WAN 3.0: native 30-second AI video generation

30 seconds, in one continuous take. The more useful part is that you no longer have to pick the number: describe the action and smart duration reads the pacing you implied, then sets the length to match. Because the clip runs unbroken, light, motion and pacing hold all the way through instead of resetting at a splice. Need longer? Extend any finished clip and keep building from a take that already works.

The 4 references you already use are all here - text, image, audio, video. WAN 3.0 adds two more. Drop in a document (.doc, .pdf, .ppt, .xls) and it reads the thing. Or just give it a URL: a product page, an article, a research paper, your own site. The spec sheet becomes the ad film. The landing page becomes the launch video. Long, multi-part instructions get parsed properly too, so a detailed request survives all the way to the final frame.

Put words on screen and WAN 3.0 renders them legibly and accurately, as text you can actually read - which counts for most in busy, information-heavy scenes, where there is the most to get wrong. Detail sits closer to real footage throughout. References hold at pixel level, so characters, objects, scenes, styles and audio all stay themselves across the sequence, and motion and emotion carry more range. Pin the first and last frame, and adaptive ratio shapes everything in between.

You can find WAN 3.0 in Picsart's AI Playground, where a single prompt runs against 150+ other models at once and you see the difference before committing. It generates at 1080p, at roughly one to two seconds of render per second of finished video, with every input type in the same panel. WAN 2.7 sits alongside it: 15 seconds in fixed blocks of 5, 10 or 15, up to 5 reference images, and it still holds the 4K advantage. Pick 3.0 for length, input range and readable text. Pick 2.7 when resolution matters most.