|
|
View previous topic :: View next topic |
Author |
Message |
niq24601
Joined: 07 Jul 2008 Posts: 323
|
Posted: Sun Mar 07, 2010 2:34 am Post subject: The ROCKTIMIZER: Full-band optimal paths |
|
|
Update 2010/10/25:
Happy Rock Band 3 day! I have an update on full band pathing for RB3 at therocktimizer.com. Please read it and email or comment here if you have thoughts on full-band pathing in RB3.
Update 2010/07/17:
There are a couple of janky Squeeze Ins that are technically buggy paths because of the way I handle OD phrases that take the OD meter above the maximum, e.g. Alabama Getaway (1st activation) and Cherub Rock (3rd activation). THABEAST721 and I tried a couple of these, and the Cherub Rock squeeze is definitely hittable ... you have to give up the front end squeeze and instead activate late, then hit all the OD phrases as late as possible except the last one which you hit as early as possible. He got it on the 2nd try. We tried the Alabama Getaway one about six or seven times without success. It seems to be much harder, probably because there's only one chained OD phrase instead of two, and each chained phrase gives you more timing window to abuse.
Update 2010/05/24:
New features! If you click on the Guitar or Bass link from a path page, you'll get a more detailed activation summary, including distance to the nearest nearest drum activation (in measures, not in beats), as well as the current vocal phrase, with the activation point highlighted. Coming soon: notes for squeeze outs and no-whammy.
It's a little ghettofabulous right now; it suffers from the same measure counting bugs as the song as a whole (measure counts on songs in anything other than 4/4 tend to bring the fail); it doesn't tell you where to activate in the middle of long held vocal notes; sometimes it's just plain wrong, etc. But it's better than nothing and I'll keep working to improve it.
Also at this point I'm going to stop putting every single update on SH and just refer folks to The Rocktimizer home page where I'll keep track of updates in more detail, including the status of the (small) whammy overestimation bug (this one affects fewer songs and is less likely to alter paths).
Please keep reporting bugs in this thread or via email at [url="mailto:info@therocktimizer.com"]info@therocktimizer.com[/url]. Remember that if you find a visual or notation error in the path (e.g. it says something like "3/sk0" when it should say "3/sk1", etc.), I can fix that without spending a bunch of time or money. Fixing a single broken path also won't break the bank, but if I find any huge bugs that may be a problem.
Please note that I'm not going to take ALL the fun out of full band pathing. Going forward, DLC paths will be published roughly 1 month after a song's release, so the DLC rivalry can still use recent songs and pathing will be important. On-disc songs for a hypothetical RB3 will be published either 3 or 6 months after release. RBN songs won't be pathed because I'm not made of money.
It's done. I've calculated optimal paths for all DLC through Feb 15th (Rust in Peace/Super Bowl medley). The full band pather works in roughly the same way as Kawigi's drum pather or SmugDuckling's guitar/bass pather. The major difference is that I have to break the song up into 1/8 notes instead of counting individual ticks. This leads to occasional errors because I get whammy or vocal activations slightly off, but the number of times that happens is pretty small. The overwhelming majority of paths (I would guess somewhere 95-99%) are correct to the nearest eighth note.
These are optimal paths, not optimal scores. The listed Estimated Maximum scores do not include squeezing. Squeezing can increase the optimal score by as much as 2-2.5%. Songs with talkie vocals or lots of short vocal notes (e.g. "c-c-c-c-c-can" in I Wanna Be Your Man) can have even more squeezing. This is why the thread says "optimal-ish" instead of optimal. It's unlikely that there's a better path out there, but the score might be off by a hair.
(earlier instructions and updates deleted)
EDIT: Wow am I a tool, I forgot to put in the thank yous. Lots of people lent me a hand; let me see if I can remember them.
- ajanata, SmugDuckling, and LordMathias, who all helped me understand how to process the MIDI file with score information.
- IceMage, RJUCPU, LordRylen, and ProfessorJoe for helping me understand vocal scoring.
- Dennis and Dan, friends of mine who helped me bat around ideas to imrpove runtime performance so the whole thing would take days instead of months.
- The members of ATOUT who playtested a couple of paths for me.
