Lightricks (LTXV), an audio-visual model that does text to video or image to video, complete with matching audio. It’s uncanny as hell and gives movements and expressions that can best be described as “imagine a pod person poorly imitating the overacting from game shows and commercials”, but is still a step up from any other video generation I’ve seen including WAN. Also it’s open source and runs locally, although I’m not sure what hardware is actually required to make it work at all; I’ve heard it can be squeezed onto even older gaming cards and still run albeit slowly.
I hate it so much, despite also getting enjoyment out of seeing closed corporate models eating shit to lighter weight open models. This is something that shouldn’t exist and which makes absolute garbage output, but it is still funny that it’s beating Sam Altman’s infinite fountain of grift.
And they’re losing out to things like LTX, which is fucking horrifying stuff that can run on just normal consumer hardware.
What is LTX?
Lightricks (LTXV), an audio-visual model that does text to video or image to video, complete with matching audio. It’s uncanny as hell and gives movements and expressions that can best be described as “imagine a pod person poorly imitating the overacting from game shows and commercials”, but is still a step up from any other video generation I’ve seen including WAN. Also it’s open source and runs locally, although I’m not sure what hardware is actually required to make it work at all; I’ve heard it can be squeezed onto even older gaming cards and still run albeit slowly.
I hate it so much, despite also getting enjoyment out of seeing closed corporate models eating shit to lighter weight open models. This is something that shouldn’t exist and which makes absolute garbage output, but it is still funny that it’s beating Sam Altman’s infinite fountain of grift.