16GB GPUs: Enough for Text-to-Video AI?

Simple Image-to-Video and Text-to-Video generations are within the reach of 16GB VRAM GPUs. I test two Pytorch code examples for you, and run them on a free Google COLAB T4 GPU (15GB).

Video generation models as world simulators? Or just learning from millions of video game sequences?

Text-to-image diffusion models as generative video models are able to generate short clips of video from a text prompt or an initial image.

OpenAI's SORA and Gemini 1.5 PRO with 1 mio prompt …

code colab diffusion diffusion models examples free game generative generative video google gpu gpus image image diffusion image-to-video pytorch simple test text text-to-image text-to-video them video video ai video game video generation world