Page 1 of 2

Scanahand 3 - a little underwhelmed.

Posted: Thu Jan 28, 2010 3:33 pm
by mully
honestly the only feature I really dig is that I can set a font style now. And I mean really dig.....

the improved trace algorhythm isn't that much better and I'm still missing a (even if VERY basic) kerning editor.
If it had a new algo called HYPER that simply gets rid of all those raggedy edges - I'd be happy and I'd be upgrading....

oh yeah upgrading isn't that cheap either.

think I'll have to wait till the next release.

for everyone starting out: it's still a very cool app.... just not worth the upgrade price imo


Re: Scanahand 3 - a little underwhelmed.

Posted: Thu Jan 28, 2010 5:02 pm
by Erwin Denissen
We do have some cool enhancements (for example improved metrics engine and auto kerning) on our todo list. Most will probably make it before we release Scanahand 4, but we haven't decided yet.

Manual kerning is not something we intend to add, since in our opinion such feature really belongs in a more sophisticated font editor like FontCreator.

Re: Scanahand 3 - a little underwhelmed.

Posted: Wed Jul 06, 2011 4:09 pm
by Timo Kähkönen
I suggest to High-Logic to make a deal with Italian Igino Marini, who has GREAT letterfitting tool (or service)

The quality is said to be better than in Illustrator Optical Kerning. Please see this comic, that is lettered with font which is kerned by Igino Marini with his iKern tool: ... lang=en_US

If someone has interest enoug, I can make here a sample that has three adjacent alternatives for helping comparation of different kerning results: without kerning, with iKern, with Optical Kerning.

Comparing iKern and Optical Kerning:

Could this be the right way?

Re: Scanahand 3 - a little underwhelmed.

Posted: Wed Jul 06, 2011 9:35 pm
by Timo Kähkönen
Here is the comparation between No kerning, iKern and Adobe Optical Kerning:

The font was hard kernable, because the italic angle varied, but my eyes see light when looking iKern result. Adobe Optical Kerning is nearly similar than iKern. Without kerning this font is not nice: see AVA and TAT.

Re: Scanahand 3 - a little underwhelmed.

Posted: Wed Jul 06, 2011 11:00 pm
by Dick Pape
Any idea what an Ikern would cost on simple handwritten font?

Re: Scanahand 3 - a little underwhelmed.

Posted: Wed Jul 06, 2011 11:19 pm
by Timo Kähkönen
Dick Pape wrote:Any idea what an Ikern would cost on simple handwritten font?
Year 2008 Igino Marini gave me price: 100 EUR per font or 300 EUR per font family (regular, italic, bold, bold italic). I got one font free of charge.

In my ears 100 EUR sounds rather expensive, but if you have to make it manually (it lasts several hours), the price is okay. I suggested Marini to establish a web service, where users can kern their fonts and pay with credit cards. I have not found such service yet.

The iKern service consist of fixing spacing (left and right side bearings of glyphs) and making kerning pairs or classes. By fixing spacing to optimal, the count of needed kerning pairs reduces dramatically.

Re: Scanahand 3 - a little underwhelmed.

Posted: Thu Jul 07, 2011 5:03 am
by Bhikkhu Pesala
Timo Kähkönen wrote:Could this be the right way?
I very much doubt it. Paying a license fee to a third-party is not a good way to make money unless it radically increases the number of copies sold. It would be much better to spend some development time on improving the auto-kerning within FontCreator. That would be a one-off cost, and the intellectual property would remain with High-Logic to use in all future versions of Scanahand and FontCreator, with gradual improvements building on that development work in future versions.

Re: Scanahand 3 - a little underwhelmed.

Posted: Thu Jul 07, 2011 12:57 pm
by Timo Kähkönen
I have to remind that Scanahand has been on stock 3 years and FontCreator at least 8 years. And still there is missing autokerning and autospacing with quality of iKern or Adobe Optical Kerning.

Igino Marini said to me, that it has been several years work to make iKern tool. I have tried to make autokern-tool by myself, but without success. My brother, who has matematician pro, denied to even try.

