[pyddr-devel] Level an Difficulty sorting in songselect

[Frank Foeth]

Hi, Pavel,

Sorry I missed your mail, I'll rerigg my mail program to not just get
mail on my demand.

> > What I would currently dearly love to incorporate into any selection
> > interface is a function that objectively establishes the difficulty of a
> > dance.
> > Another thing I would like is the ability to find groups of songs with
> > specific difficulties. E.g. dances with jumps; or songs with rhythmic
> > difficulties. Currently I can't even see if a dance has jumps, let alone
> > the number of possible jumps used, etc.
> 
> I would love to see such capabilities as well. I've pondered taking a
> shot at creating an algorithm to parse the steps and figure out how
> difficult they would be (it's an interesting machine learning problem
> that might even constitute publishable research). Unfortunately, it
> would be to be a big project (and might require recruiting actual
> players of varying skill levels and familiarity with songs to test it).
> 
> There is also the problem of defining the difficulty of a dance
> objectively: there is variability between players (for some, the
> limiting factor is their ability to read steps, while for the others,
> it's ability to move fast enough), within a player (a song might be full
> of "surprises" that are only hard the first time through), and between
> styles (with bar vs. without).

One of my hobbies is climbing. In a hall unfortunately, but anyway. They
have similar rating problems, and they ignore them. Especially short
people can't climb more difficult routes built for long people and
sometimes visa versa. Of course you rate a route according to what a
higher level climber thinks of a route, and opinions vary, but the
important part is that you have a fair idea before starting in which of
the 15 or so categories in a hall (outside perhaps 20 categories) the
route lies. Even more so because climbing strenuous routes before you're
warm enough kills your muscles for the day. 
In dances more categories are possible, because it's easier to compare
them. And if the user can sort using his grade, using an imperfect
system to roughly sort new and failed dances is probably wellcome.

> Though, we might have better luck having being able to generate the
> "radar" diagrams that are present in some DDR games, which break the
> difficulty up into jumpiness, speed, unpredictability, and other
> aspects. We might want to look into whether SM can do that, and how.

Actually, I'd like symbols. Those radar things may be great from a
design standpoint, but they are hard to read. A jungle of different
(sized) boots for chaos and wings (or Hermes' sandals) for air would
work a lot better for me.

> > What you wrote below will probably bring the latter closer, but not the
> > former.
> > 
> > Before I start commenting on what you wrote, I'll add something about
> > why pydance and not e.g. stepmania.
> > 1. The current interface is more concise and clear (a boon to people
> > like me), SM's interface is hell for a first time user. 
> 
> I am not a "first time user" by any standard, and I still prefer
> pydance's interface. IMO SM interface is too glitzy and cluttered.
> 
> > > Something I think we could do instead is have each song-difficulty
> > > combination be a separate entry in the song list, when the sort mode is
> > > "difficulty" or "level". So, without folders, there would be several
> > > entries for each song, each one with just one difficulty. If we want to
> > > get really fancy, pressing LEFT or RIGHT (to change the difficulty) will
> > > move the cursor to the entry with that difficulty.
> > 
> > Just to check that I understand: you want to replace the
> > songitemdisplays with danceitemdisplays, and link families of the latter
> > (=song) together to mimic the behaviour currently implemented. 
> > If this is the proposal, I see no problems.
> 
> I think that's pretty much it.
> 
> > As you do not seem to have much time to spend on this, and I nearly
> > burned away my motivation on the songselect adaptation, I propose to
> > keep the pacing down a bit. For reasons I won't discuss on a public
> > list, I can't make promisses as to whether and when anything will be
> > finished. I'll write a new response after reading about (and hopefully
> > understanding) virtual folders.
> 
> Agreed. I think that we can do this gradually. I'll try to get
> SongItemDisplay/DanceItemDisplay into the right shape in the next few
> days. (That is, to allow it to display only a subset of difficulties in
> a particular entry, without causing any regressions.) We'll see where to
> go from there.

Great,

Looking forward for the result. If you'd like me to sketch a framework
for the virtual folders, please tell me.

Yours,

Frank