Slashdot Mirror
The World's Smallest Legible Font
hasanabbas1987 writes "From the article: 'Well 'technically' they aren't the smallest fonts in the world as if they were you wouldn't be able to read even a single letter, but, you should be able to read the entire paragraph in the picture given above... we did. A Computer science professor called Ken Perlin designed these tiny fonts and you can fit 500 reasonable words in a resolution of 320 x 240 space. There are at the moment the smallest legible fonts in the world.'"
Won't the actual legibility of the font have a lot to do with pixel size and spacing? Sure, you can pack that font into a tiny space, but if it's all broken up in jaggies, can you READ it?
Chaos maximizes locally around me.
There is the odd word here and there that I can't quite make out.
I would tell you which words those are but...
the world's smallest legible font.....peaks the smallest interest in my world
Reading the font is also made easier by virtue of it being a text many of us would recognize. Our minds would fill in the gaps, even if it wasn't completely legible. I suspect it would be harder to read a paragraph with font that small if the text was completely unfamiliar.
This author takes full ownership and responsibility for the unpopular opinions outlined above.
How many words can fit on the 3 by 5 inch flash card? Equations? Diagrams?
ts;dr
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
Beyond the first line, the text becomes too dense and I lose track of what line I'm on. I only need that, in order to know what it says, but that doesn't really count.
Is there fuzzy-ness built into the font or is that just firefox interpolating when I'm zooming in? One pixel is way to small on my screen to be able to read this, but maybe with larger pixels it would be ok. I guess I have to try to zoom in in gimp.
Skip the blogspam, go to the source: http://mrl.nyu.edu/~perlin/homepage2006/tinyfont/index.html
Just read it backwards, word for word. I have to admit it was a bit harder, but it was still legible for me. Considering that this is maybe three point font, I find it pretty noteworthy.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
Like the latin, small print is an obvious bad-faith action and should be disallowed. Schools specify allowable fonts and margins, there is NO reason to let lawyers do less than a law professor.
excitingthingstodo.blogspot.com
"There are at the moment the smallest legible fonts in the world."
I will give you all about 10 seconds to spot the glaring error.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
How many pixels would be required to hold one LoC?
If the PNG on the Slashdot article is derived from the linked source article then I am concerned that it may not be representative of the actual research as the source article offers the image as a JPEG - which will almost certainly have degraded the image quality.
That's pretty amazing. Except that the letters a, e, and o are nearly indistinguishable. To prove it is the smallest legible font, one would have to show that a long enough sequence of just the letters a, e and o could be spelled back by a reader. aeoeoaoeoeoaoeoaoeeeoaaaoeoaoa. I doubt it.
Practically speaking, that would mean a word like onomatopeia would be hard to identify. Of course, the context in which a word shows up probably accounts for more than half of the reason a reader can identify that word so quickly in a sentence.
It may be 'readable' if I tried, but considering its hardly legible I stopped after the first line.
Fitting more on a page isn't always the right goal, making useful information visible and readable is the goal.
This sort of silly shit is only of interest to advertising agencies who like to put out commercials telling you a bunch of bullshit that isn't true and cover their asses by saying its not true in print so small you can't read it.
Yes, I CAN read it. No, I wouldn't read anything presented too me like that, neither will anyone else, and thats most likely the point.
Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
I programmed a Sinclair ZX-81 you insensitive clod!
If libertarians are so opposed to effective government, why don't they all move to Somalia?
I was able to read it pretty well, but I think that was due in no small part to it being familiar content. If it weren't the Declaration of Independence, I probably would have had to strain a bit, so it probably has as much to do with gestalt theory as anything else.
Taking guns away from the 99% gives the 1% 100% of the power.
Very true. And a similar font existed in varius Apple][ graphic mode rendered text. 5 pixels tall 3 pixels wide + 1 pixel for line and letter spacing and you had the font.
Or if you go for the least pixels, use Braille points. 6 (2x3) points font.
Léa Gris
Ummmm... they made a small font! Who cares? How does this crap even make it to tech sites? On another note, I once drew a stick figure with mud when I was a kid!
Hasn't everyone who has ever taken a course where the professor says you can have a 3x5 note card during the exam already discovered this?
How about using the positions of the red, green and blue areas of the pixels as their own mini-pixels to make even smaller fonts?
Granted, it would only work on certain display types.
