A Rock Band: Harmonies Project Release Chronicled
- This topic is empty.
-
Posts
-
Over the course of the Rock Band: Harmonies Project, I have often been asked (by people using the project and people interested in authoring for the project) how long it takes to chart a given song. The answer was always the same: “it depends”. It depends on a lot of factors, including the length of the song, the number of harmony parts, the nature of harmony parts (copy-paste friendly block harmony stuff or hand-authored stuff), and the nature of the source audio used to author them. It’s also something that depends heavily on the abilities/proficiencies of the person doing the authoring.
Nevertheless, on a few occasions, I kept track of the amount of time it took to author songs, but as time went on it became readily apparent that the time spent actually authoring on the chart was not all that reflective of the total time spent working on the chart. So, instead, I decided to keep a record of all of the time spent authoring a given release, and I figured I would detail it here.
The process for a given release is started six weeks ahead of time, but since we release every three weeks, I’m roughly doing the same amount of work every three week period (for the current release and an upcoming release). In this case, I was tracking this release for the week of November 8th, 2013.
Pre-Authoring Preparation – Total Time: 2h, 58m
Breakdown: Preparing Files for Audio Capture: 65m; Obtaining Audio: 1h, 53m
As explained, the pre-authoring prep part of the process (alliteration woo) involves all of the steps taken before authoring. This includes pulling the original LIVE files off my console, ripping the songs.dta and .mid files from the packages, and doing all of the work related to capturing isolated audio for authoring purposes. Sometimes this part is easy, as someone else may have the audio ripped (e.g. Phase Shift converts), but most of the time I need to get these files out of the audio. The most complex method for this involves modifying the songs.dta files to make a bunch of duplicate entries (making sure they don’t collide), isolating audio by adjusting the volume levels, and capturing it in real-time from the game.
As expected, this last part is more or less the most time-consuming component of this phase. It requires playing through songs two to three times a pop (for RBN songs where we need isolated drum audio), and a cache build. Sometimes multiple cache builds if something conflicts in the files (song id, name, track number on the album) or if I forget to include the RB3 disc information. Also lumped into this time is the time it takes me to dump them into a sorted working folder. This release is probably a fair time reflection of most releases as I only had to do a second cache rebuild for a couple of songs and had to capture drum audio for both RBN songs.
Authoring/Charting + Submission Review – Total Time: 12h, 57m
Breakdown: Authoring/Charting: 10h, 27m; Submission Review: 2h, 30m
I’ve gotten to a point where I can get pretty zippy authoring harmony upgrades. This week’s releases had upgrades ranging from a chart time of 9 minutes (Give It to Me) all the way up to 94 minutes (Spiderwebs), for an average of about 35 minutes per track. I would say the normal average dips a little higher — five charts at 15m or less certainly helps to bring it down. As expected, the easiest charts are copy-paste block harmony charts with few parts, while the most time consuming are ones that require lots of by-hand authoring of the new parts. It’s also worth noting that some components of authoring people may overlook are phrase marker adjustment (old RB tracks had two sets of phrase markers used in head-to-head, which I need to reduce to one set) and HARM2 phrase authoring — this latter part is a pain in the ass when I get phrases running into Overdrive phrases, because I can’t change the original overdrive authoring (scoring impact) as it stands today, so we end up with those neat graphical glitches.
Review time for submissions varies a lot based on the person sending in the submission and the complexity of the chart itself. If it’s a chart an author has requested a lot of help on, especially a complex one, it’s not uncommon for me to spend as much time if not more time on review than it would have taken me to author (especially if I need to provide the author with feedback). One of the big issues I need to address with submissions is note/harmony alignment, especially for the contributors for whom quantizing doesn’t always seem to work. Most of this gets caught and fixed in the in-engine review segment, but there is a lot of “zoom in really far after quantizing and make sure note start/end and lyric events line up right” going on. This release had a ‘typical’ number of submissions, though in some weeks I’ve had far more and in others I didn’t have any.
It’s worth noting that this week had very little in the way of releases ‘in excess’ of the 20 + 2 pattern we’ve been following — those weeks with keys charts, pro drums charts, and things like misheard lyrics charts also add to the time, both for charting/review and for video capture/testing.
