Log in

No account? Create an account

Sat, Jul. 29th, 2006, 03:21 am
metaeducat10n: Wikipedia Puzzle Ball in real 3-D

So I had the idea of building a real 3D model of the Wikipedia logo in the open-source 3D modeling program Blender. Rather than tackle my first 3D task all by my lonesome, I solicited some advice on what direction to take from an internet bulletin board.

Long story short: an artist named Hazard picked up the project and made the whole model...as well as a video tutorial (you'll need DivX to view the movie). There are of course lots of ways to render the model, including texturing the pieces with unicode characters, but here's one he did with a basic shiny globe appearance (click for bigger view):

I look forward to seeing what people do with it, especially in animations! The next big step is to identify a symbol to put on each piece (if the globe is going to spin around, it needs symbols on the back too!). If anyone can help with that, I've sketched up a bit of research on the topic, click it to see:

(As a sidenote, I thought the VfD on the Wikipedia logo article is kind of interesting; I would have thought it would have been in the Wikipedia: namespace or perhaps on Meta-Wiki. But the argument is that Google Logo sets precedent.)

Sat, Jul. 29th, 2006 11:34 am (UTC)

That's neat - I'm looking forward to the final version. :)

Sat, Jul. 29th, 2006 01:19 pm (UTC)

Very cool looking. :)

Sat, Jul. 29th, 2006 02:20 pm (UTC)

That's gorgeous. I hope you squeeze as much utility out of this idea as possible.

Sat, Jul. 29th, 2006 03:04 pm (UTC)

I have to admit that Rosemary Righter's zinger quoted in the article is pretty funny:

"Just above the omega, at the point where, on human heads, they used to perform frontal lobotomies, bits of the jigsaw are missing."

Sat, Jul. 29th, 2006 08:39 pm (UTC)
metaeducat10n: time will tell

I guess people are placing their bets, and history will judge based on the perspectives and opinions they take now. But I—personally—would not want to be one of the old-guard media maligning a project with so much potential influence on the future. I'll predict Wikipedia and its direct descendents in content to outlive any TV network or newspaper out there.

The London Times—for which Rosemary writes—uses The UK's Royal Coat of Arms for a logo. Seems like a bit of a glass house to be throwing stones from...when you've got a unicorn and the phrase "God and my right, shamed be he who thinks ill of it" representing your newspaper. Metaphors abound for monarchy, arrogance, and extinction. :P

(no offense intended to our British wiki contributors)

Sat, Jul. 29th, 2006 04:08 pm (UTC)

That's brilliant!

I hope some of the results will be GFDL'd, or copyrights assigned to Wikimedia (see http://meta.wikimedia.org/wiki/Logo#Policy_for_new_logos).

It would be very cool to have some high-end graphics to use in presentations, or to send to TV companies doing stories on us....

Sat, Jul. 29th, 2006 04:45 pm (UTC)

I can't believe the result was "keep." The Wikipedia logo has no notability and comparing it to the Google logo is a joke. I've seen the same biased phenomenon on articles related to relatively unknown Wikipedia volunteers and employees. It's like anything on Wikipedia gets include just by being associated with Wikipedia.
(Deleted comment)

Sat, Jul. 29th, 2006 06:30 pm (UTC)

Yeah, there weren't as many votes as you'd expect. It may still be hard though, as it seems Wikipedians will often vote to keep almost any articles related to Wikipedia.

Anyway, I forgot to say, nice job on the 3D model! It's especially cool that the guy made a tutorial for it.

Sat, Jul. 29th, 2006 08:53 pm (UTC)

How about taking the time that you would have devited to nominating stuff which already passed AfD and spending it on new pages patrol instead? It will do the project far more good.

Sat, Jul. 29th, 2006 05:37 pm (UTC)

Have you tried asking User:Nohat? He/She is the one who designed the current version of the logo.

Sat, Jul. 29th, 2006 10:41 pm (UTC)
metaeducat10n: contacting Nohat

I went to Nohat's User Page and saw someone has already given him a heads up on this thread. Good.