- Little fee per licence is not bad, if Marini wants to give a licence.
- But if High-Logic has time and money, i have to say: Make it. It's possible. Marini has coded it and Adobe has coded it, so why not High-Logic?

Re: Scanahand 3 - a little underwhelmed.

Posted: Thu Jul 07, 2011 1:34 pm
by Timo Kähkönen
An alternative way to add kerning pairs is to use Adobe Illustrator's script engine. Adobe has Optical Kerning, which can be selected as kerning method between two chars. Adobe has also javascript engine in it's layout programs.

Using standard javascript is possible to iterate over all possible (or needed) glyph pairs, add them Optical Kerning as kerning method and measure the kerning value between letters. After iterating all, unneeded (too low kerning value) pairs can be removed.

I tested this and I don't find any other way to get kerning values than saving two files and comparing them: one svg file without kerning and other svg file with kerning. In svg there is the x position of every letter. Comparing these x values of second letter of pair, can be used as kerning value. Of course the unit has to be taken into account, because kerning values in font are as em, which is usually 2048 in TTF and OTF.

The problem is that the speed of javascript in Illustrator is poor and it lasts few minutes (hours) to get results.

Re: Scanahand 3 - a little underwhelmed.

Posted: Thu Jul 07, 2011 4:32 pm
by Timo Kähkönen
Seems to be that the kerning value can be picked easier: comparing visible bounds (width) of unkerned and kerned text.

Below is the code. Save it as kern.js and in Illustrator run it: File / Scripts / Other scripts ... (and open kern.js)

- add textframe with glyph pair "TA"
- make it unkerned and get the width (noautokern)
- make it kerned by OPTICAL KERNING and get the width (optical)
- calculate (optical - autokern) * em/fontsize and you get the kerning value of pair "TA"

Now there is only missing the iteration through glyph pairs. And reducing the count of pairs using some minimum value, eg. 50. And maybe some HSB and RSB adjusting to reduce count of kerning pairs. And adding glyph pairs with FontCreator or other font editor.

So, this means that Illustrator owners can calculate kernings rather easy - while we are waiting HighLogic to implement it's own autokerning and -spacing.

Code: Select all

// This script gets kerning value of pair of characters
var font="ik2_TAAIMOUNUUSIFONTTI13";
var pair="TA";

if (documents.length==0) documents.add();
var docRef = documents[0];
var textRef = docRef.textFrames.add();

textRef.contents = pair;
//textRef.selected = true; = 400;
textRef.left = 100;
// Use here postscript name of font
var font="ArialMT";
var font="Arial-ItalicMT";
var font="Arial-BoldItalicMT";
var font="CourierNewPS-BoldMT";
textRef.textRange.characterAttributes.textFont = textFonts.getByName(font);

// Set font size to em / 4
var em=2048;
var divider=4;
textRef.textRange.characters[0].characterAttributes.size=em/divider; // 2048/4

var noautokern=textRef.visibleBounds[2];
var optical=textRef.visibleBounds[2];

var kerning = (optical-noautokern)*divider;
kerning = Math.round(kerning);

var textRef2 = docRef.textFrames.add();
textRef2.contents = kerning;
//textRef2.selected = true; = 400;
textRef2.left = 250;


Re: Scanahand 3 - a little underwhelmed.

Posted: Thu Jul 07, 2011 8:16 pm
by Dick Pape
I don't believe Scanahand is designed to give well perfectly designed fonts. You have to use a font editor, such as Font Creator, for that result. Font editors have both manual and autokerning routines.

Have you tried Font Creator kerning?

Re: Scanahand 3 - a little underwhelmed.

Posted: Thu Jul 07, 2011 8:39 pm
by Timo Kähkönen
Yes I have, but the result is not so good than in iKern or Optical Kerning. I have version 5.6. Maybe it is better in newer version. In my version for example pair "T." becomes too close.

I tested Illustrator script, made kerning pairs with it's Optical Kerning and the result was very close to Optical Kerning although I removed pairs with kerning values lower than 50.

Re: Scanahand 3 - a little underwhelmed.

Posted: Mon Jul 11, 2011 1:05 am
by Timo Kähkönen
Adobe Illustrator and maybe Photoshop, Indesign and Flash can be used as a kerning tool.

I made a fully functional javascript for Illustrator (tested on CS4). The script picks kerning values using Optical Kerning and these values are imported to font in Font-Creator. When using this like kerned font, the result is amazing:

See UTH in SOUTH and OTL in SCOTLAND. They are like ligatures!

The script is here:

The script outputs FontCreator-compatible kerning data, so applying to font is easy.

And here more kerning samples and explanations: ... ngTool.pdf

I had a little problem to find out which pairs to test. It's clear that different scripts are not wise to mix, Cyrillic and Greek letters are not usually in a same word. I searched a little and find KernMaster tool which seemed to have a comprehensive list of 4666 pairs that usually needs kerning in ISO 8859. Then there is 1156 pairs in FontCreator's Kern_extended.txt. I interpolated them and to make sure added all uppercase-uppercase, uppercase-lowercase, lowercase-lowercase combinations of A-Za-zÅÄÖåäö and got a list of 5971 pairs. I know that my list is huge, but the idea is to remove all unneeded pairs, so the result will be 2000-3000. If interest, here is my list of pairs:

Re: Scanahand 3 - a little underwhelmed.

Posted: Mon Jul 11, 2011 2:54 am
by Dick Pape
Every reference to kerning I've seen indicate it is a long and tedious process. To do it right you have to test the pairs in a myriad of settings to get the coloring correct. Perfectly you would consider combinations from other languages. Auto kerning is therefore a first approximation.

Test Words:

Code: Select all

"Ataif" ou 'Ataif'. O cake do chef d'oeuvre! Making of [of] (of) 'of' "of" of? of! of*. Making of [of] (of) 'of' "of" of? of! of*. Ydes, Yffignac, Ygrande, Ypres, Les Woevres, a Foret de Woevre, a Voire e a Vauvise ficam na Franca. Yves e D'Amboise sao aves. Lyford e no Texas e L'Anse-aux-Griffons em Quebec; Lyna, na Polonia. Yriarte, Yciar e Ysae, em Yale. Kyoto e Ryotsu no Japao, Kwikpak no delta do Yukon, Kvaeken na Noruega, Kylu no Quenia, nao em Rwanda& Walton, West Viriginia, mas <<Wren>>, Oregon. Tlalpan fica perto de Xochimilco, Mexico; os Zygos e Xylopaghou ficam no Chipre, Zwettl na Austria, Faen na Dinamarca, Varingsfossen e Vaeroy na Noruega; Tchula no Mississipi, Tittabawassee em Michigan. Twodot em Montana, Ywammum na Birmania. Yggradsil e Ymir, Yngvi e Vaden, Vidrid, Skeggjyld e Tyr ficam no Eddas. Torberget e Vag na Noruega, Ktipas e Tmolos na Grecia. Vasquez fica na Argentina, Vreden na Alemanha, a Von-Vincke-Strasse fica em Munster, Vdovino na Russia. Iterbio e Ytterbium em ingles. Touissant L'Ouverture, Sera que Wolfflin, Wolfe, Milosz e Wu Wu estao na biblioteca? 1510-1620, 11:00pm e o que sao os 1980s?
ps I would not consider those "ligature-like" combinations correctly kerned. I would manually correct the PJ spacing for the same reason.

Re: Scanahand 3 - a little underwhelmed.

Posted: Mon Jul 11, 2011 12:48 pm
by Timo Kähkönen
I tested PJONGJANG and SCOTLAND with optical kerning and yes - the result is similar to above example. That is one of the reasons, why iKern beats optical kerning.

In my eyes PJ has too much space between them, but I'm pleased with kerning of OTL and UTH.

Despite few beauty flaws, optical kerning is in my eyes better than no kerning at all. And those flaws can be adjusted manually if needed.