Technoli
That looks EXACTLY like the text I was staring at when I was a kid with my CoCo2 running a 72 column display. The fonts were all artifacted and all that. It was tough but it was at least a good thing that I was a kid and capable of dealing with it comfortably. The machine was originally intended to use a 32 column text display, but the 4 color "high res" display was too tempting for some to resist and they decided to write some word processing and desktop software for the thing. It worked...more or less... sorta... intolerable by today's standards but a feat in those days.
In JPG format.
Heres the original source (you skip two useless blogs)
http://mrl.nyu.edu/~perlin/homepage2006/tinyfont/index.html
-Woof woof woof!
Of Perlin noise fame?
"When in the poutse of human events it becanes necessarv far pne people to dissoke the pollitical bands which have comectef then widh pnother amd oo assinne annony dne power of the oaritv the soparare anf oquol statipn fo whiich the Lows of Wature anif of Nowre's Gof..."
If that's what the text is supposed to say, then heck yes it's a resounding succeff.
If I wanted a case of eyestrain, I'd have bought a shiny new 3D TV which would at least allow me to oogle larger than life boobies while hurting myself.
However, this font may produce some "hi-res" ascii movies... someone should run Deep Throat through the ascii encoder with it.
Did Ken Perlin not come through the 8-bit era? "Everybody knows" you can fit the entire English alphabet comfortably into a 3 pixel wide by 5 pixel high monospaced grid, it's been done hundreds of times. (Proportional can be even smaller, of course.)
...at what pixel density?
One thing I've noticed while studying Japanese is that I need to use a larger font size when reading Japanese than I do when I'm reading English. The characters are just too complex to differentiate at small font sizes. You can't easily distinguish a character with N horizontal lines without the character being 2N-1 pixels tall (one for each stroke, and another for the gap between strokes). There are common characters with as many as 8 horizontal strokes (The kanji for "kaku" for example, which conveniently means to write). Even today's video games (on nintendo DS, for example) have 11x11 pixel Japanese fonts that can be very difficult to read.
Any reduction in font sizes for readability must have separate standards for Asian characters, or the more complex ones will just appear as blobs.
This is only legible if you have a decent monitor and reasonably good eyeballs. Also helps if your monitors dot pitch is not excessively high and your web browser isn't set to automatically scale images (even minor artifacts are going to render this nigh-unreadable). Even then it's a bit tricky in a lot of places. I certainly wouldn't want to read a lot of text with this font.
I read the internet for the articles.
I remember back in high school, some teachers would allow you to use a 3x5 note card for tests. Anything you could write on the note card, you could use in class. Think how much information you could cram onto it using a font this small.
The only thing necessary for evil to triumph is for it to be pitted against a slightly greater evil
> And a similar font existed in varius Apple][ graphic mode rendered text. 5 pixels tall 3 pixels wide + 1 pixel for line and letter spacing and you had the font.
FTFY.
HGR. 280x190, with color bleed. Vertical lines would be green or pink.
e.g.
http://www.retrocpu.com/apple-ii/images/games/c/castle_wolfenstein.png
> A Computer science professor called Ken Perlin
Understatement of the year -- this guy _invented_ Perlin noise.
I consider it a quite large but ilegible font.... or is it just my eyes?
Couldn't see a link to actually download it after going thru to the actual original source page.
Try again. I bet somebody converted to jpg before converting back to png.
8 lines are a "waste." I had a 5 pt font on my HP48SX/GX. Someone even hacked in lower case case. (Technically, 6 pts, since you need 1 line for spacing.)
See the editor included with the "Jazz" assembler.
http://www.hpcalc.org/hp48/programming/asm/
Forgot a pic of the font ...
http://www.hpcalc.org/hp48/programming/asm/sshots/jazz68.gif
I "read" a lot of that text from memory; my brain knows what the words are supposed to be because it's the danged preamble to the danged consitution. If they post a test with unfamiliar prose, I bet I'd have much more difficulty reading it.
Now I can fit ten times as many documents on my hard drive!
That's 1337 in 1776.
I eat only the real part of complex carbohydrates.
Which on my display is a shedload easier to read than the cited example.
Course I had a hp48...
Many small devices detect screen pivot these days, but this is a single orientation font.
This reminds me of the Slashdot article on how humans actually process words while reading This opens up chances for even further optimization by focusing on enhancing the readability of letters that mostly appear as first or last letters of words !