There seems to be reluctance on his part, and this is one reason why I didn't think to ask up front about his desire to participate. It's fairly clear that people do wish to see more coherence in the symbols, but he has said:

"The characters on the sphere do not represent anything in particular. There are no plans at this time to change the Wikipedia logo."

Not making the changes out of fear of jeopardizing the copyright reminds me a little of this story (borrowed from a page on intellivisionlives.com):

Mark and Keith Robinson, who was programming TRON Solar Sailer at the same time Mark was programming Vectron, both disliked the built-in Intellivision font, and they both hated that the "at" sign (@) was used as a copyright symbol on the title screens. They developed custom fonts for their games and made a special point of including a true c-in-circle copyright symbol in the character sets. They proudly showed off their custom title screens to Management, pointing out that for the first time Intellivision games would have correct copyright symbols.

Management vetoed their use. The argument: if there was ever a court fight over the legitimacy of the copyrights on the old games, Mattel could argue that the "at" sign should be accepted in context as a copyright symbol, since the Intellivision can't generate a true one. Ah, but Vectron and TRON Solar Sailer would show that the Intellivision could generate a true copyright symbol, thus jeopardizing the copyrights of all previous Intellivision games. Mark and Keith were ordered to use "at" signs, instead.

Feeling that this was about the stupidest thing they had ever heard, Mark and Keith went directly to Roy Ekstrand, head of Mattel's legal department, and presented their case. His decision: Mattel could argue that technological advancements since the earlier games now made it possible to use correct copyright symbols, where previously it wasn't. It would be safe to use the real copyright symbols.

Gabriel Baum, VP of Applications Software, was furious that Mark and Keith had gone over his head, and told them to "never do it again." Then he told them to use their real copyright symbols.

They later wondered: Was it worth risking their jobs over something as trivial as having the correct copyright symbol on their title screens? Their conclusion: "Damn right it was."

In response to requests to provide the sources for the logo, Nohat has said:

"Unfortunately, with issues of corporate identity, such as logos, it is important for Wikipedia to maintain control over its trademarks, and unfortunately this is one area where the spirit of openness that is central to Wikipedia must be curtailed, for the good of the foundation and the project."

...but as I think should already be apparent, there's much to be gained from leveraging the broader community's involvement in helping the logo evolve. Companies go through revisions all the time and they are able to protect their branding despite radical changes. If Google feels comfortable mutating their logo on their homepage for various holidays, I don't see why Wikipedia would think that their rights will be compromised by permitting variations.

Clearly, however, it's not official until it's official. So it may take time before Wikimedia elects to use a new logo on the site; that's ok. In the meantime, though, free agents can make some nice whiz-bang openings for their presentations at conferences...I really don't think anyone can convince me this is a bad thing.

Sun, Jul. 30th, 2006 05:23 am (UTC)
nohat: Re: contacting Nohat

I hope that my comments here will help provide some perspective.

First, even if I wanted to, I don't have the authority or the technical permissions to change the logo and any changes would have to go through a long and unpleasant community approval process that I am not keen to repeat.

When I made the logo, I was an unemployed and not-yet-graduated former college student. I have very little time anymore for Wikipedia. I had to resign from the Arbitration Committee last year, and nowadays I barely have time to made a few edits a week to Wikipedia. I love participating in Wikipedia and I am honored that the logo I made represents a project that is very well-known.

I made the current logo almost 3 years ago. I took the final rendering from POV-Ray into Photoshop and futzed with it until it looked right. What you see is the result of that. Since then, whenever I have re-rendered the logo with a slight change to the imagemap, I have never been able to reproduce the exact set of postprocessing steps I made to get the logo to look the way it does in the "official" version. At first, whenever I was petitioned to make some small change—fix the Japanese characters, or the Devanagari, I attempted to do so, but my fixed version never looked right. I no longer attempt to change the logo. Honestly, I am loath to even open the file in Photoshop anymore.

