A bunch of changes here before I head out, covering most of what people mentioned. Another spacing error crept in as well as the recurring PDF problem. PDF 10 on top.
The usual UPMs are 1000 (usual) & 2048 (the microsoft standard). You can use something else but you might not want to. Did you mis-type?
It seems like the whole thing is starting to tighten up. Would you post a PDF with some text so I can what's happening better? Also what about posting an example of how you imagine the font being used. That would helpful too.
The Fontographer manual suggests a multiple of 1024 fo the em square for TrueType, so that's what I set it at. Naturally 2048 will double the pitch of the grid, which would be handy.
I'll give a shot at some layouts. I'm not really sure what a good use for this is right now. It calls attentiont itself a bit too much for body text, and is a bit bold for that anyway. I might come up with a lighter weight version once this is nailed down and see how it works. Maybe with slightly looser spacing it work for advertising copy. Would you buy a luxury car from this font? Eh... maybe not.
I think I overdid the Q—rather than going with a loop instead of a reverse on the tail, I did both. Is the result too heavy/busy for such an infrequently used letter?
I took a cue from the curve of the k to reshape the h, m, and n. Is it an improvement?
I think the hmn are improved. I don't know that it completely matches the face yet exactly. But you should look at your b I have it match the new letters a bit more. Or make the others match the b. Or something else but getting them all on the same path. Speaking of which ... Maybe think about that p too. It's an odd man out. All of this is a big deal beacause you will gain in legibility going some ways but I know you wanteded your own solution. I would keep trying things.
Similarly I think you can make a case for keeping the g & q with the structure they have. I think you might want to bring the d in line with them to some greater degree. the g is looking abit too closed right now. Almost like an odd 8. I like the opening on the y - even though it's got the same structure the opical difference is huge. Optics are what count.
RE; the cap Q I think you can make it work ( maybe) by adding weight to the bottom of the curve before it loops. You might have to drop the tail slightly to do that. But I bet that would work. I like the Q. I want it to work.
The top of the 3 feels off balance - maybe it's too heavy. The 8 feels denser & darker that the other #s. Maybe open it up a bit.
The "n-ary product" symbol is the uppercase pi. I note that you changed the W but not the w. I'm not entirely sure the dot in the zero works. The Q does seem a bit extravagant. Also, the tail of g doesn't look good--try squaring it out some. The point inside the k is too sharp.
That's what I thought you were talking about. Fog only gives the character name as "product". This uppercase pi does not come below the baseline, which is why your comment confused me. Are you saying it is supposed to?
The g hasn't changed since the v.6 beta. How does it attract attention now that it has been stable through 4 iterations?
The thing is when you have bigger problems with a font they are more obvious and get pointed out first. Also, people don't have time to look for 1000 things to fix - just 10 or so. And how would you feel if you got a list of 1000 problems? Overwhelmed or irritated or depressed I expect. You must do whatever you want with your font. But I think Nathanael Bonnell ( no.3) is probably right.
I posted a Q&D page layout showing the font at various sizes. It's not very good as a layout, but it gives an idea of what it looks like in use.
Yeah, I was saying it's supposed to.
Looking at the font actually used to set text, it looks pretty good now. A good quirky feel. I maintain that the Q is too much. I think that (along with the Q) your most distracting characters right now are g,u, and maybe a. The u should probably have a different serif scheme--no other letters have a similar one. I'm thinking top serifs perhaps like Weiss, or maybe just as though taken from a rotated Agamemnon n. And the g will probably be okay once you square it out like I suggested. Oh, also x is a bit wide. Next sample post a pangram or two so we can see the whole thing in action!
I realize ∏ in a Latin font is meant as a mathematical symbol and not for writing Greek, but looking at the wikipedia page Eben linked, ∑ is set the same way in equations. Should that be similarly oversized too?
Looking at these characters in several other fonts, I can find no consistency to this rule. Hoefler Text has both at standard cap height, in Helvetica they both drop their bottoms to the decender height, in Chicago they both go below the baseline, but not that deep, and in Courier they extend both above the cap height and below to decender level. If there is one correct way, then three of these fonts must be wrong!
I don't think that's the case, It's just understandably confusing.
Pangram sheet now in the PDF collection.
Yes, I forgot entirely about the summation symbol (the Sigma); that one also descends. Wow, we must be losers, spending this much time quibbling about whether math symbols descend below the baseline. (But for the record they do.) By the way, check out the stress scheme in other fonts' summation sybols. That's the generally accepted way to do it, though you could probably get away with the current one.