I thought this kind of font fell under the purview of pixel fonts. So it's not really anything new except that this particular font uses shades of grey to indicate certain shapes. I'm hard-pressed to call it truly legible. It's borderline; I have to concentrate a bit to identify certain letters and some of it I'm deciphering mostly because of context.
I do think this font may have a future as legal info and disclaimer copy.
On my Mac, 8.1-point Helvitica is more readable, has more leading, and takes up only a tiny bit more space. With a tweak to your system prefs*, 7-point Helvitica takes up about 2/3 the space and is equally readable. *Go to System Preferences > Appearance > Turn off text smoothing ..., and change the setting to 4.
"Reading the font is also made easier by virtue of it being a text many of us would recognize."
Personally, I recognise it because this looks like any other sans serif ~8 pixel-high font from any low-res game of the early 90s or so. Nothing special AT ALL here.
I programmed an Atari 800, so neener!
I'm really a low 5-digit Slashdotter, but this ID is where I am now.
Exactly. I'd like to know where this Comp Sci professor got his qualifications, and whether he's old enough to have been around when computers had a res below 800x600.
if I remember correctly, Tasword on ZX Spectrum used something similar.
my 24" LCD display is only 320x240 maximum resolution. I'm so glad that someone has made fonts that will help fit text on my screen. They show up nice and large on this display. What's everyone complaining about?
jsut athnoer menagiensls ltitle psrhae for you to dcoede. Why do we wtsae our tmie dnoig tihs?
My compound eyes were annoyed trying to read them, and I gave up. :(
Ant(Dude) @ Quality Foraged Links (AQFL.net) & The Ant Farm (antfarm.ma.cx / antfarm.home.dhs.org).
This news brought to you by the smallest web host ..
Hey don't blame me, IANAB
I imagine InstallShield will license these at once for click-through software licenses which need fine print -- really fine print.
"Imaginary solutions to real problems."
Meh, this one has smaller letters, and is still marginally readable. Although with the linked text, the words make it easier to work out what it says.
Ask me about repetitive DNA
like credit card terms
if this is supposed to be a new economy, how come they still want my old fashioned money?
Ok, so he did a sub pixel font, which is a win for LCD screens, but that creates one big problem: the most popular small-form-factor devices these days have screens that automatically change their orientation when you rotate them. Suddenly those beautiful sub pixel fonts become a bunch of gibberish.
Tired of FB/Google censorship? Visit UNCENSORED!
I'm looking at it with bifocals and without, no joy, I would have to use a reading glass to see this.
I killed da wabbit -Elmer Fudd
1280x1024 (native res) on a Dell 1905FP, I'm not more than 3 feet away, and it looks like crap. I can barely make some of the words out. I do wear glasses, but that would mean I *should* help, one would think. Scaling it up merely makes it look even shittier.
This means I will be able to fit twice as much of the Internet in the same space as the old internet.
Nice, the linked article got a hit in Trend Micro.
Web reputation result: This URL is currently listed as malicious.
Well, more likely something loaded on that page is.
I tried every decent and legal way I could think of to resolve the issue w/the business before I rented the chicken suit
Call me a traditionalist, but a font is not defined to be antialiased. That's a color image, not a demonstration of a font.
The only thing worse than a Democrat is a Republican.
Link in the article is 404.
Ahhh, what an awful dream. Ones and zeroes everywhere... and I thought I saw a two.
Yes, most of us old-timers remember tiny bitmap fonts on various systems and GUIs. Some of us even made our own for various purposes.
This is a new thing because it's treating the RGB subpixels of an LCD as individual pixels, effectively tripling the horizontal resolution of the font. I don't remember seeing anyone on any system doing that. This allows characters to be more true to their intended forms. For example, the letter W can now be done in a much narrower space - 3 pixels wide to have a recognizable W is pretty good.
In Braille, there needs to be space between the dots in each character and again between the characters and lines, so you need to more than double that. Go check your local ATM, pay phone, or restroom door sign.
BTW... 6 points... while I see what you were saying, points are a typographical unit (approximately 1/72 of an inch) so you should probably say square pixels instead!
Agree, this is not legible, especially when enlarged. And, here's my font from a good while ago which is not only slightly smaller (or would be if it was variable pitch) but also a good deal more readable. Can be enlarged without loss, too.
The creatures outside looked from Alt-Right to Antifa; but already it was impossible to say which was which.
There is another tiny font in development Flea's Knees that also exploits subpixel rendering to aid legibility.
I am becoming gerund, destroyer of verbs.
