Direction · 7 of 7
Finish a small experience end to end before expanding — and log every change with its impact on the schedule.
The 3D prototype was born on 09/23/2026 with a written goal: deliver a playable Windows demo over two weekends (09/26–27 and 10/03–04), using Godot, models produced in Tripo and human animations, with an internal completion target of 10/04 and submission to Tripothon S1 on 10/05. The priority order set by the product was development speed, then graphical quality, then capacity for technical expansion; the visual reference is My Time at Evershine — art direction, not a commitment to match its scale or polish.
Two decisions changed course after that. On 09/29, the author decided that the 3D becomes the full game, that the 2D becomes the reference for rules and that the migration is the priority: the plan became 32 issues in two milestones, Jam 04/10 and Pós-jam. On 10/01, he decided to keep the 2D and 3D games in independent repositories, with the Godot project at the root and local systems. The board is the source of acceptance criteria; the issue is the unit of work, and the document is its why.
Project decisions
Finish a small experience end to end before expanding — and log every change with its impact on the schedule.
The 3D prototype was born on 09/23/2026 with a written goal: deliver a playable Windows demo over two weekends (09/26–27 and 10/03–04), using Godot, models produced in Tripo and human animations, with an internal completion target of 10/04 and submission to Tripothon S1 on 10/05. The priority order set by the product was development speed, then graphical quality, then capacity for technical expansion; the visual reference is My Time at Evershine — art direction, not a commitment to match its scale or polish.
Two decisions changed course after that. On 09/29, the author decided that the 3D becomes the full game, that the 2D becomes the reference for rules and that the migration is the priority: the plan became 32 issues in two milestones, Jam 04/10 and Pós-jam. On 10/01, he decided to keep the 2D and 3D games in independent repositories, with the Godot project at the root and local systems. The board is the source of acceptance criteria; the issue is the unit of work, and the document is its why.
9 principles
What is not negotiable
Small and whole before big
Finish one experience end to end before expanding the number of systems, characters or settings. The full gameplay vision (planting, animal husbandry, gathering, mining, fishing, crafting, economy, combat, quests, friendship, story) is not to be confused with the set committed to for four days of production. New functionality must replace equivalent effort or stay outside the jam commitment.
Keep Godot; Windows; single-player; third person
Godot Standard and GDScript, no migration to Unity or Unreal; Forward+ as the first option, conditional on testing on the target computer. Windows as the platform, Steam as a later destination with no required integration; browser, mobile and consoles out of scope; no multiplayer or network architecture; third-person camera with SpringArm3D.
Tripo for practically every model, no Blender
Characters, houses, objects, equipment and vegetation come from Tripo; delivery must not depend on Blender, retopology or manual modeling. The first mandatory test was a complete character working in a build outside the editor: a 65-bone prototype (09/23) and the Carro Quebrado house (09/24) validated the pipeline before volume production. The rig ended up coming from Tripo itself (humanoid Auto Rig with a Mixamo skeleton), not from Mixamo.
09/29: the 3D becomes the full game
The 2D becomes the reference for rules, the migration is the priority and the jam goes in with whatever is standing. What the 3D already has stays. Three items from the original document fell outside the jam milestone and went to Pós-jam: the LLM conversation with memory and planting; the save tied to the slots ended up shipping on 09/29 anyway. If any of them needs to come back, the decision is the author's, and the issue's milestone changes along with it.
10/01: independent repositories
The 3D lives at the root of acentauric/myths-valley-3d, on the main branch; the 2D follows its own line in acentauric/myths-valley. No development or test command requires checking out the other; there is no synchronizer for rules, data or audio. The four shared files that forked (the axe as an equippable item) await the author's decision: promote the rule to the 2D or declare the fork.
Two visual styles, and then one
Up to Build 10, SETTINGS → Scenery and time chose how the whole valley was built: Tripo (master line) or Procedural (initial test), never mixed, so that the master line could be decided by looking at both in the same place. The procedural style received no new art and was not removed in an incidental change: its removal was a task of its own, with gates, and happened on October 10, 2026, in Build 11. Today the valley is Tripo only.
