OpenAI "Sora" - Creating video from text
16.02.2024
Exciting news! Just yesterday, OpenAI announced the groundbreaking Sora model, a text-to-video model that can create realistic and imaginative scenes from text instructions.
With Sora, users can generate videos of up to a minute in length, with vivid imagery and faithful adherence to their prompts. The developers argue that the model understands not only what the user has asked for in the prompt, but also how those things exist in the physical world.
From a technical perspective, Sora combines diffusion models with the transformer architecture to obtain diffusion transformers. This allows training data to span across different durations, resolutions and aspect ratios.
Stay tuned for further updates, once the model becomes available to the general public I may be back with some examples 😉️. In the meantime, check out my video montage showcasing Sora's capabilities and have a look at the interesting technical report. Are you impressed?