Skip navigation

Typography effect

Apr 9, 2012 7:30 AM

  Latest reply: Mathias17, May 23, 2012 6:54 AM
Replies 1 2 3 Previous Next
  • Currently Being Moderated
    May 21, 2012 9:29 AM   in reply to cdeonis

    This PDF from a discussion years ago in this forum demonstrates some simple methods of using ordinary straightforward path manipulations to find the centerlines of glyphs.

     

    JET

     
    |
    Mark as:
  • Mathias17
    451 posts
    Feb 20, 2012
    Currently Being Moderated
    May 21, 2012 9:41 AM   in reply to JETalmage

    Great!

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 9:42 AM   in reply to JETalmage

    Thanks for posting that PDF. That's essentially the method I use. The visuals in the PDF will help others following this thread.

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 10:07 AM   in reply to photodrawken

    If one of your previous examples was the result of tracing 72pt. Bauhaus93 in Illustrator, then I ask again: how did you do that? Step-by-step, please...

     

    Ken,

     

    Again, you are using CS6. I am not. So I can’t give you precise values for CS6. The settings I used in CS5 are similar to those I depicted in my first post, except (mainly) that I set the Max Stroke Weight higher, because the Bauhous font you specified is heavier (i.e.; more pixels to be considered “a path” by the autotrace routine).

     

    You threw out Bauhous 93 at 72 pt. as if it were some kind of deal-breaking proof that Illustrator’s centerline autotrace was dramatically inferior to that in Draw. You did not, however, specify the raster resolution at that size. Understand: all the settings relate to counts of pixels, not scale. So even if all the settings in CS6 are identical to those in CS5, the specific values I used would only be meaningful if you and I rasterized the 72 pt. type at the same resolution.

     

    So don’t fret over specific values. Just understand what the various settings adjust. As I advised, carefully read the tooltips. They (at least in CS5 and earlier) explain what the various sliders affect. (Plus, the controls parameters are really not that much different from those in Draw.)

     

    The orange example ( Bauhous at 72 pt., as you specified) is the one that was done in Illustrator CS5:

     

    • Turn on Preview.
    • Turn on Strokes, turn off Fills. (That’s essentially putting it in “centerline mode.”)
    • Max Stroke Weight: Set to a high enough value to exceed the “widths” of the thickest character strokes (in pixels.)
    • Min Stroke Length: Start with a low value. Nudge that value upward until you minimize occurrances of unwanted short strokes (like the “facets”  it tends to draw at the blunt ends of some characters).
    • Path Fitting: Determines how far the trace is allowed to deviate from the apparent “center” of the region of pixels. Think of it as similar to “smoothing”. Lower values results in more anchorPoints.
    • Minimum Area: Set to a low value. With a raster image of solid-colored type, it shouldn’t really matter how low; it determines the minimm area of contiguous pixels that should be attempted to trace, as opposed to being considered meaningless noise. With rasterized type, there shouldn’t be any meaningless noise.
    • Corner Angle: If you see the autotrace failing to make corners where it should, lower this value. If you see it making corners where it shouldn’t, raise it.
      When satisfied that tweaking the settings is as good as it’s going to get, expand results. Select the group of paths that the autotrace generated. Set the stroke weight as desired.

     

     

     

    Again, the above is relative to CS5 and prior. I don’t know how tightly those adjustments relate to whatever rework was ostensibly done in CS6. But you should be able to figure it out. And, of course, after all that twiddling, you end up with alot of time wasted on a bunch of useless sloppy garbage that really should have just been drawn deliberately anyway—this is pretty much equally true in AI and in Draw.

     

    The simple truth is, as has been stated repeatedly by multiple correspondents in this thread, (including myself), autotracing is a very poor approach to this task of drawing uniform centerlines of type glyphs.

    Some just don’t seem to hear that, and assume that everyone who has said anything about centerline autotracing is actually recommending it for this purpose. I am not, and I stated this multiple times. But centerline tracing is useful for other things, and the repeated claim that there is no centerline tracing in LiveTrace warranted correction.

