How AI Tools Built an Animated D&D Show
The team behind a YouTube D&D project used Midjourney, Seedream, Seedance 2.5, ElevenLabs, HeyGen, and Blender to create 77 Arcane-style animated inserts for the third episode.
The team behind the YouTube project «Как кости лягут» used generative AI to create 77 animated inserts in the style of Arcane for the third episode of a Dungeons & Dragons show.
- Character portraits and references. Based on photos of the players, the team generated hero portraits in Midjourney. In Seedream, each character was shown in full-body view from four angles with eight different emotions.
- Scene assembly. The references were uploaded to Seedance 2.5 with instructions on which image to use for the face, costume, and lute. Camera movement and actions were timed second by second. With a single prompt, the team built a scene where a bard takes a chord mid-somersault and finishes off a wolf.
- Voice, lip sync, and editing. NPC voices were created in ElevenLabs, lip movements were adjusted in HeyGen, and the finished scenes were composited into the recorded session. Even the dice rolls were handled: 42 dice were modeled in Blender and 180 rolls were rendered.
The toolset is collected in the Syntx AI aggregator. The third episode is available on YouTube.
The case shows how modern AI pipelines can assemble complex animated inserts from specialized services, from character generation to voice and lip sync.
Why it matters
AnalysisIt demonstrates a practical AI workflow for game and video content, where multiple specialized models replace parts of traditional animation production. It also highlights the importance of structured prompts, references, and careful editing.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
Enthusiast Built the Most Impressive AirDrop for Apple Vision Pro
Using GPT-6.2 Astra, ML-SHARP, and Tripo, a developer created a pinch-based interaction that lets users enter a photo on Apple Vision Pro.
Complete Guide to Higgsfield Cinema Studio 4.0: Scene Management, Soul 2.0, and API Automation
In-depth practical breakdown of the major Higgsfield update: Cinema Studio 4.0, consistent characters with Soul 2.0, API-based generation, and a ready-to-use UGC pipeline.
LLM (Large Language Model)
A fundamental class of neural network architectures based on autoregressive transformers, predicting the probabilistic distribution of subsequent tokens and demonstrating emergent properties of abstract reasoning, code synthesis, and logical inference.
Sora and AI Video (Video Generation from Text)
A cutting-edge class of generative neural networks capable of creating photorealistic dynamic videos from text or static images (OpenAI Sora, Runway Gen-3, Kling, Luma Dream Machine).