In what way is the d not in line with the g and q? When I started, at least, the a, b, d, p, and q were all derived from each other through rotating, flipping, removing the stem, &c, as was the g with the addition of the hook decender. in fact, the h, n, and m were in this family too, just with their bases opened. I went with the 2 story a because the earlier æ wasn't sufficiently distinct from the œ. Apart from that, are you suggesting I revert to the original shapes, or that I replicate the curve on the n to those letters that don't have it yet?
I thought the blended serif-curve ripple was an identifying characteristic of this font. Is that worth eliminating? It seems that would be removing a cohesive relationship rather than adding one.
Damn, that sounds plaintive. I'm just trying to understand the thoughts you're contributing. Most likely it's a matter of my lack of experience and limited vocabulary that I'm just not getting it. I truly value and welcome your advice, and I hope to avoid any misunderstanding.
I may be trying to hold you to a higher or simply different standard than you are interested in. Which may well be unfair. But, I think this face might actually sell with more of the kinks worked out. (You should get other folks opinions on that)
But that makes me think: what sells more? Display or text? 'Text' is answer. And the closer to a text face a display face is ( & this is a display face) so that you can use it for pull quotes & stuff the better it will sell. So the suggestions I have been making have been geared towards making the face a bit more suitable for text. If thats not your thing- no problem.
So. Assuming you could be interested in the text thing: About the b h n : One of the features that help identify a b h or n is the notch where the leg joins. If you look at the face FF dax you see an alternative solution. But conventionally speaking, and your face has some of it's roots in convention, there is usually a notch.
About rotating & mirroring: Conventionally the q should not mirror the p. The d aught not to mirror the b. Sure they do in some faces like DIN. But it is a 20th century affectation having to do with modernist reductivism. I don't see that focus in your type. There are many other alternatives but first let's look at the traditional one:
These are from Sansa Slab (Ourtype). See how the forms differ? Look at d the q! I see the d & q as a more vaiable place for your wave. Maybe the r too. Maybe not though. I would maybe make it more narrow. Maybe a different kind of wave aught be available for the b h & n. So I think you should retain your ripple but think about ways to integrate & vary it with the greater diversity of forms found in text faces. Applying 'rules' can backfire if it interferes with the distictiveness of a letter.
BTW - You are right to be concerned with cohesion, but there are many ways of achieving that. The wave was not enough. There is stem weight, optical weight matching, spacing repeated themes in the ornament, and the ends of strokes.
If you can clarify what you want from the face the choices are easier. For instance: if you plan on the face being used at 36pt and larger then your g with it's narrow opening may be okay. Looking at large text it doesn't bother me at all. But if you want people to be able to think of using this font at 12, 14, 16pt etc; then I would definitely open that thing up right away! Context helps. Keep thinking about your intended use & the choices you make will be easier & smarter. 100% of these suggestions are refutable in the right context.
It would be great if this could work as a text face. As it is, it's a bit too bold and tightly spaced, so I'll have to loosen it up and make a lighter weight version once it is fully nailed down. (To say nothing of corresponding italics. I've never done those before.)
And, I have opened up the g in the next iteration quite a bit. I also increased the em square to 2048 (surprisingly painless).I'm still poking around for other things to tweak. I should be ready to show those developments later today.
In the mean time, here is a sample to compare with your example of Sansa Slab, plus a few extra letters:
In version 11ß, the bowl of the b will have a similar top to the archof the h,n, and m. but as I understand it, you're suggesting that follow on to the p as well, and also be reflected on the underside of the d, q, and maybe the g.
The wave I speak of is most clearly evident on the top of the r, but it is present on the top of the p, q, and g as the curve of the bowl flows into the serif, and similarly on the bases of the b and d. To apply your suggestion (in the manner that I'm understanding it, which could be wrong) would wind up putting notches in the d and p that aren't there now. Am I getting this right?
You almost have me. What I am saying is that the glyphs which have bowl or arms that emerge from the left to right ( nmb&p & maybe r too ) will all benefit from being notched as in the Sansa example. Whereas the qd & g all seem to take to the wave shape ( on the top ) more naturally. Looking at the b I see is has the wave on the bottom. So there is your connection still in place.
Assuming you agree that this is worth trying out I should add that the n & m do have that 1st serif that make them semi cohesive already but the way the arms are coming out don't look quite right to me with the rest of the design. What if you used something more like the b's bowl top join. That feels more relaxed.