I know that there are much more talented and skillful graphic artists out there than me. I just took what I thought was a mediocre logo concept that was executed extraordinarily poorly, and made it seem a little more palatable using my limited skills and somewhat larger hubris. I was surprised that no one more talented than me came up with a better edition of the winning "logo concept".

I think the logo should be re-drawn in a vector format. Although I know how to use the pen tool in Illustrator, I have neither the time nor the skill to do recreate the logo in vector format. I wish I did, because I would like to fix the errors and improve the logo.

What I originally envisioned was that each language edition of Wikipedia would have the first character in the middle of their logo, with an aesthetically-influenced semi-random sampling of other characters around it. If anyone is interested in spearheading an effort, I would happily endorse any such project to make that happen. I am also willing to help with the limited time and resources I have available to me.

I hope, however, that if anyone takes me up on this offer, they will understand if I am reticent to assist if the plans steer towards dramatically altering the logo's appearance. Ideally, any new logo would be pixel-for-pixel identical to the current logo, except for the changes to characters, but I know as much as anyone that maintaining a high-quality high-resolution (or vector-format) image and an equivalent low-resolution web image is very tricky.

I'll stop now. I can't promise I'll remember to check back here for replies, so if anyone wants to talk seriously about the Wikipedia logo, he or she can leave me a message on my user talk page on Wikipedia with contact information. I'm happy to discuss.

Sun, Jul. 30th, 2006 08:48 am (UTC)
metaeducat10n: Re: contacting Nohat

Hello Nohat! Certainly I'm empathetic to issues of the time/etc. Reviewing my msg. my rebuttal to your talk page was a bit harsh, sorry, I just get a bit emphatic when it comes to open source issues...even if it's a logo. :)

I think the logo should be re-drawn in a vector format.

What I was wondering would be if it would be easier to implement this as an automated vectorization of a high-resolution bitmap? If the render were done to a very basic black and white image then a tracing tool could probably do a decent enough job. Drawing 3D-in-2D is a bit tough...

Ideally, any new logo would be pixel-for-pixel identical to the current logo, except for the changes to characters

Here I disagree; I think there's room for improvement on the puzzle ball's structure and how the characters are mapped onto the surface. I guess it will take some testing to see how much of an improvement the new model will yield; if it doesn't look better, and people don't like it better, then maybe the original will turn out to be the way to go.

Honestly, I am loath to even open the file in Photoshop anymore.

I have more than a few files on my HD that I feel this way about. :(

Sat, Jul. 29th, 2006 07:30 pm (UTC)
prezzey: Hebrew/Yiddish

The Hebrew is not a he, it's a resh:
But that letter is not in the word "Wikipedia".

Sat, Jul. 29th, 2006 07:59 pm (UTC)
metaeducat10n: Re: Hebrew/Yiddish

Is the character I picked the correct one? (below is from http://meta.wikimedia.org/wiki/Wikipedia_in_other_languages)

Sat, Jul. 29th, 2006 08:08 pm (UTC)
prezzey: Re: Hebrew/Yiddish

The character in the name is a dalet (ד) - they look kind of similar, but I think the one in the logo is a resh.

Sat, Jul. 29th, 2006 10:09 pm (UTC)

This is very cool, but would it be possible to put it behind an LJ-cut?

Sat, Jul. 29th, 2006 10:51 pm (UTC)
metaeducat10n: Image Placeholders

I'm afraid the picture is the post, more or less! If you subscribe to the notion that all pictures should be behind cuts, check out the image placeholders feature in your Viewing Options. You can make it so that pictures on your friendslist show up as links that you can click on...and you can do this for all images, or just images of certain sizes.


Sat, Jul. 29th, 2006 10:59 pm (UTC)
metaeducat10n: thumbnail

...though I changed the image to a thumbnail, if that makes it any better for you...

Sat, Jul. 29th, 2006 11:09 pm (UTC)

Okay, no (to more than half the identifications of characters).

Starting at the right hand side on the top and going in columns:

1) He identifies it as tlh. I'm pretty sure it's from the (traditional) Mongolian alphabet.