     

    It has never been unusual in this forum (and most others like it) for discussions to spur off onto related topics. Quite often, its such spin-offs that prove most valuable. Just because someone starts a thread, doesn’t mean that the thread is thereafter and forever “owned” by that person. It’s an open forum.

     

    JET

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 10:20 AM   in reply to cdeonis

    I would approach this (as a starting point at least) as monica sugested in the very first reply post.... This would also allow for the text to remain live.

     

    If you wanted cool effects (depending on what exactly) then you may have to outline the text and negatively offset path as your starting point. But this technique is very shape dependent. In type for example, to achieve the best results you would have to use a fixed width, sans-serif font.... (as in the examples you posted)

     

    If you are using a font that is not fixed width then manual is best. See 2 egs below:

     

    Fixed Width

     

    t1.png

    Same technique non fixed width font, notice the top "bar:" of the inside stroke a different thickness... and fonts that have greater variation in the glyph widths will return even worse results.

    t2.png

    non ideal fonts  even worse results.

     

    t3.png

    in cases like the one above, manual is best.

     

    and an accurate way to do this is:

     

    1. Outline the font

     

    2. Offset to negative (not as an effect) untill small as possible but that you can see all the slabs still

     

    ot1a.png

    3. Then delete points you dont need/use as a guide/make the inset path a guide whatever you prefer. Draw lines aligned etc, Offset for thicker lines if you want like so: (then you can apply this to even non ideal fonts)

     

     

    f1.png

    and then work on the effect you want:

     

    G

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 11:57 AM   in reply to Grant H

    For intricate lettering effects I don't leave typefaces active/editable. I'm not placing such lettering effects into body copy on a printed page. It usually resides in some kind of big title or logo; in that kind of graphic it's far more safe to convert the lettering to outlines in order to prevent alterations to the lettering. Someone on a different computer may not have that typeface. You can provide a copy of the typeface, but that often will violate the license of that typeface. Even if the other person has a copy of the same typeface he may not have the same exact version of it. Some typefaces have numerous version releases and perhaps releases from more than one vendor. Throw in various clones, like the Swiss 721 or Nimbus Sans clones of Helvetica. Throw in differences between Type 1, True Type and OpenType versions of the same typeface. Add to that the differences in how the typeface may be treated as the design jumps between Mac and Windows platforms. Finally, a graphic with live lettering designed in an old version of Illustrator or CorelDRAW with certain letter spacing may see that letter spacing change (sometimes radically) when opened in a later version of the same program. I've certainly seen things like text on path effects made with something like CorelDRAW 6 get totally hosed when opened in CorelDRAW X5.

     

    I can understand the need to keep type active from the perspective of certain very geeky web developers. Active type on a web page and "@FontFace" techniques will make a web page more search-able. However, it's also possible to append data to graphic images to make them search-able too. If I work hard embellishing a logo or lettering with a complex effect I'd rather have it converted into "finalized" vectors or even a raster-based image to keep it from being altered based on another user's fonts and/or software.

     

    Anyway, that's just my opinion. Others can do whatever they like. :-)

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 1:46 PM   in reply to JETalmage

    Yes, thanks for the link to the PDF.  Very helpful for the original question.

     

    Ken

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 2:05 PM   in reply to Monika Gause

    But Bob the Sign Guy has offered yet an even better and bigger issue that and a method that actually could lead to a much better blend feature as well and which also might actually bring about a way that would be rational for the engineers to build such a feature so his contribution seems to be very big indeed.

     

    Well done Bob!

     

    I mgiht submit such a feature request based on your approach but seeking a way to create an interface for it and to not  have to manully break apart the letter forms.

     

    Very cool.

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 3:03 PM   in reply to JETalmage

    JET,

     

    Thanks for the steps, but it seems like AI CS6's tracing behaviour is different from AI CS5.  Your point about the rasterize settings is a good one, and it almost allowed me to get the trace that you did.  In my previous example, I rasterized the type at 300dpi; here, I used 72dpi before tracing:

    ai trace 72ppi.png

    Still not the result you got, nor the result I get with CDR's centerline trace.