As for the r I would either give it a b like arm connection or alter the wave for that glyph. I realize that it was the concept glyph but I think it could be better. You could raise the height of the r's arm & ball too. And or you could try to make the glyph less wide. It's creating an odd break in flow as it is. The a might be slightly less wide too.
It took me a little longer than I expected but there it is, version 11, including a lot of the changes we've talked about. I do not like what's become of the r, and I'll likely revert back to the previous version.
Here is that short sample I posted earlier done up in the new version:
New PDFs are up on top too.
I like your first 'r' better also. Although the ball terminal works in itself, it's too slanted and sticks above the x-hight too far. And the serif is too distict. But you already saw that. ;)
The decrease in size of the loop in the Q opens the shape up, good thing. The rest of the overall little changes are improvements, I think.
I think some of this is exactly right. The b p & h now have a notch but they also look utterly natural. They fit right in. About the lc r. I agree it isn't working. Oddly enogh it looks you took both of the ideas I was suggesting for the r at once! I was thinking high or less wide. It's actually incredible cool to see both just from an interest point of view and I think you did a great job but I think one or the other is right way - especially now. I would try less wide and model it partly on the n. The g looks much more readable at small sizes. There rae still some letter relationship things left - like the am&b all seem to be thinking about different x heights. Do you see what I mean? I think I like the a as the model for ideal height the best. What do you think?
The a is a bit taller than the rest of the lc, a side effect of swithching to the two-story design. but looking at it that is probably a better height to aim for with the rest of the letters. It will mean making adjustments to almost half the characters, but it should be worth it.
Jelmar, I think you could be seeing an illusion in the Q. The loop is larger, not smaller, than in the previous version, but there is an overlap in the loop that got rendered even-odd by Canvas exporting to PNG and needs to be made solid. It should render correctly in the PDF samples.
I was looking at the PDF of pangrams and it struck me that when you are done with the next round of changes it might be most productive to start altering spacing and then maybe kerning.
I have been wondering what effect the rolling shapes would have when set into text. If for instance they would produce an unpleasant unevenness or if they would be colorful and attract the eye. Of course it's a little preliminary to say, but I think it's the later & better of the two. Nice. I especially like the z. It's really doing it's job.
Ohh, yes I see. I was mistaken, I switched them around.
Well I guess I like the previous Q better then. Although you did reduce the weight of the loop, which also works.
I think now might be a good time to look at how far this has come along. Remember, this is something I started a several years ago before picking it up again last October. Here is pretty much what I started with (corresponding with the agamemnon 4 PDF):
And here is where I'm at as of yesterday (v.11ß):
What do you think of how this is evolving so far?
Come to think of it, I should try that same layout with an even earlier version. I can't think of why I didn't.
Well here it is, derived from the oldest backup I could find, created 3/17/01, last modified 9/1/03.
Yes, it's gotten much better. What's next? Will keep on with it - or put it away for a while?
I have another update to show. There are just a few kinks to kick out of it.
I solved one missing glyph problem (seems inherent to Canvas 8) but for some reason the lc y acute and multiply won't generate in the TrueType font, and I'm having loads of problems when I output to PS1 format. A whole bunch of glyphs go missing and the space comes out four times too wide.
I might have to boot into OS9 and see if that solves the problem, though I dread having to do that.
PDFs seem to generate just fine from the print dialog in TextEdit, so I'll be using that for my character sample. I'll have to see if I have anything else that will do as well and allow a 2 cloumn layout for a text test.
Well, the above issues remain unresolved, but here is revision 12:
I normalized the x height overshoot of rounded lc to match the a, redid the r again, along with æ, œ, more subtle tweaks to move some serifs that weren't lining up across the board.
Loosened up the spacing a bit too and I think it's good, but the kerning pairs still need work. Any flaws you can cal my attention to would be appreciated.
The new PDFs are multipage documents. You may want to only print the first one or two of each.
Maybe there is missing info in the glyph itself, or wrong info. I ended up pasting my outline data into a whole new font 'holder' to get rid of a corruption problem. I don't know that any of this is what will help you - but it might. The other thing is you might want to search the Typo-L archives and typophile archives for similar problems that may have been solved already or create new threads on typophile with one specific problem per thread. There is also a MSN group for the company Fontlab - and you could post there as well to see if anybody can help you get a grip on the tech.
Aside from the technical issues I've already mentioned, is there anything that still sticks out as inherently flawed? How does the spacing look now?
