> ink- and light-traps through blurring.

Funny - I tried some experiments exactly like that, about a year or so ago! I was looking for automatic incorporation of trapping. But in the end I had to conclude that it doesn't really work...

You're doing a Gaussian blur then a cutoff, right?
Well, it looks cool!

hhp

Hmm...I'm going to guess he employed the roughen command in Illustrator using smooth points and a very low distortion.

Rodolfo - Make it unicase with lowercase-glyph
alternates and you have a buyer here.

The straight blur/bitmap treatment doesn't
produce the bowed curves; same with the roughen.

Is it a Fontlab trick (unfamiliar with that
program) or some variation on punk/bloat?

bj

Hrant - what i meant to say was he did something to bow the straight areas *before* the blur/bitmap. The blur/bitmap alone will round corners and clog counters, but won't make the "wavy" effect.

He might have also used the "wave" filter in photoshop. There are infinite settings...

bj

Actually, when you apply a cutoff to a Gaussian Blur, that "wavy" effect is exactly what you get!

hhp

If this is made publicly available everyone who
ever used Impact would kick themselves for not having
waited for Tyson.

Thanks... I guess...

The only think that bothers me is the quick end to the
descending 'J'. Can't it be lengthened without spacing
issues?

Spacing heavy type is a problem because of J,T,L,F and
the other characters that create "holes". But I think this
can be done here, yes.


Rodolfo - Make it unicase with lowercase-glyph
alternates and you have a buyer here.

You know something? That's exactly what I had originally
thought of doing - especially because I didn't want to
make a full UC + lc sets -, but ended up doing an UC-only.
I'll think of it.


About the Tyson B treatment (this sounds scary! ;) :



It's all done in Photoshop: the image of the blurred
characters has its tonal range flattened (I used 20%
to 80%) and spots of "blurred tone" (one "master spot"
created with the airbrush tool), in the light and dark
extremes of the full range, are applied in the
appropriate places (this manual part, through copy &
paste, took only a few minutes - especially since many
of the spots fall at the same heights). The tonal range
is normalized (to the base image's values), it's loaded
as a selection, converted to paths and ... voil

Actually, when you apply a cutoff to a Gaussian Blur, that "wavy" effect is exactly what you get!

Explain this, please.

Grande Capeto,
Novamente, mandou muito bem!

I found that the ridge of the base of the G in
the A version, was very exaggerated (Would be
inktrap?) and that the S, has in its starts/end
leaving for the edges. Also I agree to Coles on
the descending of the J...

Tyson is really good name that matches with a
grotesk face. But, unhappyly already he was used
to nominate a typeface by Neville Brody.

Look at:
http://www.fontfont.de/packages/dome6623/tys1900/tys1900_xx.html

Fala, F

Rodolfo,
I hate to make the spoilsport, but if you plan to release Tyson, you must know a typeface named Tyson (by Neville Brody) already exists in the FontFont library.

Algorithmical treatment of outlines is very interesting. I use it often when designing logotypes.

Sorry, I didn't see Fabio already pointed out the naming problem. I'm too in a hurry!

Rodolfo, give me until this afternoon.

hhp

I'm not sure if this has been mentioned or discussed yet. In regard to this typeface. I can see this in motion, as if it is wiggling or vibrating. I do like that quality. Although maybe it's just me. That said, is it the monitor or is it my brain trying to straighten out the edges?

Rodolfo, I just realized: I did what you did.
(No filters necessary: Blur -> Levels -> Threshold.)

hhp

you must know a typeface named Tyson (by Neville Brody)
already exists

You know what would be nice? A searchable universal
typeface-name bank database. Oficially updated by all
the foundries and independent developers. I once saw a
site that intended to be something like that, but it was
someone's effort only - an impossible task, of course -
and wasn't very usable.


That said, is it the monitor or is it my brain trying
to straighten out the edges?

Tiffany, that algorithm should theoretically affect only
the corners and leave the straight and curved edges
relatively unchanged, though it doesn't really work like
that. It creates "bows" and is quite asymmetrical. :/
Next time I'll do something different!


(No filters necessary: Blur -> Levels -> Threshold.)

Hrant, this doesn't work. This creates exactly the
effect the traps should be acting against:

Hmmm. I must be missing a step - maybe a "Minimum".
But I remember my results being the same, and the
whole thing being automatic. Give me a bit of time...

hhp

The step I was missing was an Unsharp Mask,
a nice big one, after a nice big G-Blur.

Instead of doing a cutoff, here's a Posterize:
http://www.themicrofoundry.com/other/faketrap.gif

hhp

Hey, Hrant,interesting! I would like to see the original
image. For the sake of completeness (and the
scientific method), I've tried to reproduce your
results, with the same base image that I was using
above. With some tweaking of the g.b. and u.s.
controls, the best I could achieve was this:



But I'd have to try at a larger scale, too.

