Manual TrueType Hinting Coverage and Costs

Michael Jarboe

I'm curious what level of manual TrueType hinting 'most' (I know this is a big generalization, but still…) independent foundries offer with their webfonts? Is there an average here?

I'm leaving out autohinting with this list:

• Basic ClearType
• Basic ClearType/DirectWrite
• Grayscale
• Black&White

And are said foundries paying tens of thousands of dollars per typeface family to get this done?

Mike Duggan

I would imagine that anyone doing manual hinting on Webfonts, does so to target ClearType and ClearType Direct Write. Greyscale or Font Smoothing and Black and White are less relevant today. I dont know about the cost.

John Hudson

Mike, what's the status on rendering in the Metro environment? In Win10, is this still fractional positioning with supersampled greyscale, or is it back to CTDW?

Deleted Account

If one leaves ttf autohint out, the average price per glyph goes way up to a few dollars. If one keeps it in, $.56 per glyph is what commercial web fonts need, (which is not black & white or greyscale hinting).

I think, if someone is trying to sell hinting for tens of thousands per font, users should get their bellies rubbed by the hints every morning while the hints are making their breakfast and walking their dog, but that's just me.

Michael Jarboe

David, would you recommend autohinting with manual corrections, then?

Michael Jarboe

With the 'tens of thousands' I meant, say a typeface family that includes a total of 14 styles, say 300 actual glyphs per style (not including component glyphs) at a few dollars per glyph which would amount to over $12K. I'm just curious how independent foundries are tackling this, without compromising quality.

James Puckett

I'm just curious how independent foundries are tackling this, without compromising quality.

I think that the majority of independent foundries just compromise on quality and let the vendor apply autohinting. 

Michael Jarboe

James Puckett said:

I'm just curious how independent foundries are tackling this, without compromising quality.

I think that the majority of independent foundries just compromise on quality and let the vendor apply autohinting. 

Yeah, I guess the actual question I'm trying to get at is "How can independent foundries best hint their webfonts without compromising quality while still being as cost efficient as possible?"

It's just been drilled into my mind that manual hinting is the highest quality possible, yet the time and costs associated are significant.

I'm thinking of the independent foundries (however few their are) that don't work with outside distributors, controlling their own distribution and offering high quality webfonts. I imagine it'd be most cost effective to bring someone in-house to handle all the hinting work as opposed to outsourcing it.

Paul van der Laan

Most of the Bold Monday webfonts are manually hinted by myself, and a few others have been outsourced. As Mike Duggan already noted – we hint for GDI Cleartype in the first place. In most cases DirectWrite will automatically look good too then.

I believe that if you want to make a difference as an independent font foundry then you should focus on quality. Investing time to learn TrueType hinting was well worth the effort in our case because we can offer good quality hinting to our clients that order custom fonts.

kupfers

You could also just ask for a ballpark quote. Paratype is probably the largest hinting service, good quality, fast turn-around, and potentially very affordable if you only need ClearType (most common, I agree). Or ask Micah Stupak, or Noe Blanco, or any of the others mentioned in this thread: http://typedrawers.com/discussion/comment/13917

Miles Newlyn

Outsourcing our GDI Cleartype hinting (mostly to Monika at Fontwerk) has become routine for us. We haven't always done it before initial release, mostly when funds become available.
We recently lost a big license because someone client side incorrectly assumed that only the big foundries do well hinted work. We didn't get the chance to prove otherwise :-(

Deleted Account

"David, would you recommend autohinting with manual corrections, then?"

I do, as long as the corrections are to the outlines. Manual corrections to ttf autohinted glyph instructions, I don't recommend. 

I agree with Paul that if you have custom clients, the Sky is the limit for hinting quality, as well as price. Your Indy web fonts then fall out of your expertise. I came at it, custom hinting first, and web type fell out of that, our investment in web type hinting was easy, and so I'm not as well positioned to advise as Paul.

But, we have seen data indicating a migration to widows CFF for the web, (which is great, as that renders more compatibly with print) and with the rise in resolution, we feel ttf autohinting is fine for all but reading edge products, which we do by hand, or through a hinting vendor. 

However, no amount of hinting can save all designs on Windows with any kind of cleartype rendering, if Windows is lighter than print rendering. So, our type design process starts with size recommendations to the user. While it annoys type designers at work, to ask them what size this design is for — Eventually they get the message and not only pick a size, but review the design more closely at the recommended size.

This also gives the type designer ample opportunity to add offspring styles to make up for any size-limiting issues in their design, if I don't do so first. In existing families, this goes on all day with users trying to use fonts outside their "hidden" size range, asking for new styles, and poof, you wake up one day with MillerSuperExtraLargeBanner. Why not start when the family is young. And for non-Latin scripts of many kinds, there is no other sensible way, to me.