I've already widened the feet on the Ohm from what I've shown. I'm still a bit concerned with the terminals on the C, G, S and c, e, s groups. Might as well bring 2, €, §, £, ¢, and $ into the discussion too. Should I force them all to match each other, or are they fine as they are, or do some need fixing and not others? I'm at a bit of a loss at the moment.
I've copied and pasted the characters into a new FOG file. Sadly, the guides and templates did not carry over, but there was so much dead weight in there I suppose that's for the best. If I need reminders of the past I can look in the old files.
About the lc : I can thing of two things to try : You could pull the upper arm back slightly & see if you like that better at text size. OR you could add weight to it and move the spacing slightly further to accomodate it.
I think the e is fab. You could add a little weight to the bottom - not an issue at display size but it would make the text size work better. Maybe the s could benefit from some similar tratment above & below - but be careful. Even small changes are a big deal!
I basically like your CG& S. What issues do you see?
The main thing I am noticing at text size is that the X height portions of letters are maybe too contrasted with the overshoots. The w is dwarfed. The m & n seem to jump. I don't think you have to match the usual text convention. But the 'deviancy' could be softened.
BTW if all this seems too text centric the other thing you could do is go back & make a display version! That would be an awesome pairing.
What about a ball terminal on the 2? The Pound Serling might be better with one too.
BTW I really like the german double s. Could I see it in use?
I think the euro is okay but it is a little light. What about a heavier initial mark on the right?
The Bar D ( Eth) & Bar L have bars that are too heavy. The lc eth has the cross bar at too acute of an angle. and it is too far down. Have you seen this? < a href=http://briem.ismennt.is/2/2.11/index.htm>(link) Thorn and eth: how to get them right
What is next to that Yen Sign? It seem interesting!
The infinity is pretty weird. What's the idea there?
What if the fi lig joined at the top rather than the bottom of the dot?
What precedent is there for a curl under the 'mu' sign? That looks odd to me.
Apart from a funny feeling the sterling sign is going to fall over I think I like the Currency symbols. They seem to match the spirit of what your doing.
The Section symbol seems to belong in another font though. It's so quiet & small. Maybe try making a new one from a Cap S. Ahve a look at some other fonts. Also it should be bigger and taller - that is the Height should be that of the Cap but it should descend to the depth of the lc g.
I saw a lc k like yours the other day on a font caled Olive Green Mono. It's on the italic. I suggest you have a look at it! It has a motion woth emulating.
you mean this ?:
It's a personal logo of a sort. The father of my girlfriend at the time was a curator at the National Museum in Taiwan, and quite an accomplished ink painter and calligrapher himself, helped me with the design. I forget what it literally translates to, but it kind of sounds like "Jason". It's the first thing I ever literally carved in stone.
Since then I've put it in the "apple" key space in my fonts. Maybe it's not right, but I don't think it's right to put the Apple, inc. logo there either unless they're licensing the font.
The reverse curl of the mu is indeed rare (I've usually seen them extend left and hang straight), but I have seen it in other fonts. I'll have to get back to you on which ones.
Too many older fonts never even bothered with matching shapes for all these special characters and just substituted from whatever their basic serif or commercial pi font was. I hope that doesn't go on too much anymore.
I have read Briem's pages, but it looks like I'm due for another review.
If the fi lig connected to the top of the dot instead of the bottom, the arch of the f would be much wider and more open. The hook of the regular f is very tight in comparison already. Is it wise to make them more dissimilar?
I'm not sure what I was thinking with the infinity other than I did not want it to be a sideways eight.
This is something I've been meaning to ask in a more general context (rather than specific to this font alone) but exactly what features differentiate a text and display face (and other optical size categories, if that's the right term) of the same weight?
While I haven't found another mu with a similar recurve to the one in Agamemnon, the following have at least a noticeable rightward hang to the descender:
ITC Veljovic italic
What I noticed a lot more of, though, was the use of an italic mu in otherwise upright fonts, and quite frequently a mu from an alltogether different font, often without even regard to matching weights. These are from the big foundries like Adobe, Bitstream, URW, etc. Those that do have a matching mu often look like a u with a p descender snapped on the left upright, though about as often there is a lachrymal descender.
The C, G, S, and € each have a different spur on top. I think the one on the C is the weakest, while the one on the S is a bit too strong to directly transplant to the rest of the group. I just feel there must be some way to make them more consistent without making the designs fall apart individually.
Ouu. I found a very good example of a right-hooked mu, in Halter from the Apostrophic Lab.