Last edited by niq24601 on Mon Oct 25, 2010 4:55 pm; edited 24 times in total |
|
Back to top |
|
|
niq24601
Joined: 07 Jul 2008 Posts: 323
|
Posted: Sun Mar 07, 2010 2:35 am Post subject: |
|
|
FAQ, Known Bugs/Incorrect Paths, and Feature Requests
In general, if you find a path that looks broken please send me a PM or add a reply to this thread.
Q: We FBFC'd the song using your path and ended up with a slightly higher/lower score. How'd that happen?
A: Squeezing, lag, and the difficulty estimating vocal scores can affect overall score by as much as 1%.
Q: We FBFC'd the song with a different path and ended up with a much higher score. How'd that happen?
A: That's a good question, I'd like to know about it! The most likely culprit is that your path involves a reverse squeeze or activation squeeze that my pather doesn't think is reachable. Let me know in a reply and I'll see if I can figure out what happened.
Q: We tried your path but it tells the vocalist/drummer/guitarist/bassist to use overdrive in a way that's impossible. WTF?!?!
A: Please tell me! It's possible that my handling of overfilling the OD meter leads to invalid paths, which I need to fix.
Q: Will you provide paths for RBN songs?
A: I'm not going to do it myself, but ...
Q: Will you release the source code or binaries so that we can calculate the paths ourselves?
A: Hopefully. There are legal issues where I have to convince my employer that I haven't done anything that will screw up their patents. Stay tuned.
Q: What about miss paths?
A: It would take about two days on a typical song to try each OD phrase individually. But if you know which OD phrase(s) need to be missed in the miss path, it's possible to re-calculate a specific path.
Q: What about easy/medium/hard difficulty?
A: It cost $200 to path all the DLC on expert. I'm not going do it to myself, but I will try to release binaries that will allow others to path songs on hard.
Q: You spent $200 to path all these songs? Are you batshit insane?!?
A: That's $0.20/song. People are spending $2.00/song to play the songs, and I only have to path them once. Anyway, yeah, maybe it's a little insane but I only needed to spend it to catch up. Going forward I can do all the processing on my home machine at no new cost.
Q: What about TB:RB?
A: Since GD:RB and RB3 will include harmonies, I will add support for harmonies ... eventually. For songs with harmonies I will include two optimal paths: one for solos singing and with ALL harmonies.
Q: What about BNS/early drum hits/late drum hits?
A: If you can come up with a list of songs where these things might matter, I can re-run those songs under different assumptions about drum activations.
Below is a list of known bugs and caveats. They are sorted in the order of importance. The topmost item will be fixed first, followed by the 2nd item, etc.
- Because the pather approximates activation points, it may select paths with impossible vocal RSes or ASes. Please let me know if you find one; spotting these is very tricky and I'm not a vocalist.
- Will be fixed ... soonThe vocal path ought to list stuff like "AT", "BOD", "FRS", etc., but doesn't.
- Will be fixed ... soon. If a sustain lasts through the marker that starts an OD phrase, the pather thinks that sustain can be whammied, e.g. Alive. This affects way fewer songs than the other whammy bug so I can fix it easily.[/url].
- In some songs where instruments stay in activation through the end, the activation doesn't appear in the chart, even though the score includes it. Examples: Cups and Cakes (drums), Rob the Prez-o-Dent (Guitar & Bass).
- Path grouping in the summaries doesn't always do the right thing (e.g. Ironic).
- Drum squeeze in/out to chain/avoid chaining overdrive IS considered. Drum early/late hits to show/hide drum fills is NOT considered. In general late hits are unlikely to change the place in the song where the activation would occur but might change the score. Early hits however might result in substantially better paths.
- Paths for fast-tempo songs where each BEAT is two quarter notes (e.g. Aesthetics of hate, Bodhisattva, Give it All, etc.) are divided by quarter note instead of eighth note. This means they're more likely to get incorrect paths. Let me know if you find one.
- When the guitar and/or bass overfill OD, the pather will often select paths where the guitar/bass decide not to whammy. Obviously whammy or not, it doesn't matter. You should look closely at the path to see if "no whammy" is truly required. Examples of required no-whammy include Drain You and Disposable Teens m62-70.