And while we're on the topic of migrating fonts from desktop to web, the other major adjustment Indies should make, is that the web needs to know every registered feature a typeface style contains from the OT tables, not from visual inspection or post-it notes in the vendor's font database. There is going to be enough of that for unregistered features. It is then a simple database and function set, up to your vendor and user, to get the right glyphs and features into the right spots in the licensing and productization.

We, as developers can then make sure our fonts work on all platforms, apps and page description languages, in the recommended size range, with whatever hinting or outline corrections are required. And for vendors, if a user has written web code that employs registered OT features, and they want to change fonts, they won't have to rewrite their code because a type designer prefers small caps in the Lowercase aka "pro" position, or any such registered feature nonsense. 

I think I've just gone over my two cents, so I hope it's worth three.

Jason Campbell

at House I do a lot of the manual hinting, and if we need additional help due to time constraints we outsource to the usual suspects. David's numbers seem right. I would typically hint for Cleartype, hinting for greyscale or B+W would only be done if a specific customer needed it, and then it would be very expensive. 

Mike Duggan

JH >> Mike, what's the status on rendering in the Metro environment? In Win10, is this still fractional positioning with supersampled greyscale, or is it back to CTDW?

It is fractional positioning with supersampled greyscale

PVDL>> we hint for GDI Cleartype in the first place. In most cases DirectWrite will automatically look good too then.

This can actually be made better. The latest release of VTT supports a new set of commands, in the high-level font hinting language (“VTT Talk”). These new commands (RES Hints (Rendering Environment Specific) can be used to hint for a variety of rendering environments. Unlike VTT Talk’s existing commands, which more or less directly map to a series of TrueType instructions, the new commands map to TrueType functions, which determine the rounding granularity dynamically. See the VTTDemo.ttf that ships with VTT for examples of this.

Michael Jarboe

Another related question, is there a point in which one could abandon TT hinting altogether, and just go direct to market with CFF webfont versions?

I'm thinking of display typefaces such as 'Hairline Display/Poster', or condensed display styles in which the recommended size would be above 24-30pts or even higher. Is there a certain threshold in which TT hinting can absolutely be abandoned and CFF versions would be completely appropriate?

James Puckett

Another related question, is there a point in which one could abandon TT hinting altogether, and just go direct to market with CFF webfont versions?

DirectWrite handles CFF fonts well enough, but isn’t in versions of Windows older than Windows 7. For Latin typefaces we’ll be there once Windows XP and Vista become commercially insignificant on the web. The real end-of-life for XP is 2019, when Microsoft will stop supporting XP for embedded systems, which means users can no longer hack embedded updates to keep desktop installs secure. Hardware failures will have forced many enterprise users off of XP by then as well. 

Non-Latin is probably going to require hinting longer than Latin. Hinting some writing systems will probably be necessary until 4k displays are standard for desktops, which is probably at least ten years away simply because the lamps in recent LED displays last so damned long.

Michael Jarboe

Interesting James. I'm thinking back to this Typekit blog post too in which some fonts at large sizes are served as CFF for smoother rendering on Windows:

http://blog.typekit.com/2011/08/31/improved-windows-rendering-for-adobe-fonts/

Which makes me think that if the typeface family in question is intended for larger sizes above a certain threshold (24pt and up?, 30pt and up?) then why not simply serve CFF webfonts from the jump?

Nick Shinn

I think that the majority of independent foundries just compromise on quality and let the vendor apply autohinting. 

That’s what I do, mostly.

I don’t think it’s much of a compromise, with high res screens.

Mike Duggan

Not all have high res

Belleve Invis

John Hudson said:

Mike, what's the status on rendering in the Metro environment? In Win10, is this still fractional positioning with supersampled greyscale, or is it back to CTDW?

DWrite. And you can actually drop hints and use a GASP entry to force bi-directional antialiasing.

Belleve Invis

James Puckett said:

Another related question, is there a point in which one could abandon TT hinting altogether, and just go direct to market with CFF webfont versions?

DirectWrite handles CFF fonts well enough, but isn’t in versions of Windows older than Windows 7. For Latin typefaces we’ll be there once Windows XP and Vista become commercially insignificant on the web. The real end-of-life for XP is 2019, when Microsoft will stop supporting XP for embedded systems, which means users can no longer hack embedded updates to keep desktop installs secure. Hardware failures will have forced many enterprise users off of XP by then as well. 