I don't know how much of a precedent that counts for in font court, but there it is.
This is something I’ve been meaning to ask in a more general context (rather than specific to this font alone) but exactly what features differentiate a text and display face (and other optical size categories, if that’s the right term) of the same weight?
This is topic much discussed. There are ways of searching typophile - ( admitedly not all the search boxes work for me) but google can help. In general though it's things like x height being in a certain range and a a whole bevy of other tendencies. Really your face is never going to be used to set a novel or a whole story in a magazine even - so it will not be a text face per se. But it could be used for pull qotes and a whole range of other near text tasks now that it would not have been acceptable for before. Actually, do you agree it is better now - or not so much?
About the serifs- (CGS etc.) Your right about their needing to be different while maintaing consitancy. My adice to is to go look at the many & in some cases contradictory things that have been done & find a model which you think suits your font best & try it. Alternatively - trust your eyes. Find what feels right. Try things.
I was looking at the mu in the new Arno Pro from adobe & it actually kinda close to yours. It's basically down with a slight curve in & then a ball-ish blob. It's less wild than yours if you see what I mean. But realated. It comes with the trial version of the latest Photoshop. Maybe it's in the font shops too. The bold 36pt has a weight kinda like yoyr too. Have a look if you want.
Well, just to see how it looked, I did actually transplant the spur from the top of the S to the C, G, and €, and y'know, it works well enough, and much better than what was there before. I think the spur on the Euro mark is supposed to be angled, though, so I might have to alter that one some.
I've calmed down the mu. It still has a curved stem but the descender ends in more of a blob than a hook.
I ought to have the new one up tonight, or maybe early tomorrow.
Here is version 13 (associated PDFs up top as usual. randomtext shows ß, ð, and þ in use, though the passage is nonsense.)
In addition to the points touched upon earlier, I raised the lc serifs to the x-height, reversing the slopes on most of them. This results in some of the serifs being tapered, which I couldn't quite figure out how to match on the v, w, x, and y. Maybe it doesn't matter. At least there isn't that big jump between the serif and the humps of the m, &c.
I have printed out the latest. I am looking foreward to checking it out.
Overall I am finding the face really really pleasant. I especially like the k oe ae r s e h.
My opinion is not informed enough so take it with a grain of salt but I like the german double s too. Maybe ask a german speaker about that.
But of course I found things that bugged me too. So here is what I noticed when I printed the pangram sheet:
- on the second line 'funtidrk' the ti seem to be clashing. It's a spacing kerning issue. But before you address that you might want to alter the 't' which looks odd & a little weak compared to the robustness of the other glyphs. The ending that works for l is not strong enogh for the t
- with such a robust face the 'eth' ( reverse italic looking d with a slash) will need to be ajusted the slash is still in danger of being swallowed by the bowl. Why not extend the height of the ascender to that of the h & d?
- I am still finding the Cap M looks pidgeon toed. Can you alter those serifs at the bottom? On a related note the serif in the upper right of the Cap N seems overlong to me.
- If the end of the question mark stroke ( near the dot) was just a bot lighter and perhaps had form more lik ethe end of the lc r I think it would be better.
- the lc u seems too wide. Try reversing that lc n perhaps. It needs to have that bite out of it too - like the nmr & so on.
So on the t, do you think it should have a full hook, as the decender of the j, or a flat base like the f?
I think the t would look bizzare with the flat base. But try it & the reverse j hook & some other things besides. I suspect it will need it's own shape. But you can start from those places. Also you could try clamping on the trail end of the lc z for fun too.
Dang, I have a drawing of the t to share and now the "insert image" doohicky isn't working.
Anyway, I tried it with the ball terminals from the j and z, and they just seemed too out of place above the baseline like that, so I took the flare terminal from the c and shortened it up a tiny bit. With a lengthened crossbar, I think it looks pretty good. I hope the description is enough, since I can't get the picture to upload.
Let's see if this works:link to picture on ImageShack.us
Can't seem to hotlink an image either.
Jason, what is the latest PDF. I want to peek and give you my 2¢ but am too darn lazy to play catch up.
Miss T: The last three PDFs at the bottom of the list of attachments are the most recent. Each begin with "agamemnon13-" and refer to the same font version. "allcahracters" shows all the characters in the font at 48 pt. except the multiplication sign and y acute, which for some reason Fontographer won't include when generating the font file. The other two show the font used in blocks of text.
The image linked above your post shows changes to the t after those PDFs were made.