Wan is one of our favorite models. It has come a long way over the first version. Wan2.5-Preview, which was announced last night, introduces some of the features we have come to love in Veo 3. Instead of treating text, images, audio, and video separately, Wan2.5 can produce cinematic videos with synced audio. By training on text, audio, and visuals together, it achieves much tighter alignment. It supports multi-person vocals, sound effects, and background music. Here is the prompt I used:
A bacon weave unfolds across parchment like a sushi mat. Gloved hands spread seasoned ground beef in a perfect rectangle—pepper jack sticks, jalapeño strips, prosciutto, and crispy onions layer down the middle in fast beats. Hands roll it tight; sauce brushes the seam. Smash-cut: a clean slice reveals the molten core. Hold a half-beat on the bacon-seared spiral.
This model give you the option to make 5 and 10 second videos. Wan 2.5 also has an amazing image model that follows instructions better and supports more art styles. It support instruction-based image editing. You can try this online at this time.
[HT]