- Also for reasons I don't understand the pather sometimes suggests that guitar/bass not whammy even though there's no reason not to (e.g. during overdrive even if there's no OD phrase at the end that you are trying to avoid chaining). In general, it's safe to whammy unless you see guitar bass drop out of OD [i]just before
picking up an OD phrase.
- Vocal paths are sometimes listed wrong; I know I've seen examples but I need to go back and find them.
- Changes in the drum multiplier due to drum fills covering notes are not considered. This means that final scores for songs with early fills (e.g. Lenny, Just What I Needed) have slightly incorrect scores.
- Paths where the vocalists wastes OD (by filling more than a full bar) list "4/sk0" instead of "5/sk0" or "6/sk0" etc. Examples: Livin' at the Corner of Dude and Catastrophe. In these paths the number of skips counts from the fourth OD phrase. For example, in Livin' at the corner ... , the first activation is listed as 4/sk1 rather than 5/sk0.
- Measure scores (listed below the drum chart) are sometimes wrong if the song is in a time signature other than 4/4.
Known songs with bad paths:
- Clouds over California (broken unison bonuses) -- "Fixed" 5/17, though hit by the whammy overestimation bug; bass should activate at 104 or so.
- Cherub Rock (broken unison bonus) -- Fixed 5/17, but the new path may have an impossible Squeeze in on drums.
- Visions (broken unison bonus) -- Fixed 5/22
- Chop Suey (impossible RS)
- Ramblin' Man? (impossible RS, other weirdness)
- Psycho Killer (impossible RS)
- Feel the Pain (image is correct but vocal notation should be 3/sk1-3/sk0-2/sk3 or 3/AT-3/sk0-2/sk3)
- Eye of the Tiger (guitar can activate IMMEDIATELY at 9, squeeze out at m13, and then be at full OD bar by m26
- More as people find them...
Last edited by niq24601 on Sat May 22, 2010 9:43 pm; edited 20 times in total |
|
Back to top |
|
|
MasterNobody
Joined: 11 Jul 2008 Posts: 2791 Location: Duxbury, MA
|
Posted: Sun Mar 07, 2010 6:22 am Post subject: |
|
|
Wow, is this really ALL the DLC??? And if it is, how do we find exactly what we're looking for? This is revolutionary as fuck... _________________
|
|
Back to top |
|
|
niq24601
Joined: 07 Jul 2008 Posts: 323
|
|
Back to top |
|
|
Sarg338
Joined: 07 Feb 2008 Posts: 5143
|
Posted: Tue Mar 09, 2010 11:39 pm Post subject: |
|
|
I said this for Guitar/Bass, and I said it for drums (although it was already there, I WOULD'VE said it if it wasn't).
Needs more Hard.
kthx
EDIT: More I read this reply, the more I think it could had been read in a dick-ish way because I never really commented on it and just asked for something.
This is an awesome thing you are doing. A year ago, no one would have imagined optimal FB paths.
As I said, since I'm a hard player, hopefully you expand this to the rest of the difficulties so my band could try these out. Of course, no rush I'd rather you get the bugs worked out and stuff before thinking about doing that. |
|
Back to top |
|
|
niq24601
Joined: 07 Jul 2008 Posts: 323
|
|
Back to top |
|
|
eyecawl
Joined: 01 Dec 2008 Posts: 57 Location: Cleveland, Ohio
|
Posted: Wed Mar 10, 2010 6:35 am Post subject: |
|
|
Just out of interest, would you mind discussing the algorithm that you used to compute the paths (unless it's brute force)? Good work on this project. |
|
Back to top |
|
|
bclare
Joined: 21 Jun 2008 Posts: 6048 Location: Boston
|
Posted: Wed Mar 10, 2010 9:14 am Post subject: Re: Full Band Optimal Paths (DLC -- On-disc coming 3/31) |
|
|
niq24601 wrote: | Please note that I'm not going to take ALL the fun out of full band pathing. Going forward, DLC paths will be published roughly 1 month after a song's release, so the DLC rivalry can still use recent songs and pathing will be important. On-disc songs for a hypothetical RB3 will be published either 3 or 6 months after release. RBN songs won't be pathed because I'm not made of money. |
lol
Thanks for doing this though, really great resource to have. Do you think you might ever present the information in a more organized and sortable way, like apathtoofar has done with guitar/bass paths? _________________