2) He identifies it as Arabic/Farsi/Urdu. This is correct, although he incorrectly judges that an "alif" is included, and that a rteh is added for ur.

3) Correctly identified as ko.

4) Incorrectly identified as the Latin character "U" or the Tifinagh character for (I believe?) "w". It is in fact a Thai letter. Not sure which one, but if you really want to know, I would figure it out.

5) It is indeed Japanese, but it is not the right character -- it says "kwi", for some reason on the globe even though this syllable does not occur in Japanese words and very rarely even in foreign words.

6) W is obviously correct

7) Cyrillic character

8) This is indeed a Hebrew character, but not the one he chose. Rather, it is a Resh.

9) Obviously omega.

10) This is indeed a Chinese character, but it is not the same one as he indicates. It means something different and has an entirely different sound.

11) This is a Laotian letter, not the Latin letter B.

12) I honestly have no idea. I can't see it very well. Possibly a Coptic letter.

13) This is a Khmer letter.

14) This is not the Latin letter H. It is a Tibetan character.

15) This character is identified correctly, but the characters in the identification are displayed in the wrong order.

16) This is not, in fact, a stylised "Y". It is, I believe, a Sibe letter.

And honestly, it would be easier to just ask Nohat (he designed the logo).

I am responsible for nearly all of the localisation of the logos (text that goes underneath and decision of the fonts for it).

I would recommend in addition:

1) The Divehi Wikipedia logo: http://dv.wikipedia.org/wiki/Image:Wiki.png People's opinions on Divehi script vary; some people think it is very beautiful, others say it "looks like bird poop" (literally).

2) The Armenian Wikipedia logo: http://hy.wikipedia.org/wiki/Image:Wiki.png

3) The Inuktitut Wikipedia logo: http://iu.wikipedia.org/wiki/Image:Wiki.png

4) The Buginese Wikipedia logo: http://bug.wikipedia.org/wiki/Image:Wiki.png

I can probably come up with a few more if you so desire, and if you are alright with it, I can provide additional letters from alphabets which currently are not used by a Wikipedia (mostly African ones, but including a few from elsewhere in the world as well)

Sun, Jul. 30th, 2006 05:36 am (UTC)
nohat: Almost entirely correct

Node's analysis is almost entirely correct. Number 1 is in fact a Klingon character, as identified by the original poster, not Tibetan. Node correctly identified 14 as the Tibetan. Every character on the globe comes from a different writing system. I intentionally picked characters that seemed most representative of the writing system rather than characters that were the first letter of "Wikipedia" in that writing system. This was mostly because the Greek would be beta, and that is not significantly different in appearance from B in the Latin alphabet, and wouldn't be identified to the casual observer as being Greek in the same way that omega does. Same with resh rather than vav for the Hebrew—when standing alone, vav just kind of looks like a styled l or I. For the other writing systems, of which I had very limited knowledge at the time, I just picked the symbol from that system's section of Unicode that looked the most striking to me.

Sun, Jul. 30th, 2006 09:08 am (UTC)
metaeducat10n: localization

Okay, no (to more than half the identifications of characters).

Thanks for the analysis! Awesome! Although I'll point out that the only characters I was testing against (the ones in boxes) were the first letters in "Wikipedia" from the Wikipedia-in-other-languages logo page. So those were the only candidates I was considering—despite knowing that Nohat was picking from a broader palette of symbols. The identification was thus a rough match against language...

I can probably come up with a few more if you so desire, and if you are alright with it, I can provide additional letters from alphabets which currently are not used by a Wikipedia (mostly African ones, but including a few from elsewhere in the world as well)

Absolutely—that would be fantastic! The amibitious goal is 56 unique symbols...but as I've pointed out some of those can be pretty bizarre as they would be on puzzle pieces typically removed from the globe. I liked Nohat's original plan of making these characters the start of "Wikipedia" in each language, rather than just a random symbol...if that's possible!

