Detailed Introduction
Image to Video (I2V) Workflow for COSMOS Open-Source Model
What Does This Workflow Do?
This workflow takes your input image, adjusts it (crop/resize) to match the ideal render size for COSMOS, and then automatically generates a suitable prompt for COSMOS to work its magic! The process leverages Florence for automatic image description to streamline prompt creation.Key FeaturesOptimizations Included: The workflow incorporates Sage Attention and Torch Compile for enhanced performance.
How to Use
1️⃣ Load Your Image: Add your input image. Check the note next to Load Image for guidance on ideal size and other caveats.
2️⃣ Queue the Prompt: Once the image is loaded, simply queue the prompt.
3️⃣ Use Ultra or Ultra PRO Gpu for quick results.That’s it! All other settings are pre-configured and ready to go.Enjoy the MagicClick Queue and let COSMOS transform your image into a stunning video.
Have fun experimenting! 😉