usage of the word legible that I was not previously aware of.

Probably due to me being old.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact