Two more questions about nodes.

Goran Soderstrom's picture

I must be on top of the list now, regarding asking questions about FL, but all you guys are so great in helping me along the way so I go on ;-)

This time I have two questions:

1. Start Points.
If you look at this image, this particurlary letter contains only smooth connection nodes. However, the need for start points automatically converts them to corner nodes instead. Is this OK? If not, how to achieve the letter?

2. Extreme points.
By looking at this image where the red arrows are, do I have to put points there? What are the recommendations? I would prefer _not_ to put nodes there to make the curve smoother. An insert of an nod there will attach to the "grid" and mess upp the curve a little bit (shift it to the nearest grid that is) – and make the curve a little bit distorted.

Thanks in advance.

hrant's picture

> do I have to put points there?

If you want to be safe (like with some RIPs) and/or you need to hint robustly, yes.

> ... to make the curve smoother.

If you mean to avoid grid coarseness, this is a classic conundrum (resulting from an em of 1000 often being too coarse, especially for italics). One trick (that I learned from Gary Munch) is to slightly move the tangent point below to get a smooth "total".

hhp

Rob O. Font's picture

"do I have to put points there?"

If you are planning on using this primarily in high resolution environments, no.
In fact, it's probably better if you leave it as it is.

Goran Soderstrom's picture

Thanks for those two different answers ;-)

How about these start points? The funny think is that if I make the start points into smooth tangents — they anyway get converted into corner points the next time I open the font, by them selves.

It seems that a start point can’t be a tangent. (?)

twardoch's picture

Goran,

> The funny think is that if I make the start points
> into smooth tangents — they anyway get converted
> into corner points the next time I open the font,
> by them selves.

Obviously, this is a bug. Thank you for reporting it here, I have notified our developers.

A.

Tim Ahrens's picture

[edit]

dezcom's picture

Adam,
That conversion seems to happen most often when an outline is copied and pasted or scaled if that is of any help.

ChrisL

crossgrove's picture

Other point types also change when you add a component and then decompose it. I can't think of a reason for this, and I hate it.

paul d hunt's picture

this type of point scheme is foreign to me (and most likely any FL users who didn't start with Fontographer). I don't ever get any problems like this using the Options>Glyph Window>Appearance>unchecked "node shape shows point and connection type" configuration. why would anyone prefer the old FOG point scheme? I don't get it.

crossgrove's picture

Because apparently it's beyond crucial to interpolation what the point type is. More so than in FOG.

Goran Soderstrom's picture

Other point types also change when you add a component and then decompose it. I can’t think of a reason for this, and I hate it.

I’ve noticed that tangents sometimes convert themselves to "corner nodes" if they are not really smooth when decomposing OR if I merge countours. That is, if I have moved another nod or enlarged the glyph or something — and not double clicked on the tangent to correct the angle of the handle. For me it seems to follow this pattern anyway.

I don’t ever get any problems like this using the Options>Glyph Window>Appearance>unchecked “node shape shows point and connection type” configuration. why would anyone prefer the old FOG point scheme? I don’t get it.

Ah, I turned it off now to see what happened, and suddenly there was only two types of nodes. A smooth (with or without a handle on both sides) and a straight. Didn’t know of that one. Now I just have to figure out what is best for me to work with. Are there any big differences in the behaviours with these two types of point handling compared to the tangent, or is the circle just replacing the tangent?

Something new to learn, everyday — I like it!

Because apparently it’s beyond crucial to interpolation what the point type is. More so than in FOG.

Can you explain more why this is so? Since you both are two established type designer with two different opinions, I would very much like to listen your thoughts on this.

twardoch's picture

> Other point types also change when
> you add a component and then decompose it.

As always, we appreciate specific sample files that illustrate this kind of behavior and can help us replicate it on a particular example. Please send them to adam at fontlab dot com, with a step-by-step description on how to replicate the problem.

Regards,
Adam

Rob O. Font's picture

1. Start Points.
a. FL handles them "specially", i.e. You cannot edit them exactly the same as all other points, so you frequently have to move the start point just to be able to correctly move a "curve point" along a "curve", e.g. The Start point is needed by the format but it should be totally and completely transparent to the user in the editing process, (ala Fontographer), and the point type the designer defines should be an inviolate choice though the cycle of storage, but in don't work that way, and so far, making the point has not had an effect.
b. whatever niceties are brought by having "curve" "tangent" and "corner" points are controlled by the user interface and database of FL, (something they refuse to add to their TT editing it seems, so all points can only behave as "corners" ;-o\\\\\ ) These "point types" do not exist in the PS or TT format, only the start point does.

I think I've got that right, but maybe I'll be proven wrong, soon.

twardoch's picture

> (something they refuse to add to their TT editing
> it seems, so all points can only behave as “corners”
> ;-o\\\ )

David,

I'm sorry to realize that your memory seems to be rather flaky, sometimes. Only a few weeks ago, you and I corresponded about the TrueType points directly, and based on this, we had an internal discussion at Fontlab Ltd. I told you that you will likely see some improvements in TrueType point handling in the next release of FontLab Studio. This is actually the opposite of "refusing", at least where I live.

I agree that the startpoint issue is a very problematic one. There have been some improvements along the way but still not sufficient. I have reported the "losing of tangent type" problem to our developers. Please direct other specific problems related to startpoints directly to me (adam at fontlab dot com), ideally with samples illustrating the problem. I will try to exercise some pressure on Yuri about this -- well-documented test cases showing what exactly is wrong are always helpful in such situations.

Regards,
Adam Twardoch
Fontlab Ltd.

Rob O. Font's picture

Dear Adam,

I began my sentence on this forum with a list of FL "bugs", a year or two ago? Start point was there I believe, and I was told it was fixed when I purchased 5. I have not complained that it still sucks time and effort from my work. I'm just wondering how patient you'd be if your Mac came with a restart button on the bottom of the mouse or your car came with an ignition switch located in the Congo?

As for the TT situation: This is the first I've heard that you received my "HELP QUICK!" mail and illustration from...July 24th. If that is the memory problem you're referring to "sorry, I forgot"! There are BRAVE FONTS AND DESIGNERS DYING OUT THERE — Anti-aliasing is spreading under the black wing of totalitarian typographic ignorance, and independent designers should be able to fix it before the public pukes, (subconsciously of course, both fix and puke).

Just exactly how many useless functions should we be rooting for before the fundamental ones are available? I don't know, I don't care and I've lost interest in making these points opaquely and without alienation. If this is not better done in the transparent light of day, so others might layer their opinions on to mine, yours or whoever's, sorry. How exactly is it that you think FL got into a situation where numerous basic functions are broken? Exactly! Listening in dark corners to bad ideas from typographically incomplete minds.

Time's up, there are clients at the door, and the hinges are not installed.

Kindest regards,

David Berlow
The Font Bureau, Inc.

hrant's picture

> Anti-aliasing is spreading under the black
> wing of totalitarian typographic ignorance

Automatic anti-aliasing, not the good kind. :-)

hhp

twardoch's picture

David,

> a list of FL “bugs”, a year or two ago? Start point
> was there I believe, and I was told it was fixed when
> I purchased 5.

I know that the startpoint behavior was improved in 5 vs. 4.6, but there are still some problems with it. However, the roughest problems like doubling outline points upon outline conversion were fixed. Our main engineer Yuri promised some further improvement in the next release of FontLab Studio.

> I’m just wondering how patient you’d be if
> your Mac came with a restart button on the
> bottom of the mouse or your car came with
> an ignition switch located in the Congo?

Well, actually, my PowerBook came with a power button that, when the computer was put to sleep, does not work every 4th or 5th time, requiring me to take the battery out.

> As for the TT situation: This is the first
> I’ve heard that you received my“HELP QUICK!”mail

There are some situations where there is no immediate solution to a problem. However, I have forwarded all your TrueType-related comments to our developers and they promised some improvement in the next version. This is all I can say -- for sure I cannot agree this is "refusal", though it most certainly not always is immediate relief.

> How exactly is it that you think FL got into
> a situation where numerous basic functions
> are broken? Exactly! Listening in dark corners
> to bad ideas from typographically incomplete minds.

We listen to those who dare to speak. Where have you been in, say, 1998 or 2000? (Probably still trying to give Macromedia Fontographer a mouth-to-mouth.)

This was when my typographically incomplete mind was infusing my bad ideas into the FontLab creators' minds, and they were listening. I think I am responsible for dozens of potentially useless (though I do think otherwise) functions in FontLab Studio, so is Luc(as) de Groot, John Hudson, Gary Munch and countless other typographic ignorants who have kept a dialogue with "the Russkis" ever since the inception of FontLab. But it is not too late to join in :) You're most cordially invited!

Best,
Adam

Rob O. Font's picture

> Our main engineer Yuri promised some further improvement in the next release
> of FontLab Studio.
Just make it like any other point. And thank him for getting rid of the double point nonsense, that was really not helpful.

>We listen to those who dare to speak. Where have you been in, say, 1998 or 2000?
I, unlike others, have spoken out on This Tool, since I started using it in 04, and never before that. I know of no one who is adept with both tools who disagrees with me, I being one of the few trying to do so. If you want proof that Fog is a superior drawing tool, look at how far FL has come towards it since 00, when Yuri told me they were just going to copy it's UI.

>(Probably still trying to give Macromedia Fontographer a mouth-to-mouth.)
:), and now that you can be eye-to-source with Fontographer, why don't you just get them to finish copying it.

> I think I am responsible for dozens of potentially useless functions in FontLab Studio,
Did someone ask for start point psychosis then?

And thanks for the invitation to join in, but there seems no proper triage to get the basic drawing functions up to speed, (and keep them there) for who, I guess you'd call, Typographic Completes, (including teaching). In the end, it reads PS and TT fonts at least, so I don't have to worry, it just makes ya wonder for starters, how do they get it...?

Nick Shinn's picture

On a completely unrelated note, I just copied a tif file in Photoshop and pasted it into the mask layer of Fontlab. Brilliant. [edit: actually, I used to do that in Fog. I'm getting confused. The worst is when you bang out some keyboard shortcut from Fog that doesn't work in Flab, and banging it out again doesn't undo it, and you have to go figure what it was you made happen.)

I have gotten used to the FontLab drawing tools after a couple years; the thing that is preferable about them, which is directly related to the question at the head of this thread, is that the points always snap to the one unit grid, which makes it possible to insert a node on the curve, and then tweak the adjacent paths as per Mr Munch, at a much higher magnification than Fontographer, to get the curve nice, and anti-aliased too. So I would score those three factors -- snap to one-unit grid, greater magnification, and smooth curve -- as an improvement over Fontographer. Although, of course, its drawing tools were simpler and more intuitive. Such is progress, slick and tricky.

Rob O. Font's picture

"a much higher magnification than Fontographer" Really?

The magnification of Fog went to the max, that being 1:1 with the UPM of the font. You can't go any higher than that Nick. If you think so, I've got a bridge in Brooklyn for ya. If you need more magnification, you can make the UPM 2048 instead of 1000, but otherwise fog 'n FL are the same.

Tim Ahrens's picture

I always found it hard to understand that people preferred the Fontographer interface to FL, already with FL 3.0.

There is only one exception: The node deletion in Fontographer is much better. And integer coordinates are not an excuse. It would be great if node deletion could be improved, you would not believe how useful really good node deletion is.

Changing the UPM for zooming purposes is obviously a ridiculous suggestion. I wonder what you would say if this was necessary in FL?

