Thursday, April 24, 2008

(Re)Learning Finale, Part 6 – Wrapping it up

(You might wish to read Part 1, Part 2, Part 3, Part 4 and Part 5 first.)

*note - this is a fairly long post in which I talk about Finale's interface and underlying mechanics, some things the program does well, and the miserable situation brought on by marketing needs. Have fun!

What have I learned about Finale?

I've been reacquainting myself with MakeMusic's Finale 2008 over the last couple of weeks and have come to a new appreciation of the software. I have a much greater familiarity with the kind of thinking that created it and the market pressures that have made it what it is today. I am also much more comfortable entering and editing music within the confines of the system. At this point, I would rate myself conversant, but not yet to my level of expertise and efficiency in Sibelius. I've had to achieve many graphic effects the hard way. The plus side is that I now have a much deeper understanding about how the underlying programming works and can use that to my advantage in the future.

Tools and Modes

At its core, Finale is a program from the late 1980s that has persevered mightily. Over time, it has accreted new features and refined some of its more glaring flaws, but its basic way of doing things hasn't changed much in 20 years. Every aspect of the program is compartmentalized, which lends itself very well to a particular style of workflow. If you want to 1) enter notes, 2) add dynamics and directions, 3) add slurs and lines 4) layout the page and tweak positions, Finale works fine. The advantage to this kind of tool-based design is that working on articulations doesn't affect slurs and lines; adjusting spacing for one measure doesn't necessarily affect the spacing of any other measure. It serves as a mental focus – to make the user contemplate only one aspect of the music at a time.

The downside is inherent to the design. Because the user can only deal with one aspect at a time, switching between many tools to accomplish a given task is often required. The recently-introduced Selection Tool simplifies the process; I can drag most items, and a context menu allows basic edits to be made. Double-clicking any given item generally switches to the appropriate tool. However, creating any item requires the user to be in a particular mode.

Aza Raskin at Humanized wrote a very useful discussion about the dangers of modes here. The basic gist is that having to stop and recognize what mode the system is in before taking an action interrupts the train of thought. Resuming the task after determining the mode adds a cognitive burden, even to experienced users. Incorrectly determining the mode of the program causes a mode error – if I think the program is set up to input articulations when it is actually in Expression mode, holding down H and clicking a note will produce a sf under the note, instead of my desired harmonic diamond above the note. To avoid such an error, I will tend to clear the current tool by pressing ESC, and then select the desired mode of entry. These interruptions of the train of thought make editing music in Finale much more about the system used than the flow of the work. (Happily, Finale does have some quasimodal inputs – using the right click + drag to move the score around is a prime example)

With that said, it is certainly possible for experienced users to produce quality work quickly. Getting in the flow of the program and knowing how things work goes a long way toward increasing efficiency. Of course, it doesn't hurt to have an encyclopedic knowledge of the arcana of music engraving. No computer program can be a perfect engraver – a well-trained and judicious eye is necessary for high-level output.

Experience and Automation

In years past, engraving a page of music required the precise hand-positioning of every single mark on every single page. From page numbers to slur curvature, there was no such thing as feasible automation. Typeset music could never match the nuance and subtlety of a page engraved in tin or composed from acetate rub-offs. When computers came into use for engraving in the 60s and 70s, hand-positioning of every symbol was still required. As software became more sophisticated, an increasing percentage of the page could be automated. Today, many standard engraving rules are followed without the user's input. Many modern users don't even know what "casting off" means, because fitting measures of music on a staff is not only automated, but so easily modified. Getting it right the first time is simply no longer necessary.

Finale poses a particular challenge. On one hand, I should be able expect the program to follow the most basic engraving rules. Stem lengths and directions, beam angles, instrument name positioning, etc. On the other hand, I am responsible for every mark on every page that leaves my workstation. Generally average to poor automatic placement on Finale's part means that I can take nothing for granted. Every line has to be carefully nudged into place and reexamined periodically to ensure that it hasn't been repositioned without my input. Every beam and every articulation has to be adjusted slightly. What ends up happening is that the very automation that we prize from the software as time-saving in fact requires us to spend more time making sure that it doesn't screw things up. Sometimes it feels as though Score users have it right – very little automatic anything. Sure, it may take a little longer, but it's right the first time without any second-guessing.

