Esacode

Docs / Demo projects

Text to video

Open text-to-video models make short clips on your NVIDIA GPU, each streamed live and played whole as it is made; keep the good ones and browse them.

Open text-to-video models from Hugging Face make clips of up to 10 seconds on your worker. Each video plays whole on the View page while the next one is being made; keep or remove each one, then save the kept ones as a dataset.

The flows

FlowWhat it does
1 · One prompt, several videosWan 2.2 TI2V (5B, 720p), the default model, turns one prompt into 4 videos of 5 seconds (seeds 1 to 4).
2 · A list of promptsFour different prompts, two variants each: eight videos.
3 · Small to large: the modelsThe same prompt and seeds on several models, up to LTX-2.5 with its sound: their videos side by side and the time each takes per video.
4 · Browse the kept videosThe dataset where you saved the kept videos, in the review view.

What you need

An NVIDIA GPU: 24 GB holds the default model; 8 GB works too, more slowly, with the rest of the model kept in the computer's memory (about 32 GB of it). The first run downloads the model, tens of GB (the log shows how much), once per worker. The largest models (LTX-2.3, LTX-2.5, Wan 2.2 A14B) want a GPU with 80 GB, such as a cloud worker. The worker needs Run code from projects turned on (Computing > your worker > Worker settings).

Try it

  1. Open Projects > Demos and choose Text to video. Read its readme.
  2. Open flow 1 in the Blueprint and press Run: it runs as it is on your own workers (it waits until one with an NVIDIA GPU is online). Switch to the View page to watch it. Its runs and results are yours only.
  3. To change it, choose Clone into my workspace: a copy goes into your workspace, with its readme, blueprints, the code of its blocks and its files, saved to a GitHub repository of yours (link GitHub first, under Profile > GitHub).

Learn more

Images and videos from text.