The jam cut: what goes in and what waits
In: quests at the anchors that exist, the backpack, stamina on the HUD, the key table for the new screens, the first Tripo batch of handheld items, and the translation. Out: Phase 2.5 (extending the region is KML drawing and import), animals (they need rigs and clips) and interiors (they need rooms, which is a system, not a piece). A new screen four days before a submission is how you lose a submission. The rule of the cut: nothing goes into the demo without a gate.
Everything is translated, and the translation is the author's
pt-BR, English and Spanish in all player-facing text, not just the menu — a decision made before Phase 3 so the very first quest step would be born with _en and _es. The cost is tripling the content; the gain is that the valley no longer has a ceiling on its audience. The migration delivers the structure and the gate; the translated text is Ramon's.
The 2D is the reference, not the owner
The valley has its own quest mechanism (CadernoDoVale, a single progress line, no checklist) by the author's decision: a new quest here can have a different pattern, format and order. The works shipped with their effect in place before the art ("effect now, art later"). The almanac is the collection (L), the web of talents is on K, the village on P, the backpack on I and the panel on J — the same keys the 2D tutorial teaches.
8 rules
In practice, day to day
- Log every change of course in the decisions document with its impact on the schedule; new functionality replaces equivalent effort or stays outside the commitment.
- While there is an open task in the Jam 04/10 milestone, it takes priority; Pós-jam follows the dependencies described in each issue.
- When a slice closes an issue, the corresponding phase is struck off the plan in the same commit: a board and a plan that disagree are two plans.
- Each Tripo batch is approved before it runs, with the cost shown by the interface posted in the conversation before the click; 'modelos-3d' issues spend credits and require an approved cost.
- Don't treat the decisions document as the Tripothon rulebook: deadline, time zone, format, video length, categories and rules for prior projects are checked against the current form before submitting.
- Clearly identify what already existed in the 2D base and what was produced for the prototype; don't claim eligibility without checking the applicable rules.
- The deliverable is a Windows build packaged as a ZIP with all required files, an identifiable commit or tag, run instructions and the list of assets and credits.
- The redistribution license for each model (including test prototypes and the Tripo models after any cancellation of the plan) is confirmed before publishing.
Where it shows up
- The first local test (09/23/2026)
Standalone Godot project, a humanoid prototype with 65 bones (FBX without clips, then GLB with clips exported from Tripo), a 1.78 m controller with running, collision and camera; Forward+ started on the GTX 1660 Ti with Max-Q Design. - The first integrated house (09/24/2026)
The semi-automatic Tripo Studio → GLB → Godot workflow validated with the Carro Quebrado house: 65 credits, task fef21312-…, 1,915,732 → 273,548 triangles. - The board (09/29/2026)
32 issues labeled 3d, split into Jam 04/10 and Pós-jam, with area labels: sistemas, interface, conteudo, mapa, modelos-3d and qualidade. - The split (10/01/2026)
The project opens at the root, with local systems and tests that don't need the 2D checkout; models and WAVs in Git LFS; the opening and controls count failures and were falsified to confirm failure via exit code. - The build identification
"v0.1.0-dev · Build #6" (September 28, 2026), the historical milestone of this adaptation; the current distribution is Build 12 (v0.12.0) for Windows; the 2D is at 0.2.0-dev and Build 14.
Where it comes from
The documents behind this area
- 3D prototype decisions — what was decided and when
- Development history — the steps before the standalone repository
- From 2D to 3D — why the two games stay separate
- 3D plan — the game's direction and next steps
- Authored composition — how the valley is edited in Godot
- Assets with Tripo — the official 3D model workflow
- The living valley — day and night, residents and the valley's sound
- 3D changelog — what changed in each build
- Game version — the number stamped on each build