     

    The choice of Bauhaus93 at 72pt. was to limit us to using the same starting point for the discussion, not because that font is particularly important.  Still, it's a good font to use for comparing CDR and AI in terms of achieving a centerline trace.

     

    It seems you have a setting in CS5 that's not in CS6 -- the "Min Stroke Length".  The other settings might be named differently, but look to be the same in CS6.  You can see from my screenshot of the Trace Settings that I'm using the other settings as you described.

     

    Based on the results I'm getting, I still have to say that AI CS6 does not have a centerline trace capability.

     

    Ken

     

    P.S.,  I'm in complete agreement with the separate discussion point made by several folks here that a centerline trace will not give the desired effect originally posted.  Right now, I'm thinking that CDR's centerline trace is a time-saving starting point for the manual work needed to achieve the effect.

     

    Message was edited by: photodrawken to add postscript.

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 2:52 PM   in reply to Wade_Zimmerman

    Bob the Sign Guy musybe very skilled because as far as i cansee there is still a lot of work to do in order to get satisfactory results.

     

    Screen Shot 2012-05-21 at 5.47.44 PM.png

     

    I wpould say if you are not skilled you probably have to commission someone that is and I would be very impressed if someone could make this into an actual automated feature.

     

    Look at the problems with the lowercase "a" it took me about five to ten minutes to get it reasonably correct.

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 3:09 PM   in reply to Wade_Zimmerman

    @ Wade: in order to develop vector graphics filters that will properly do center-line effects on lettering (as well as other effects derived from it like 3D chiseled prismatic effects) a number of new technologies would have to be in place.

     

    The filters would need new levels of smarts to figure out what to do when the filter encounters a corner on the letter. They don't know how to do that so well. Even more difficult: how to handle areas where letter stems join -like the loop and diagonal stem on a "R" letter. The filter would need to know how to follow the curve all the way through the gap created by the diagonal stem. Right now it takes human intervention to rebuild the gap in that curve. Not to toot my own horn too loudly, but it does take some experience as well as tough subjective choices regarding areas in letter joins to make the effect look good. Typeface choice is very critical. Easier to do in all caps in a face like Gotham. Very difficult if you want to do it to mixed case lettering in something like Antique Olive. I'd stay away from serifed faces.

     

    Perhaps new font technologies could build in some of the elemental shapes of letter forms for graphics filters to use. Typefaces are built upon these kinds of shapes. The shapes wouldn't be a visible part of the letter but could give guide lines on how the letters could be embellished correctly.

     

    Unfortunately I'm not optimistic any new type of font technology would be developed to offer this sort of function. Arguably there's not a lot of money to be made in selling fonts made in a new technology due to piracy, availability of free fonts and the likelihood a new type of font technology like this would meet considerable resistance. For one thing, filter effects are typically seen as cheesy by those on the more high brow, influential side of typographical design. They would prefer letter forms be unaltered and not embellished. I can certainly agree with them in regards to distorting letters. That drives me nuts.

     

    I suppose if Illustrator had some sort of shape prediction technology that was both smart and adjustable it could make center line effects a snap to create. IMHO, we're a long way off from that.

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 3:54 PM   in reply to Bob the Sign Guy

    Actually I am not so sure either of use were correct about the possibilities of such a feature, I use to own several multiple master fonts tha allowed you to adjust te weight of the charcacters from very thin to very bold.

     

    I wonder if a combination of multiple masters fonts, blending and offset path can work with an image trace feature.

     

    Or just a way of turning text into temproary or live effect multiple master fonts.

     

    I wonder?

