SwiftI2V: Efficient High-Resolution Image-to-Video Generation Method

SwiftI2V generates high-resolution videos from images via conditional segment-wise generation. It preserves fine details and realistic dynamics at 2K resolution, fixing issues in existing end-to-end models.

SwiftI2V introduces conditional segment-wise generation for image-to-video synthesis at 2K resolution. It divides video creation into manageable segments conditioned on the input image to preserve details and add motion. This tackles limitations of prior end-to-end models that falter on high-res outputs.