Sun, Jul. 30th, 2006 10:21 am (UTC)
node_ue: Re: localization

I would imagine it's possible, yes, but given the relation many alphabets have to one another, it seems likely that they are going to end up looking like certain characters in other languages, at least in certain cases. If that's not an issue, though...

I don't know how good your font support for different scripts is; if a certain character doesn't display, tell me and I can give you an image of it.

1) Amharic: ዊ
2) Arabic/Urdu/Farsi/Etc: و
3) Bengali: উ
4) Cherokee: Ꮻ
5) Gujarati: વ
6) Divehi: ވި
7) Hebrew: ו
8) Devanagri (Hindi, Marathi, Sanskrit, Bhojpuri, Sindhi): व
9) Armenian: Վ
10) Japanese: ウ
11) Georgian: ვ
12) Kannada: ವಿ
13) Thai: ว
14) Sinhala: ව
15) Korean: 위
16) Malayalam: വ
17) Tamil: வ
18) Gurmukhi: ਵ
19) Syriac: ܘ
20) Bopomofo: ㄨ
21) Buhid: ᝏ
22) Cyrillic: В
23) Hanunoo: ᜯ
24) Hiragana: う
25) Khmer: វ
26) Lao: ວ
27) Limbu: ᤘ
28) Mongolian: ᠸ
29) Burmese: ဝ
30) Oriya: ୱ
31) Osmanya: 𐒓
32) Babayin: ᜏ
33) Tagbanua: ᝯ
34) Tai Le: ᥝ
35) Telugu: వి
36) Tibetan: ཝ
37) Inuktitut, Cree, etc: ᐅ
38) Chinese: 維
39) Yi: ꀁ

I will post the rest in the morning

Sun, Jul. 30th, 2006 04:23 pm (UTC)
metaeducat10n: Re: localization

Awesome! I bet if we render a model of the ball and have it spinning around and zooming in on the letters...then put the video on youtube...lots of people would end up looking at it. That way, if there were any nits to be picked over the choices of letters someone would speak up pretty quickly. So you probably don't need to be too perfectionistic, as long as there's a list and a documentation of the rationale...

it seems likely that they are going to end up looking like certain characters in other languages

Closer duplicates could be on opposite sides of the ball. Perhaps in some cases it could be more than one letter (e.g. if "Wi" looked better than "W" we could consider that...)

The goal is to get a hi-res bitmap file of each letter to use as a texture (the suggestion made on the blender thread was 1024x1024, I don't know if that was supposed to be with the character taking up the full space or 1/9 of the total area in the center of the square). But the hope would be that the character could be "squarish"...I believe.

I don't know how good your font support for different scripts is

My browser doesn't seem to be doing a great job, and "view source" isn't giving me unicode values for some reason. If I had the unicode values I could look them up in character charts, and possibly dig up a nice scalable vector version...

Something I've been wishing for would be for Wikipedia (or another wikimedia project) to have an entry for every Unicode character, with an SVG of the letter. There's Category:Letters_by_alphabet, and I think there's a lot of purpose served by the articles on things like the article on W. I'm sad that the unicode working group doesn't have a "letter wiki" with free graphics for each character already—their reference versions of the letters are in PDF files with proprietary fonts (?!)

It certainly seems that there is a precedent for doing a good job of documenting symbols on wikipedia. See Symbol Dictionary's Ankh vs, say, Wikipedia's Ankh...it measures up pretty well on a few such entries, but there's a lot of symbols that don't have articles.

Thu, Aug. 3rd, 2006 11:40 pm (UTC)
metaeducat10n: more letters

Hi Mark,

Have you gotten any more of these letters thought out? If the letters were chosen and rough unicode numbers (or even just links to some semblance of what the image should look like) I can produce the texture bitmaps...

Just want to keep the ball rolling, so to speak, so that the blender guys don't lose interest before it's done!

Sun, Jul. 30th, 2006 07:39 pm (UTC)

Interesting. Maybe we can have some sort of fun short movie?

I need to learn Blender/Maya/some other 3d program :/