     

    Or of course the best feature is of course experience and skill development.

     
    |
    Mark as:
  • Currently Being Moderated
    May 21, 2012 4:21 PM   in reply to Wade_Zimmerman

    Multiple master font technology was interesting, even though it's now arguably dead. I miss the multiple master slider palette in Illustrator 7. Anyway, MM fonts were versatile but all the instances were derived from master letter outline shapes. The font technology I'm describing would take things a step further -basically embedding shape primitives into the letters to help effects filters make the right decisions. The outline letter shapes (or glyphs) that make up a typeface are the "destructive" end result of merging a lot of other smaller parts that weren't all previously cut or welded together.

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 7:37 AM   in reply to cdeonis

    Maybe the designer is trying to keep his "recipe" secret. The truth is type designers use a lot of tricks with blends, welds, cuts, etc. on open and closed paths to build up letter shapes.

     
    |
    Mark as:
  • Mathias17
    451 posts
    Feb 20, 2012
    Currently Being Moderated
    May 22, 2012 7:50 AM   in reply to cdeonis

    That's ok. After JET's and Bob's input here, I highly doubt there's anything the originator of the image's you like could possibly add to further expand our undestanding of how the effect is done. Let him keep his "secret".

     

    I don't like it when designers are tight-lipped. I share information freely. I'm not intimidated by what others might do with anything "proprietary" I come up with - I'd rather see what they do with it and learn from them, thereby improving myself, but whatever.

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 8:00 AM   in reply to cdeonis

    Just to quote cdeonis, post #41:

     

    I wrote the actual designer and will share if she responds.

     
    |
    Mark as:
  • Mathias17
    451 posts
    Feb 20, 2012
    Currently Being Moderated
    May 22, 2012 8:10 AM   in reply to Jacob Bugge

    Hey, Jacob.

    Just to be clear, I was referring to the maker of the OP's posted image, not cdeonis.

     

    This thread has been awesome. I've ignored the power of blends for way too long!

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 9:04 AM   in reply to Mathias17

    I know, Mathias.

     

    It was just a comment to the use of a masculine pronoun, by you and Bob.

     
    |
    Mark as:
  • Mathias17
    451 posts
    Feb 20, 2012
    Currently Being Moderated
    May 22, 2012 12:10 PM   in reply to Jacob Bugge

    Oh haha, right.

     

     

     

    *EDIT     Got an annoying newlsetter email from photog David Jay just now, and it had this applicable quote in it:

     

    "Talents and gifts do not reduce or diminish when shared; they expand and increase like the widening ripples from a pebble dropped in still water."- Kevin Hall

     

     

    Yep.

     

    Message was magically edited by: Mathias17

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 1:30 PM   in reply to Bob the Sign Guy

    Bob the Sign Guy wrote:

     

    @ Wade: in order to develop vector graphics filters that will properly do center-line effects on lettering (as well as other effects derived from it like 3D chiseled prismatic effects) a number of new technologies would have to be in place.

     

    The filters would need new levels of smarts to figure out what to do when the filter encounters a corner on the letter. They don't know how to do that so well. Even more difficult: how to handle areas where letter stems join -like the loop and diagonal stem on a "R" letter. The filter would need to know how to follow the curve all the way through the gap created by the diagonal stem. Right now it takes human intervention to rebuild the gap in that curve. Not to toot my own horn too loudly, but it does take some experience as well as tough subjective choices regarding areas in letter joins to make the effect look good. Typeface choice is very critical. Easier to do in all caps in a face like Gotham. Very difficult if you want to do it to mixed case lettering in something like Antique Olive. I'd stay away from serifed faces.

     

    Perhaps new font technologies could build in some of the elemental shapes of letter forms for graphics filters to use. Typefaces are built upon these kinds of shapes. The shapes wouldn't be a visible part of the letter but could give guide lines on how the letters could be embellished correctly.

     

    Unfortunately I'm not optimistic any new type of font technology would be developed to offer this sort of function. Arguably there's not a lot of money to be made in selling fonts made in a new technology due to piracy, availability of free fonts and the likelihood a new type of font technology like this would meet considerable resistance. For one thing, filter effects are typically seen as cheesy by those on the more high brow, influential side of typographical design. They would prefer letter forms be unaltered and not embellished. I can certainly agree with them in regards to distorting letters. That drives me nuts.

