xtx.api provides a lightweight R interface for cross-modal generative models —
systems that transform text, images, audio, and video into new media.
It is designed to work alongside:
- stt.api — speech → text
- tts.api — text → speech
- xtx.api — cross-modal generation
Together, these packages form a minimal, composable toolkit for multimodal AI workflows in R.
xtx.api is intended to support:
- Text → Image
- Text → Video
- Image → Video
- Audio → Video
Backends may include:
- Diffusion-based image/video models
- Talking-head systems (e.g., SadTalker, D-ID)
- Hosted APIs (OpenAI, Replicate, etc.)
| Container | Repo | Port | Purpose |
|---|---|---|---|
| sadtalker | cornball-ai/faster-SadTalker-API | 10364 | Talking head video |
| diffusers-api | cornball-ai/diffusers_api (private) | 8000 | Image generation |
library(xtx.api)
xtx_generate(
input = "A cinematic shot of a robot walking through fog",
modality = "text2video"
)