In any case, the problem with this type of operation
lies more in the tracing phase. The paths-generating
algorithms aren't good enough.

Nice, Stephen! The problem is the initial "populating"
of the database. Then, updating becomes easy. ;)

> each foundry could update their own listing.

How do you prevent abuse?
Like I have a list of dozens of cool names, and I could put those up to prevent others from using them.
(BTW, unlike Rodolfo, I see a huge value in naming.)

--

Rodolfo, I like yours the best, but it's not automatic... BTW, you might want
to include the original "G" too. And can you give me that as an EPS?

> The paths-generating algorithms aren't good enough.

Are you sure? What are you using, at what resolution?

hhp

(BTW, unlike Rodolfo, I see a huge value in naming.)

It's not that I don't see value in naming. It's that, for any
cool name that's already taken, I can think of a dozen other
equally cool ones that aren't.

Rodolfo, I like yours the best, but it's not automatic... BTW,
you might want to include the original "G" too. And can you
give me that as an EPS?

Sure.

> The paths-generating algorithms aren't good enough.

Are you sure? What are you using, at what resolution?

Well, Photoshop (for lazyness), at a body size of 600 pixels.

Wait, you're using Photoshop to auto-trace?
I didn't even know it could. But tracing algorithm quality varies a lot (even -no, especially- within the products of a single company), so try Streamline, or even FontLab. And data from digital processing (like you're doing) tends to be very "clean" - it should trace exceptionally well, if you have enough bits - try 1000.

hhp

Not auto-tracing an image: "converting the selection to paths" -
well, that's essentially la m

I think even Fog's tracing would be adequate, with enough resolution (by "1000" I meant the height in pixels).

hhp

Regarding Stephen's name database idea, I think this is an excellent thought and would be very welcome. The main problem with the Goudy Center database is that they wanted to charge money for name searches, which put people off using it or even bothering to submit to it.

FontLab now has a nifty 'Check at MyFonts.com' button in the FontInfo naming panel, which goes off and checks to see if MyFonts lists any fonts with the name you are using. This is handy, but of course is limited in that it only checks the one source. If another freely accessible database were available, FontLab could check multiple sources and so provide more reliable feedback.

Regarding the kind of abuse that Hrant suggests might happen (people submitting cool names for as yet non-existent typefaces to prevent other people from using them), you can only legally claim trademark on a name that you are actively trading goods or services under in the marketplace. You are also legally obliged to defend a trademark: there is no automatic protection. So if I start selling a pair of fonts called Nour & Patria, Hrant would be obliged to sue me or lose his trademark on the names. I suggest that names should be accepted for the database only if the submitter is claiming them as trademarks, i.e. submission to the database constitutes a trademark claim.

I suggest that names should be accepted for
the database only if the submitter is claiming
them as trademarks, i.e. submission to the
database constitutes a trademark claim.

Then you get into a swamp, as local laws regarding
this are different around the world, aren't they?
(Even the suing thing.) This is supposed to be
international, I guess?

Right. I've got a lot of the data, I just need to put
a (inter)face on it.

BTW, Stephen, in what format is this data? This would
be a really great resource if the persons or company that
would give you technological support created a flexible
database format where one can hang pieces of data of
several types on each register. I mean, at least: an URL,
maybe an image, links to other related registers in the DB
(with a flexible definition of what a "relationship" is), a text
field with references, etc. And of course everything
searchable according to a number of possible query "views".
Wow, now I'm dreaming! ;)
Maybe you could get tovarich Yarmola into the idea? ;)

Rodolfo - If this is made publicly available everyone who
ever used Impact would kick themselves for not having
waited for Tyson.

The only think that bothers me is the quick end to the
descending 'J'. Can't it be lengthened without spacing issues?

You know what would be nice? A searchable universal
typeface-name bank database. Oficially updated by all
the foundries and independent developers.

There is The Goudy International Center, but who knows
when it was last updated. The website hasn't been
touched since '97.

I've been working on something very much like you're
suggesting. The idea is that is free database in which
each foundry could update their own listing.

"Nice, Stephen! The problem is the initial "populating"
of the database. Then, updating becomes easy. ;)

Right. I've got a lot of the data, I just need to put a (inter)face on it.

This topic probably needs a new thread, but I'll briefly
describe the project: TypeTracker is more than a
database of font names, but a general reference
directory of fonts, designers, foundries and resellers.
Each of these 4 databases will be cross-referenced and
searchable with links to font samples and other
relevant pages on the web.

The owners of each entity will be able to modify their
entry a la the directory at Briar Press. (I've been hoping to
speak with the creators of BP about working with me to
develop TypeTracker, but I'm open to ideas and
collaboration from other designers and database engineers.)