Scourge of the Necromancer

Scourge of the Necromancer makes significant use of generative AI in its production. We want to be fully upfront about it.

Visual assets (2D images + video) — all AI-generated, then human-curated and human-edited. Every shipped image and video in the game and on the Steam store page was produced with AI tools, then reviewed, edited, and selected by hand by the developer. This covers: Steam store capsules and library art, the store-page background, all in-game achievement icons, hero portraits, the Necromancer concept art, decorative UI elements, all real-time cinematics (the defeat sequence and all victory variants), and the marketing trailer's atmospheric segments. Image models used: Google Imagen 4, Google Veo 3.1 (video), Nano Banana Pro / Gemini 3 Pro Image (Google), Black Forest Labs FLUX.2, Recraft V4 Pro, Ideogram V3, and Bria RMBG-2.0 (background removal). Final assets received human editing — compositing, color correction, retouching, frame extraction, masking, and rejection of unusable outputs.

3D models — AI-generated base meshes, then human-edited in Blender. The in-game hero character models (Knight, Cleric, Ranger, Wizard, Warrior, Alchemist) started as outputs from Tripo AI's generative 3D model service (FBX + texture PNGs), but each model was then opened in Blender by the developer for hand editing — mesh cleanup, geometry adjustments, and texture fixes — before being exported for use in the game. Each finished model was imported into Unity with hand-tuned scale, rotation, material configuration, and import-setting curation. Unsatisfactory generations were discarded and re-prompted; the Blender editing step is substantial enough that the shipped models are not pure AI output.

Music — AI-generated, then human-curated and post-processed. All music tracks in the game (menu, gameplay layers, boss phases, victory, defeat) were produced with Google Gemini's music generation, then human-selected, trimmed, looped, and mixed by the developer.

Necromancer voice — AI-generated, then human-edited. The Necromancer's spoken syllables were synthesized with ElevenLabs Text-to-Speech (Eleven Multilingual v2 model), then pitch-shifted, EQ'd, and processed with reverb in Audacity by the developer to produce the final in-game delivery. No other character has spoken voice lines in the shipped game.

Sound effects — all AI-generated. Every SFX in the game (ability sounds, ambient layers, UI sounds, footsteps, kill/spawn cues, tile state cues, boss audio) was produced with ElevenLabs Sound Effects, then human-selected, trimmed, processed, and mixed by the developer. The Necromancer voice lines use ElevenLabs Text-to-Speech (covered separately above).

Runtime behavior. The game does NOT use live-generated AI content at runtime. Every shipped asset is a fixed, pre-rendered file. Player-facing text, dialogue, narrative, lore, and game design were written by the developer without AI generation.