Non-Latin is probably going to require hinting longer than Latin. Hinting some writing systems will probably be necessary until 4k displays are standard for desktops, which is probably at least ten years away simply because the lamps in recent LED displays last so damned long.

Yeah, DW can handle most Chinese characters well under bi-directional AA, but the hints produced by sfdhanautohint can improve the clearity.

Nick Shinn

Not all have high res

What is the present status of “Retina” (i.e. pixel-free seeing) screen usage?

Belleve Invis

Nick Shinn said:

Not all have high res

What is the present status of “Retina” (i.e. pixel-free seeing) screen usage?

Almost NO PC users in China have a "retina" monitor. I think in the US and Europe the quantity of "retina PC" uses is still not large.

Chris Lozos

but the numbers are increasing.

Belleve Invis

Chris Lozos said:

but the numbers are increasing.

But there are still millions of lowres users.

Richard Fink

As somebody who's done quite a bit of autohinting, I don't have a clue how one would even go about figuring a price on a per glyph basis. Certainly not with a font sight unseen.
But as far as the need for TT Hinting - as long as hinted TT fonts give a sharper or clearer or more uniform or more whatever you want to name, in rasterizers programmed to interpret TT hints to good effect, I don't see the need for web fonts to be hinted TT fonts as going away at any time.
And BTW - don't think high end desktop. Think smartphone. Think cheap Android phone. That's where the hinting is crucial. In a world where screens of all kinds are ubiquitous, it's not millions. It's hundreds of millions of users that are affected. 
My three cents for today.

James Puckett

Nick Shinn said:

Not all have high res

What is the present status of “Retina” (i.e. pixel-free seeing) screen usage?

Google puts hi-res at 40+ percent on mobile devices running android. And I think we can safely assume iPhone users are a much bigger number as the non-retina iPhones can’t run recent releases of iOS (at least not useably). Hi-res desktop screens get dumped into “other” for market share surveys, so I can’t get details on that. Someone who has a paid NetMarketShare account might be able to dig up more information. But it’s probably a low number—in January only 11 percent of Americans were even using 1920x1080, and 1366x768px was the most popular. I think this is probably a combination of businesses continuing to use relatively old LCD displays that just won’t die and personal computer users preferring cheap laptops to high-end desktops.

Richard Fink said:
And BTW - don't think high end desktop. Think smartphone. Think cheap Android phone. That's where the hinting is crucial. It's hundreds of millions of users that are affected.

Android ignores TrueType hints. The Google web fonts API even strips hints from hints it serves to Android. Unless Windows 10 becomes popular in the developing world we probably don’t need to worry about hinting on mobile devices.

Dave Crossland

Android ignores TrueType hints

What about CFF hints? Since the adobe CFF renderer went into FreeType, and that version of FreeType has trickled down to Android, it seems important to me that CFF hints are correct for optimum mobile reading experiences.

Deleted Account

There were zero retina displays in 1990 when adobe's hinted bi-directional anti-aliasing was released. In 1999, apple, who serves all the same scripts as windows, and a wide variety of resolutions including desktop, and at the time, zero Retina displays, removed all TT hinting, and rendered type with just bi-di-aa. Today, CFF renders effectively on Windows, because it is not CT rendering.

So, why anyone thinks hinting today is required because of resolution, and that we can just wait for a few billion more high res devices to seep into the market, is something not entirely unlike a mystery to me. It should be obvious that rendering is the mother of all hints, today. 

Belleve Invis

Dave Crossland said:

Android ignores TrueType hints

What about CFF hints? Since the adobe CFF renderer went into FreeType, and that version of FreeType has trickled down to Android, it seems important to me that CFF hints are correct for optimum mobile reading experiences.

but Adobe's rasterizer does not go into GDI.

Belleve Invis

Richard Fink said:

As somebody who's done quite a bit of autohinting, I don't have a clue how one would even go about figuring a price on a per glyph basis. Certainly not with a font sight unseen.
But as far as the need for TT Hinting - as long as hinted TT fonts give a sharper or clearer or more uniform or more whatever you want to name, in rasterizers programmed to interpret TT hints to good effect, I don't see the need for web fonts to be hinted TT fonts as going away at any time.
And BTW - don't think high end desktop. Think smartphone. Think cheap Android phone. That's where the hinting is crucial. In a world where screens of all kinds are ubiquitous, it's not millions. It's hundreds of millions of users that are affected. 
My three cents for today.

In addition, Windows still takes over 80% of desktop share, and almost ALL of them do not have any high-resolution screen. Even if the recent DWrite enabled bi-directional AA by default, provide hints for them can also improved the clearity significantly.