When these cost several hundred dollars apiece. You really crammed in the 26/52 characters, 10 numerals, and bit of punctuation- 25+ bits each.
The image in the summary is really quite blurry compared with the same thing on the original article, which is crystal clear.
Original article: http://mrl.nyu.edu/~perlin/homepage2006/tinyfont/index.html
For quite a while I used a terminal emulator on an Atari 800 computer which gave an 80x24 fixed-width font. Atari 800 resolution is normally 320x192 making the characters 4 by 8 pixels including spacing. Since it was fixed width, it didn't get to make the i 2 pixels wide or the t 3 pixels wide, so taking that into account it was the same size as this font.
Actually, if you don't know what comes after When in the course of and events then you probably failed elementary school history if you live in the US. By using a well known quote the mind, already knowing what to expect, will see what is expected. If he really wanted to show the readability of his font, then he should write something original rather than using a quote every US citizen has seen and internalized to one degree or another.
Guns don't kill people; Physics kills people! - John Lithgow as Dick Solomon on Third Rock From The Sun
> A Computer science professor called Ken Perlin
Understatement of the year -- this guy _invented_ Perlin noise.
Agreed. There were two papers published in SIGGRAPH 1985 that really introduced "Solid texturing" to the computer graphics community. One of those was Perlin's An Image Synthesizer which has since formed the basis for numerous procedural texturing systems, especially those in ray tracing systems.
This doesn't seem much smaller than Tom Fine's Atari-Small font (maybe a pixel shorter?). I expanded this font to a full DOS extended ASCII version for my Doom source port.
were you expecting to see a sig here? perhaps you'd rather see the inside of an ambulance!
In the example picture in the article, the space actually occupied by characters is 282x203px. Not QVGA. Its readable though.
I even cheated in my disadvantage by actually cropping of some antialiasing thats part of the characters at the edge. So really, its somewhat larger. That is 57246 pixels.
Vertical average line height: 203 / 43 lines = 5.9705882352941176470588235294118
I assume this is the whole text: (excuse me, my copy paste for some reasons stops working when visiting Slashdot.com in Chrome) .. upto "the State remaining in the mean".
link
This is 3106 characters, including spaces and not including linefeeds. Mind that due to the small width of the image, some spaces got dropped saving a major percentage of space. This is not accounted for. I'm to lazy to actually count the characters and spaces in his image.
Spread over 43 lines, this is 91,352941176470588235294117647059 characters per line average.
Horizontal average character width: 282 / 91,35..etcetera = 3,0867285254346426271732131358661.
I know that the characters A-Z 0..9 a..z and most punctuation fit in at least 5x3 pixels. In this case line height could be 6 pixels, or less if you interpolate on say, thirdh or half pixels.
At 5x3 pixels, the same text would require (minus 43 space characters at end of lines, so 3063 characters) a 3063*3*6 pixels, = 55134.
5 because lines are separated by a whole pixel.
My solution takes less pixels. There are aproximatly 2112 pixels left, good for another 117 characters.
Hivemind harvest in progress..
I can read the font on my 1280x1024 work machine screens, so next I tried it on my N900. It has a roughly 3.5" widescreen with an 848x480 resolution, so the individual pixels are practically invisible. Unless you look really close up (where you can sort of see some jaggyness if you strain your eyes) it just looks like some kind of magical infinite-resolution screen.
I can still read it but it's really, really hard.
When I go to portrait mode it's still barely legible, but it seems slightly easier than in landscape mode. Maybe the screen was originally manufactured to be used in portrait mode?
Ow now my eyes hurt and I have to take a break.
"When information is power, privacy is freedom" - Jah-Wren Ryel
That just looks like noise to me... get it...? :)
"When in the course of human events it becomes necessary for one asshole to dissolve the political bongs which have connected them with another and to assume alimony...."
I don't think the message is coming through quite right.
"I don't care about the Constitution!" --Bill O'Reilly, November 17, 2009
It would have made cheating a lot easier.
We haven't even seen the text without clear type yet.
Open it in GIMP, choose Colors > Components > Channel Mixer, and set Red and Green to 0% Red, 100% Green, and 0% Blue. This makes a grayscale image out of the green channel, which has the effect of removing all subpixel information from the image.
Not.So.Legible...
ROFL "Assume Onions" 8p
It takes 40+ muscles to frown, but only four to extend your arm and bitchslap the motherfucker