Open-source attributions

Luna CEO's media engines run on models we host ourselves. Several are open weights released under licences that require attribution, and this page is where we give it. Attribution is a condition of those licences, not a courtesy: under CC BY 4.0 the rights terminate automatically if it is not given.

Speech synthesis

Voices are synthesised with Piper. Three of the cast voices are Piper models trained on the LibriTTS-R corpus, released under Creative Commons Attribution 4.0 International (CC BY 4.0): LibriTTS-R is derived from LibriVox public-domain audiobook recordings and is distributed by its authors under that licence. No changes were made to the corpus; the voices are Piper models trained on it.

Vision and video

Stills are generated with Stable Diffusion XL under the CreativeML Open RAIL++-M licence; motion with LTX-Video under its Open Weights Licence; lip-sync with SadTalker (Apache-2.0). Each of those licences carries use restrictions that we pass on to anyone using this studio — see our Terms of Service.

Everything is generated

People shown in films made with this studio do not exist. Every film carries a visible mark saying so, and removing it breaks both our terms and the model licences above.