Testing, Testing, Testing (and Video Capture) – Total Time: 7h, 57m
Breakdown: Preparing Files: 52m; Round One: 3h, 26m; Round Two: 3h, 39m
The testing phase is split into three pretty distinct parts. The first part is the simplest, and merely involves all of the file prep. This involves setting up the folder structure for the packager (and also for later prep of the packages for uploading), the updating of the songs.dta file (to include relevant fields for vocal parts, tonic info, keys if applicable, and the missing data already included on the RB3 disc, which I include for the sake of completeness for anyone looking to build their own RB3 Version files), and the creation of the songs.dta file.
The second phase (referred to as “Round One”) involves the first “in-engine” test. For this one, I need to create the packages with the packager, build a cache (usually takes 10m, repeated whenever I make a mistake), play through all of the songs in the release in-engine (I usually don’t sing along for this pass), and edit/correct them in Reaper. This release was pretty typical in that about 3/4s of the songs needed slight correction/tuning based on things I could only see (or only see easily) in the RB3 engine, such as note alignment, overlapping HARM2/HARM3 lyrics (helping me decide which to hide), weird glitches caused by accidental note insertion in an invalid row, lyric events missing in a copy paste, or weird phrase placement issues.
The final phase (“Round Two”) isn’t always the video recording phase — sometimes I do a second testing phase, particularly for egregious problems, before I move on to a video recording run — but it was for this release. It’s similar to Round One, in that I play through all of the songs again to check for mistakes, but this time I’m also singing along and capturing video (again, this is sometimes reserved for a third pass). Changes get made as a result of this pass as sometimes I miss things in the first in-engine pass and sometimes I only notice things (like alignment or incorrectly pitched phrases) when trying to sing along. It’s also important to note that this phase requires me to play along to songs receiving pro drums and/or keys/pro keys upgrades twice to capture audio without the track cutting out to be layered into the released video (the time in this phase also includes time spent splicing those videos together).
If this had been another release with a third pass, we’d be seeing about the playtime of each track added again (so another couple of hours, though I’m too lazy/disinterested to do the math right now).
Deployment/Administrative Work: Total Time: 3h, 25m
Breakdown: Preparing Archives 22m; Uploading Files/Updating Spreadsheet: 36m;
Uploading YouTube Videos + Editing Descriptions: 1h, 48m; Preparing Posts/Publishing Videos/Posting Links: 39m
When all is said and done, the final step is to push this stuff out to you. It’s a small part of the administrative work, but when it’s done, it means the release is done and being enjoyed in your hands, and it’s the part I look forward to the most.
The main step for this is to prepare all of the archives and upload them to our filehost. Once this is complete, I need to make sure I update the spreadsheet links, as well as update all of the relevant release information (the front page, the news page, the released tab, and each of the individual category tabs). I provide pksage with a list of the release info for the release to make sure it’s in the C3 blog post, and I usually drop notes on Twitter and Facebook to give people a heads-up. And, of course, once complete, I make forum posts as necessary to announce it (I no longer post on ScoreHero and 360ISO as I used to).
Finally, the last part involves video uploading. I have a much faster pipe than I did when the project started, meaning I can mostly get an entire run of 22 videos up overnight versus spreading it out over a week or so. The videos have a copy-pasted base description, which I tidy up, and I try to make sure everything is properly keyworded. Generally speaking, the videos are uploaded as unlisted and then rolled out/published in groups so that I can easily add them to playlists. A Twitter post is made for each song, and a link is added to the spreadsheet for the chart after each post.
When the last post is made and the last update is done, the project is done.
Total Time Investment for the November 8th, 2013 Release: 27h, 17m
A pretty typical amount of time over any given three week window. It’s spread pretty evenly over the three weeks, though weighted more towards the second have of week two and the third week in any cycle.
In any event, I’m hopeful this helps to give you an idea as to the time requirement for any of the releases. Authoring and submission review makes up about 40% of the timeline, while the in-engine review part eats up another third. The biggest way to help me get these times down would be to have an uptick in submitted charts (in good shape, of course — I’ve rejected submissions from authors that have been really, really sloppy), so, as always, if you’re capable of authoring vox, I’m always interested.
” src=”/wp-content/uploads/invision_emoticons/default_SA_smile.gif”>P.S. This also explains why you haven’t seen any customs and why I’ve declined most vocal authoring collabs.
” src=”/wp-content/uploads/invision_emoticons/default_SA_wink.gif”>Initially I was going to go a little more in depth w.r.t. the process and things that went on as a part of it, but I didn’t want to make it too wordy. If anyone has questions about any part I’d be more than content to answer them.
” src=”/wp-content/uploads/invision_emoticons/default_SA_cheeky.001.gif”>
- You must be logged in to reply to this topic.