     

    I suppose if Illustrator had some sort of shape prediction technology that was both smart and adjustable it could make center line effects a snap to create. IMHO, we're a long way off from that.

    They don't need to create any new methods in order to develop more useful tool. The method already exists, the blend shape technology can find the center between two lines. The only thing they need to do is create a tool with user defined input about the choice of segments on the letter path, because this as you said can not be automated. So basically a center line tool will be a dedicated blend tool that can work like this: Initially click with it to define 4 points - the start and end points of two corresponding segments from the  letter and this should create a new path as the center line between the two segments. After that to continue the creation of the center line, click to define 2 new point  which will extend the previously defined segments in  the desired direction. For splits like the letters R and B, clicking on the end of the center line being currently created and then define two new points on each side of the letter path should create the split and the direction.

     

    Doesn't seem a big deal to me to implement a tool like that. This will reduce a lot of work, and avoid things like releasing compound paths, splitting, adding anchor points and all that drag that consumes most of the time.

    I'm sure all this is perfectly possible and is a matter of decision and not at all anything like impossible challenge. It is a typicall problem perfect for software solution.

     

    just my 2 cents

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 1:40 PM   in reply to Bob the Sign Guy

    Bob the Sign Guy wrote:

     

    Maybe the designer is trying to keep his "recipe" secret. ..

    What secret? She said:

    I begin with the boldest weight of the font and the thinnest weight and just play with the letterforms in illustrator

    play in this context means a lot of manual work and pain. You can see from the image in the first post of this thread that the center line is rather poor. Her explanation to me, has equal lever of secrecy, if any, as the explanation you gave about your use of blends

     
    |
    Mark as:
  • Mathias17
    451 posts
    Feb 20, 2012
    Currently Being Moderated
    May 22, 2012 1:47 PM   in reply to emil emil

    Well, whatever the case, the methods behind creating a proper "centerline effect" have been very sufficiently exposed to the masses. Thanks, all who contributed.

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 1:48 PM   in reply to emil emil

    Emil, the solution is not so simple. While computers are very precise at doing certain things, they're incredibly dumb when it comes to subjective decision making.

     

    This is essentially why it's necessary to break closed paths of letters into individual open path components to do the blending. When the pairs of open paths are too long and/or contain unequal numbers of anchor points the blend effect can lose its way. It doesn't have the subjective, asethetic decision making power to get the new center line path centered in certain situations.

     

    As to a function that would simply span the gap between blends at letter joins, any automatic tool would fail at this. When I created my "BIRD" example (on page 1 of this thread) I had to carefully create some longer curved source paths to span the gaps at the joins in the "R" and "B" in order to create new paths that followed the loops of the "R" and "B" correctly. Any tool that would try to automatically bridge the gap would fail miserably because the tool wouldn't understand how to create the elemental shapes used to create the letters that were no longer there in the finished letters.

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 1:50 PM   in reply to emil emil

    I actually think Bob gave a pretty clear explanation of what he was doing he broke apart certain sections of the outline letterform to make several sections which could be blended and then joined up those areas that could be joined and then he made modification of position or suc as need if needed and often it dod not need to be alter and some letter forms lend them selves to this blend.

     

    it was clear to me. It takes experience to knwo how this is going to playout and how to do it witht he minimum of alteration but that is what it is all about.

     
    |
    Mark as:
  • Mathias17
    451 posts
    Feb 20, 2012
    Currently Being Moderated
    May 22, 2012 1:53 PM   in reply to Wade_Zimmerman

    Of course he was. The "secrecy" claim was directed at the originator of the OP image. You guys have all been great. Let's go make something now.

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 3:46 PM   in reply to Bob the Sign Guy

    Bob the Sign Guy wrote:

     

    Emil, the solution is not so simple. While computers are very precise at doing certain things, they're incredibly dumb when it comes to subjective decision making.