So, do you like anything about Finale?

Compared with Sibelius, the typography, slurs and instrument/group bracketing are significantly superior. Handling of text expressions is a wholly different approach that involves a tradeoff. Finale includes a limited drawing module and Sibelius does not. There is a general sense of being able to adjust many more settings with finer granularity, though I'm not sure how far that goes.

The basic typography in Finale won't win any awards, but it does include some crucial adjustments that Sibelius sorely lacks. Kerning is the most important of these. Being able to adjust the individual spacing between letters allows the engraver to correct prominent text so that it flows better. Individual-character baseline-shifting and super/subscript are other welcome settings.

Text expressions are an odd beast in Finale, but offer some unique advantages over Sibelius. When one enters an expression, one must choose between selecting an existing expression and creating a new one. This system creates a master list of all expressions in the score. Every time one places an f, one just invokes another instance of the master f. So if you want to change something about every given expression in the score, one simply goes to the list, makes the change once and it is propagated throughout the score instantly. To make a change for just one instance, one must create an entirely new expression. This method allows a very fine global control over the positioning and text content of all of the expressions in the score. (This also belies the program's beginnings in the late '80s – this approach saves a lot of memory in big scores.)

Sibelius, on the other hand, uses a system of text styles. Each text object is a separate container that inherits a particular style's properties. If the style changes, so does the format of the text. Individual expressions must be altered individually. Text styles can help ensure a consistent look throughout the piece. Both programs' systems involve tradeoffs. Finale requires micromanagement to make sure that every expression conforms to a consistent style (did I make stacc. and marc. both 12pt Times italic?). Sibelius requires micromanagement to make sure that each expression is properly positioned and consistent with others of its type (did I have "a2" or "a 2" earlier?). In the end, this one comes down to personal preference.

Finale's drawing module, the Shape Designer, is unique to the program. It allows custom vector graphics to be precisely created and placed in the score. Even though the module is redolent of early-90s MS Paint, it is possible to create some quite complex images. It's possible to create some rudimentary shapes with Sibelius, but for any degree of difficulty, it's far preferable to use a third-party program like Illustrator.

I've talked in some detail about Finale's instrument grouping and slurs before, so I won't retread here.

Marketing Pressures

MakeMusic has been in a bind for quite a while. They are saddled with a program with an archaic interface and an entrenched group of expert users who rely on it for their livelihood. They are in business to make money. Music notation has ever been a niche market in a niche market. Music engravers are an even smaller share of that already miniscule market. Their strategy has been twofold – market Finale and its associated products to educators, composers and recording artsts and release upgrades every year. This strategy reliably results in gimmicky features tacked on to the program for every release. Instead of fixing core issues with the program, they spend their development time and money on being able to input notes via guitar or microphone and automatically add dynamics. Now I can add an audio track to a score and mix the volume and panning of individual staves. I understand why they think they have to jazz up each new version with something, anything, to get more new people to buy the software. Finale's yearly upgrades at $119.95 (or $169.95 for those who are bad and skip years) have been described as more akin to leasing the program rather than owning it.

It is not just MakeMusic's desire to attract new users that inhibits any true innovation. Finale's expert users have become accustomed to the many quirks of the program. Dramatically changing the interface would alienate them. Instead of going through and stripping out old, obsolete code and building a fresh program, they are forced to keep supporting the last 20 years of users and files. To correct the most egregious faults, they have taken to including certain plug-in packages rather than going through and actually fixing them.

Here's an image for you: Finale is a man riding a rickety unicycle holding a large basket of watermelons. He has been doing this all day every day for many years. His clothes are threadbare and the unicycle is in need of a new inner tube and probably new spokes wouldn't hurt. The watermelons at the bottom of the basket are rotting a bit. Periodically, his manager comes along and tosses a few more watermelons on top of the heap.

What you're supposed to do is lose the watermelons and get the guy a shower, a shave, a new suit and a bicycle. Or maybe a Maserati. What you're not supposed to do is decide that maybe cantaloupe would look more enticing on top of the basket.

