Description
Accidental Sect Master is a game about running a cultivation sect. If you often imagine yourself inside the fantasy worlds of the novels you read, the author hopes this world is your gift. It is made for players who love cultivation stories, running a sect and building relationships with characters, and for anyone who wants to run a sect of their own, compete with other sect masters for talent and raise a new generation.
It is more than a map waiting for you to finish quests: the sect develops, characters grow, and other factions pursue goals of their own. NPCs run on a locally deployed Qwen3-4B-Instruct-2507 model (Q4_K_M quantization), so they can converse smoothly with the protagonist and remember what was said. The author is also considering an API option so players who want a better experience can plug in stronger models.
You can build an army and raise your own prodigies, with two tech trees to pursue, or found a ranked university of your own: early on you send your prodigies to study abroad, and once the university grows strong, other NPCs start applying to it. According to the author, the game has a sweeping setting with several intersecting main storylines; cutscene voiceovers did not make it into this build for lack of time, and assassination and infiltration are planned for later. The author has also fine-tuned a Stable Diffusion 1.5 model for a ranch beast-breeding system that may come in a future update.
The game builds on an Unreal Engine prototype that existed before the hackathon: the rough map layout and some basic features were done beforehand. During the event, work focused on sect management and character interaction, including the opening story and branching dialogue, the tutorial, resource buildings and population assignment, character management and research and education screens, plus the integration of local-model NPC dialogue and Tripo creature assets. The score uses public-domain pieces such as Canon and High Mountains and Flowing Water, which the author edited in MIDI and arranged for orchestra with help from GPT. The current release is a Windows demo.
How Tripo was used
Tripo turns creature design references into 3D assets ready for Unreal Engine. It feeds the game's wildlife pipeline and shortens the path from character concept to game resource.
- Pipeline: starting from reference art, models are generated with Tripo Smart Mesh, checked for mesh and texture quality, given rigging and animation assets, and brought into Unreal Engine.
- Example: the gray-ridged wolf in the game is a 903-face model with 2K textures, an animal rig and a walking animation.
- Regional variety: the author used Tripo to diversify the world's models so that every region feels different.
