new file mode 100644
index 0000000..568ab7e
@@ -0,0 +1,24 @@
+---
+type: idea
+title: voice to 3D model
+tags: [ai, 3d, creative, software]
+status: raw
+first_captured: 2026-01-02
+updated: 2026-04-10
+sources:
+ - sources/ideaflow/2026-01-02_talk-to-an-llm-and-find-3d-model.md
+ - sources/google-sheets-ideas.md
+---
+
+# voice to 3D model
+
+talk to an LLM and find/generate a 3D model of anything — yoga poses, engineering schematics, sculpture. "voice sculpture; voice to nano banana." a creative tool bridging natural language to 3D space.
+
+**spreadsheet evaluation:** originality 7/10, excitement 3/10, MVP 2-4 weeks. competitive landscape: text-to-3D exists (OpenAI Shap-E, DreamFusion, Meshy.ai, Luma AI Genie) — voice-to-text-to-3D is trivially composable but no product optimizes the conversational refinement UX ("make the arm longer," "rotate 30 degrees"). technically a thin wrapper on STT + text-to-3D. quality depends on underlying 3D model quality which is improving fast. tech depth 8/10, labeled [ARCHIVE] ("cool but massive scope").
+
+---
+
+## timeline
+
+- [2026-01-02] captured — voice-to-3D concept
+- [2026-04-10] google sheets evaluation — 7/10 originality, 3/10 excitement, [ARCHIVE] tier