So what's the verdict?

Finale is an immensely powerful program that can accomplish virtually any given musical situation. However, the interface and underlying workings of the program are clumsily-updated holdovers from the 1980s. Barring a substantial shift in marketing strategy from MakeMusic, it is likely that future updates will follow their well-established pattern. As Sibelius improves its fine-tuning capabilities and steadily encroaches upon Finale's market share, more and more music publishers will start preferring it. Right now, though, it is the dominant software and I am well-served by fluency in it. I can only hope that MakeMusic takes a good, hard, look at what they have and make their next release a leaner, faster, better system rather than more watermelons on the basket.

3 comments:

Lucas Gonze said...

Beautifully articulate writing on the character of software.

Speaking as a software developer and musician, I often notice how the development process (and associated business issues) affects the outcome. Sometimes there is a single big account that can't live without a difficult and hopelessly obscure feature, and bug fixing is sacrificed to land the deal. Other times the investors and business press want a trendy new feature like tagging. Or there is a new OS that requires a mammoth amount of work just to keep the same features working.

The Lilypond free software community has its own dynamics. They're very different from commercial developers', but still produce endless annoyances. They do get the bugs fixed, but at the expense of a brutal learning curve.

Matthew Maslanka said...

Thank you for your kind words! You're right that commercial and open-source software have very different ways of going about doing things with different attendant difficulties.

On the whole, I tend to gravitate toward commercial software for mission-critical projects because I can get honest-to-god support from people who are obligated to support me. For mild tinkering-around, I'm perfectly happy to download a pre-alpha copy of some guy's neat idea.

That there are only a handful of open-source projects that have gained any notoriety on a large scale speaks to the difficulty of volunteer collaborative efforts. Without a dedicated, paid, full-time core group, a project will most likely fizzle out at best after a scant handful of revisions. This is not to say that early adopters of commercial software aren't left holding the bag at times, but there tends to be more insurance that the software will still be supported after six months and not abandoned due to Real Life(tm).

Unknown said...

I know I should not be commenting on such an old post, but I want to address few misconceptions.

The LilyPond project has been around since 1996, with continued contributions from the developers and community since then, offering excellent support (in mailing lists and the excellent documentation, online or in PDF). You can even pay them to fix bugs and implement new functionality you need. Can you say that about Finale?

I have typeset large band pieces as well as horn quartets, quintets, sextets and choir repertoire in LilyPond since 2005, and I can say, without doubt, that it is the only software that produces decent output with minimal tweaking.

By minimal I don't mean the lots of tiny (or not so tiny) adjustments you have to make in Finale (I have suffered it too), but rather polish the general music appearance, a system break here, a page turn there and that's all. Since version 2.12, the generated output is almost collision-free.

I am still amazed about the huge amount of work you have to do in Finale to get something just readable. LilyPond has brought very decent engraving to everyone: I am tired of playing pieces in my band arranged in Finale that have been printed without any regard to engraving quality because most amateur arrangers do not care and do not know about engraving.

Just to name a few things you have striggled with in your blog:
- Nice system braces? Done, automatically, not just streched.
- Limitless arbitrary sytem grouping and nesting, engraved like in your 1965 example? Out-of-the-box.
- Text kerning? Implemented.
- Tremolos? On single notes, whole notes...
- Optical music font scaling? Built-in since the beginning.
- Tiny adjustments? All objects are movable and tunable (and documented :)
- Bézier slurs? Highly customizable (http://wiki.lilynet.net/index.php/Slurs). You can even have a slur that begins with dots, changes to dashes and has a solid ending.
- Set a fixed number of pages/systems per page/horizontal spacing and reflow all the music? Trivial.

A nice feature is the output as vector graphics (SVG) which enables further enhancements and processing in Inkscape or Adobe Illustrator.

Sure, it may have some annoyances (like other music software), which are being solved and improved in every single version; it is very, very far from limited or unsupported.

There is one thing I am sure of: if it was proprietary, it would be as expensive, if not much more, than both Finale and Sibelius.

Keep this wonderful blog alive ;)