I'm back I suppose |
|
Back to top |
|
|
niq24601
Joined: 07 Jul 2008 Posts: 323
|
|
Back to top |
|
|
Moomoomoo1
Joined: 16 Feb 2009 Posts: 178 Location: In a van down by the river
|
Posted: Wed Mar 10, 2010 9:16 pm Post subject: |
|
|
Wow, this is some pretty amazing stuff. I never thought this could be done... _________________
|
|
Back to top |
|
|
FatalPhoenixx
Joined: 02 Nov 2008 Posts: 11 Location: Yorba inda, CA
|
Posted: Thu Mar 25, 2010 5:48 am Post subject: |
|
|
You're not taking the fun out of it? Come on you just did all the hard work for us and now its basically who has the better skilled band than with someone who has come up with better path........ _________________
|
|
Back to top |
|
|
ProffessorJoe
Joined: 02 Feb 2008 Posts: 1681 Location: Scottsdale, Arizona
|
Posted: Thu Mar 25, 2010 6:07 am Post subject: |
|
|
ya once you release an optimizer its all or nothing. Waiting a little bit to let pathers have a chance to path really doesn't help anything. So i would say you would be better off just releasing the paths when you can instead of waiting. _________________
|
|
Back to top |
|
|
Deschain
Joined: 15 Dec 2006 Posts: 2137 Location: Austin, TX
|
Posted: Thu Mar 25, 2010 7:15 am Post subject: |
|
|
FatalPhoenixx wrote: | You're not taking the fun out of it? Come on you just did all the hard work for us and now its basically who has the better skilled band than with someone who has come up with better path........ |
Excluding miss paths, which the program doesn't account for, you can still tweak what it puts out. I'd say the program does 98% of the work, but good vocal squeezing can likely beat whatever it pumps out.
But yeah, it's going to be a hell of a lot harder to beat a supergroup attempting a Bat Country FBFC. _________________
Fraggle Rock
Lars Ulrich said, not wrote: | It's only 7 1/2 minutes! F***, we gotta put some more riffs in there. |
|
|
Back to top |
|
|
Shaftedofadown
Joined: 30 Nov 2007 Posts: 1408 Location: Hoover, AL
|
Posted: Thu Mar 25, 2010 7:43 am Post subject: |
|
|
Deschain wrote: | FatalPhoenixx wrote: | You're not taking the fun out of it? Come on you just did all the hard work for us and now its basically who has the better skilled band than with someone who has come up with better path........ |
Excluding miss paths, which the program doesn't account for, you can still tweak what it puts out. I'd say the program does 98% of the work, but good vocal squeezing can likely beat whatever it pumps out.
But yeah, it's going to be a hell of a lot harder to beat a supergroup attempting a Bat Country FBFC. |
fatalphoenixx doesn't know how to path anyway so it's actually a good thing for him lolol (just kidding)
i'm still totally fine with this generator. like apathtoofar, you can always squeeze out a better score if you're good enough. and besides, there's always going to be some glitch, or some human error that will cause a lower score (i.e. a disappearing OD resulting in no unison bonus).
and Bat Country FBFC? I still don't think it's possible. it would be insane to get two guitar/drum runs to connect... granted that a guitarist like strikebowler could just be consistent and it would just take a drum run for the fbfc. i've heard that he triple FCed it with a few people a few months ago, still no fbfc though. _________________
BigFatBob wrote: | DAAAAAAAAAAAAAAAAAAAAAAAAAAMN, SUPERMAN, THOUGH. I ain't never seen a nigga puttin' a song on his back like that before, though! |
|
|
Back to top |
|
|
NoahTheDuke
Joined: 27 Sep 2008 Posts: 178
|
Posted: Fri Mar 26, 2010 3:13 am Post subject: Re: Full Band Optimal Paths (DLC -- On-disc coming 3/31) |
|
|
niq24601 wrote: | RBN songs won't be pathed because I'm not made of money. |
I don't know what kind of protection you want to keep on your program, but if you care to, you could always release it for RBN authors to use/produce their own paths.
*shrugs* Just a thought.
Noah |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
Copyright © 2006-2024 ScoreHero, LLC
|
Powered by phpBB
|