    Yes, of course,  that's why I suggested a feature tool that works with user defined input. My suggestions are based on my experience with using bends. When a blend doesn't work as expected adding points on a proper places of the paths gives the desired result, I don't see how creating a brand new path can make a difference. The tool I suggested can make the segments as short as necessary to define the desired result. The open path needed and etc, that you are concerned with should happen behind the scenes and in the front end it should remain as one path, similar to the Width tool which behind the scene is creating a filled path identical to the path when expanded, but in the front end it is just one stroke. I'm not suggesting some magical decision maker tool, what I'm suggesting is automation of all the tedious work you are doing that doesn't require decision making like disassembling of the paths into pieces and then assembling it together. The decision making is defining the segments and points needed which you still have to do but with less effort.

     
    |
    Mark as:
  • Mathias17
    451 posts
    Feb 20, 2012
    Currently Being Moderated
    May 22, 2012 4:08 PM   in reply to emil emil

    K, now tell the dev's all this!

     

    (as in . . . suggest it so there's a hope of it being in a future release . . .)

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 4:52 PM   in reply to emil emil

    emil emil wrote:

     

    I don't see how creating a brand new path can make a difference. The tool I suggested can make the segments as short as necessary to define the desired result. The open path needed and etc, that you are concerned with should happen behind the scenes and in the front end it should remain as one path, similar to the Width tool which behind the scene is creating a filled path identical to the path when expanded, but in the front end it is just one stroke. I'm not suggesting some magical decision maker tool, what I'm suggesting is automation of all the tedious work you are doing that doesn't require decision making like disassembling of the paths into pieces and then assembling it together.

    The problem is it is impossible to perform the blends I did without both breaking apart the path into individual segments and creating some new parts that were not originally part of the letter.

     

    BirdThing2.jpg

    Notice the curved black path segments. I had to create these new paths as a blend source target in order to make the lines inside the letters follow the curve more naturally. The new black paths extend the natural curve of the outer loops well into territory consumed by the join in the original letter form. Computer filters up to this point have never been able to do this because some subjective human choices are involved in making those new path segments. The "B" was particularly complicated because I had to create the center line first and then create 3 step blends between it and the inner and outer loop segments to get all 7 lines established.

     

    If a computer filter, even one that involves human input, could do this sort of thing without breaking apart the letter form I would be really amazed.

     
    |
    Mark as:
  • Currently Being Moderated
    May 22, 2012 6:31 PM   in reply to Bob the Sign Guy

    Ok here's how a tool could work and I it should be perfectly possible to make with proper math calculations.

    the left R on the first line shows the current problem and on the right side shows what a dedicated blend tool should be able to do.

    On the second line are the 4 steps and the number of clicks and their order in the red circles. Behind the scenes each click is cutting the path and performing a blend on the defined segments and additional end points if specified.

    In step 2 the 5th click defines the end point which causes to end and the path at that point with the appropriate smoothing applied. In step 3 behind the scenes 2 path segments are cut and joint for the blend operation. In step 5 this is basically the same as step 2 but I deliberately broke the continuation of the line showing that the user can also click arbitrarily on any point of a path and not only on its end point and since the line is all corner points no smoothing is applied. I'm not a programmer but having used a lot of different programs I can tell that this shouldn't be the biggest software challenge to implement.

     

    Capture.JPG

     
    |
    Mark as:
  • Mathias17
    451 posts
    Feb 20, 2012
    Currently Being Moderated
    May 23, 2012 6:54 AM   in reply to cdeonis

    Yeah, I know what it's like to be constantly emailed questions by admiring fans. It's tough to answer everyone well.

     

     

     

     

    ok . . . nobody emails me actually . . .

    http://i3.kym-cdn.com/entries/icons/original/000/003/619/Untitled-1.jpg

     
    |
    Mark as:
1 2 3 Previous Next
Actions

More Like This

  • Retrieving data ...

Bookmarked By (1)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points