dhee-bundle-cartoon-explainer

Narrated cartoon explainer videos: Ideogram 4 cartoon scenes, Qwen3-TTS cloned-voice narration, LTX 2.3 motion + crossfades.

A Dhee bundle (pipeline) providing the bundle id cartoon_explainer. Discovered by dhee-core via the dhee-bundle-* npm convention; referenced as npm:dhee-bundle-cartoon-explainer#cartoon_explainer.

Sibling of dhee-bundle-infographics: same brief → plan → narration flow, but instead of flat data panels it renders each segment as a vivid cartoon scene (characters + visual metaphor) and animates it with generative LTX motion — which suits illustrative art where it would mangle crisp infographic text.

Pipeline

brief → outline → infographic_plan ─┬─ segment_scene_prompt → segment_image    (Ideogram 4 cartoon scene / comfy.tti)
                                    │
                  narrator_voice ───┼─ segment_narration → segment_audio       (Qwen3-TTS / comfy.tts, cloned voice)
                                    │
                                    └─ segment_motion                           (LLM per-scene motion directive)
                                                                  │
   segment_image + segment_audio + segment_motion → segment_video               (comfy.ltx_director, single-still, audio-driven, guideStrength≈0.55)
                                                                  │
                                                          final_video           (ffmpeg.concat, xfade crossfades)

Each segment becomes one vivid Ideogram 4 cartoon scene (a recurring mascot + metaphor that dramatizes the idea, key number rendered as in-scene text) and a Qwen3-TTS narration track in one cloned voice. A per-scene segment_motion LLM directive tells LTX what to animate; guideStrength is loosened so the motion actually moves. Audio is generated first so each clip is sized to its narration.

Requirements

• Runners: llm.generate, comfy.tti, comfy.ltx_director, ffmpeg.concat (built into dhee-core) and comfy.tts — install dhee-runner-tts.
• ComfyUI with Ideogram 4, Qwen3-TTS (FB_Qwen3TTS*), and the LTX 2.3 Director nodes at your self.local endpoint.

Drop-in workflows (per machine)

Replace the shipped workflows/ API-format graphs with your machine's exports if node ids differ: ideogram4.json (+ manifest), qwen3_voice_save.json (designs/saves Narrator), qwen3_narration_cloned.json (clones it per segment), ltx_director_local.json (single-still audio-driven motion).

Layout

bundles/cartoon_explainer/
  bundle.json     prompts/   schemas/   workflows/   inputs/brief.md

License

MIT