Font tracing is usually done by printing out the character to be traced at very large scale -- I've seen about 12" x 12" -- and placing it directly on a large digitizing tablet. A sequence of strokes / points is collected for the outline of the character, and then curves of somewhat reduced degree are fit to those strokes / points to both reduce font data size and reduce the impact of errors, inaccuracies, and quantization in the data capture.
Even at this huge scale, and with this amount of effort, the outline of your character will be very close to -- visually identical to! -- the starting character, but not exact. As a result, the generated font program will be quite different. For example, it may use a different number of control points for equivalent curves.
Now, one can imagine automating this process differently: Take a font file, digitally render each character, perturb it a small amount, and resynthesize the strokes to generate a new, different program for a visually identical font. This is generally against the terms of service for the initial font, however, which would make it a legal matter...
> This is generally against the terms of service for the initial font, however, which would make it a legal matter...
I doubt there's anything in the ToS for most fonts prohibiting me from rendering a short story that just so happens to contain every character and post it online for everyone to enjoy. I couldn't possibly predict that my friend who doesn't even know the name of the font, let alone ever agreed to any ToS, would take that render and trace all the characters on it.
Note that I generically said "render", not image or raster, since from my understanding, an SVG or vector PDF render of the font (not embedded, but turned into paths) wouldn't be any more copyrightable than a raster, but far easier to clone.
A font isn't just the letterforms, there's also all the metrics, spacing, kerning, and OpenType features like ligature replacement. Also modern OpenType releases contain many languages which makes the metrics even more complex. Metrics are also very refined to the point that with some fonts if they weren't copied completely there would be problems.
I suppose if you manually traced Helvetica letters, then they might have a case for misusing their font. After all, there isn't a licensing option that allows you to use the font as a template.
I'd assume the license for viewing the file is implied, but I have no clue if it would hold up in court if you viewed the file -> traced the output (which is copying the typeface, not the font.)
If you copied the .ttf file and sold it, it's a copyright violation just like any software copying. The foundry takes you to court.
If you copy by hand (at what size? at what accuracy? do you include the same hinting and ligatures?) the file will not be bit-for-bit identical. The foundry cannot sue.
Font tracing is usually done by printing out the character to be traced at very large scale -- I've seen about 12" x 12" -- and placing it directly on a large digitizing tablet. A sequence of strokes / points is collected for the outline of the character, and then curves of somewhat reduced degree are fit to those strokes / points to both reduce font data size and reduce the impact of errors, inaccuracies, and quantization in the data capture.
Even at this huge scale, and with this amount of effort, the outline of your character will be very close to -- visually identical to! -- the starting character, but not exact. As a result, the generated font program will be quite different. For example, it may use a different number of control points for equivalent curves.
Now, one can imagine automating this process differently: Take a font file, digitally render each character, perturb it a small amount, and resynthesize the strokes to generate a new, different program for a visually identical font. This is generally against the terms of service for the initial font, however, which would make it a legal matter...