Quickstart

Get image and video super-resolution running in minutes.

---

Install

uv add lexigram-multimedia-upscale
# Optional: local reference-server dependencies
uv add "lexigram-multimedia-upscale[real-esrgan-server]"  # Real-ESRGAN server (torch + realesrgan)
uv add "lexigram-multimedia-upscale[hat-server]"          # HAT server (torch + timm)

lexigram-multimedia-upscale depends on lexigram, lexigram-contracts, and aiohttp (installed automatically).

The package works zero-config: the default backend is real-esrgan talking to http://localhost:5400.

---

Minimal Upscale

import asyncio

from lexigram import Application
from lexigram.di.module import Module, module
from lexigram.multimedia.upscale import UpscaleModule
from lexigram.contracts.multimedia import MediaAsset, UpscaleProvider, UpscaleRequest


@module(imports=[UpscaleModule.configure()])
class AppModule(Module):
    pass


async def main() -> None:
    async with Application.boot(modules=[AppModule]) as app:
        upscale = await app.container.resolve(UpscaleProvider)

        asset = MediaAsset(
            mime_type="image/png",
            provider="local",
            bytes_data=b"<your png bytes>",
        )
        result = await upscale.upscale(UpscaleRequest(asset=asset, scale_factor=4))
        if result.is_ok():
            upscaled = result.unwrap()  # MediaAsset — upscaled image bytes
            print("upscaled:", upscaled.mime_type, len(upscaled.bytes_data or b""))


if __name__ == "__main__":
    asyncio.run(main())

With an application.yaml:

multimedia:
  upscale:
    backend: "real-esrgan"
    real_esrgan_base_url: "http://localhost:5400"

Calling UpscaleModule.configure() with no arguments also works — UpscaleConfig defaults apply.

---

What Just Happened

1. UpscaleModule.configure() creates a DynamicModule with an UpscaleGenerationProvider.
2. Application.boot() runs the provider lifecycle:
   • register — UpscaleConfig is bound in the container; the configured backend (RealEsrganUpscaleProvider or HatUpscaleProvider) is constructed and bound as UpscaleProvider; an UpscaleTask handler is bound; if a VideoProcessor is already resolvable, a VideoUpscaleService is bound too.
   • boot — a no-op: all HTTP happens per-request.
3. upscale.upscale(UpscaleRequest(...)) base64-encodes the input, POSTs it to <base_url>/upscale, and returns Ok(MediaAsset) — or Err(UpscaleError) for transport/HTTP failures.

The reference server must be running on port 5400:

lexigram-upscale-real-esrgan-serve

---

Next Steps

• Guide — mental model: assets, requests, backends, results
• How-Tos — task-oriented recipes
• Configuration — every config key and env-var override
• Architecture — internal design and extension points