Type a concept. Get a beautiful animated video.
Our proprietary LLM orchestrator decodes raw natural language into a deterministic abstract syntax tree (AST), semantically mapping user intent to a multi-dimensional scene graph of spatial transformations, temporal keyframes, and vector primitives.
The Code2Video engine bridges the abstraction gap, autonomously synthesizing production-ready Python/Manim architectures. A fine-tuned critic network performs static analysis and rigorous pre-flight validation to guarantee mathematical accuracy before execution.
The compiled scene graph is serialized and dispatched to our elastic A100 GPU cluster. The core rendering engine executes highly parallelized rasterization and vector rendering pipelines, generating 4K topological geometries in near real-time.
Leveraging modified acoustic models, our temporal alignment engine synchronizes synthetic TTS output against animation keyframes at microsecond phoneme-level precision. This non-heuristic, mathematically rigid synchronization is Vivacity's core architectural moat.
import vivacity
client = vivacity.Client(api_key="viv_...")
video = client.generate(
prompt="Prove the Cauchy-Schwarz inequality geometrically",
duration_hint=120, # target seconds
resolution="1080p",
)
print(video.url) # → https://cdn.vivacity.dev/v/abc123.mp4A prompt enters the pipeline, interpreted by a language model that maps intent to Manim scene primitives. The Code2Video framework writes valid Python/Manim code, reviewed by a critic model before rendering. WhisperX syncs the voiceover to animation keyframes at the phoneme level. End-to-end: under 90 seconds.
Mathematics and physics at undergraduate and competitive exam level. Chemistry and computer science are in active development. Full subject list in the docs.
Yes. Pro and API tiers support bulk generation and LMS embedding. Contact us for institutional pricing if you're a coaching institute or university.
Student tier and above produce clean downloads at 1080p or 4K. Free and Supported tiers produce watermarked exports.
Programmatic access to the full pipeline. Billed after your usage cycle. No monthly cap. Designed for platforms integrating generated content at scale.
WhisperX aligns voiceover to animation at phoneme-level precision. This is Vivacity's primary technical differentiator - the sync is mathematically computed, not approximated.
Manim requires Python expertise and hours of per-video coding. Vivacity abstracts prompt → scene graph → render → audio sync → export. The delta is roughly 4 hours versus 90 seconds for an equivalent 2-minute video.
Paid tier outputs are private by default. Free tier outputs may be used to improve the model unless you opt out in settings.