TAE for LTX-2.x latents (Seein mirror)
taeltx2_3.safetensors — a Tiny AutoEncoder (madebyollin's TAEHV family,
https://github.com/madebyollin/taehv) trained to decode LTX-2.3 / LTX-2.5
video latents cheaply. Seein uses it for the live
preview while a shot renders: the frame is drawn as it forms, so a wrong
take can be stopped in seconds. The final clip is decoded by LTX's own VAE;
this file only draws the in-progress picture.
An unmodified copy of the file Phosphene (https://github.com/mrbizarro/phosphene)
ships in its tae pack, so it stays downloadable for Seein installs.
Licence
MIT — LICENSE-taehv in this repo, copyright the TAEHV author.
Hardware compatibility
Log In to add your hardware
Quantized
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support