Rob O. Font's picture

"There is only one exception:"
Really? Someday, I'll sit down with you and design type and be done with the S, while you'll still be fumbling around in the flea hair tolerances of FL to close the contour of the K.

"Changing the UPM for zooming purposes is obviously a ridiculous suggestion"
...said by someone who's never designed an ultra black extra compressed, I'll bet.
I would say the same thing about FL, because it is precisely the same truth. All digital design programs come with a snap to grid at the base resolution of the data. It is only when you need to go into tiny detail that you know what effect that can have on your design, (assuming the product of the effort is being used at high resolution of course).

twardoch's picture

David,

the screen zoom needs to be higher than what Fontographer gives you because these days, pixels got tinier on the screen. The zoom in FontLab Studio is not infinite but is higher than in FOG.

One very effective way to show why one tool is better over another, especially regarding the user *experience*, is record-and-tell.

One of the following tools:
http://www.ambrosiasw.com/utilities/snapzprox/
http://www.miensoftware.com/screenrecord.html
http://shinywhitebox.com/home/home.html
should give you a kickstart ;)

Really, give it a try. It would be funny, at least :>

A.

Rob O. Font's picture

I'll surely get back to those links as soon as I can ;)

The tool's zoom has nothing to do for you after the units per em are overwhelmed by screen resolution. This doesn't mean that you can't look at it zoomed to where 1 upm = 10 screen pixels, but there turns out to be little reason as this enlargement gives you little true perspective on even the smallest curves and then the screen movement of a point by a single pixel is too huge to see the effct of, I think. I'll leave it there but to say there'd be an interesting comparison with lens crafting for punch cutting. I think enlargement ends with the molecule, in the best possible world, and so does the tool, but if you think it ends with the Adam...;

I wonder how many people have designed or edited a significant number fonts on upms of 2048, or 256, e.g.? 16? How about 8? :-)

The other point I made goes to the fact that I want to make screen fonts. Since the most optimization enabled multiple-size-from-a-single-master format for screen fonts is TT, then that is what I must make. I don't have 7 years and an unlimited budget to make 7 families by drawing Beziers at 1000 upm, using FL's conversion (2.048 x input upm?) and then manually or semi-automatically convert to an optimized TT contour, which is a specialized waste of time that all the CT Collection probably went through without the knowledge of those designers. So I'm just asking, can we get Back To A Tool that allows us to draw in either format and save the all the same point types or not?

Nick Shinn's picture

Resolution is one thing David, but I think what I appreciate most about the larger zoom of FontLab is the room it gives me to co-ordinate what I'm seeing on the screen with my hand movements on the tablet. It's working at a larger scale, which is more comfortable and I would hazard an explanation -- because it requires less concentration, and is therefore less stressful.

Actually, this zoom is very necessary, with all the tablets that are crammed onto the FL interface, so the other HUGE improvement in my working condition in recent years has been moving from a convex CRT (which distorted glyphs if you didn't view them bang in the middle of the screen) to a flat "cinema" screen with way more real estate.

hrant's picture

> I wonder how many people have designed or edited a significant
> number fonts on upms of 2048, or 256, e.g.? 16? How about 8?

Maybe what you're getting at is not opposed to these, but:
1) TT fonts have an ideal/official UPM of 2048.
2) AFAIK, many (most?) CJK fonts use smaller UPMs, I think 256.
3) Hmmm, now I'm thinking really small UPMs might be perfect for pixelfonts! :-)

> the most optimization enabled multiple-size-from-a-single-master
> format for screen fonts is TT

Also only TT, but I don't think we should forget the option of embedded bitmaps, which can be of many sizes in a single font file. The big question is, do embedded bitmaps (and the tools to embed them) properly support Unicode, and grayscale?

hhp

twardoch's picture

I've made some pixelfonts where the UPM size was equal to the ppem size, i.e. each virtual pixel was just one font unit large.

A.

Goran Soderstrom's picture

Any news when FontLab will fix the bug with startpoints?

Syndicate content Syndicate content