The Videos tab makes short clips five ways. Image to video animates a still. Reference to video puts the people and products in your shot into a new scene. Text to video makes a clip from a prompt alone. Motion puts your persona or product into the movement of a reference video. Avatar makes a talking clip from a photo and an audio file you upload. The tab also changes the clips you have: edit, extend, lip-sync, dub, add sound to, reframe or upscale one, or take out its background or an object. Effects and templates turn your shot into a ready-made effect, an outfit try-on or an ad. Every clip is priced before you start.
Pick a still of the persona or product and one of 45 image-to-video models, among them Seedance 2.5, Seedance 2.5 Turbo, Seedance 2.0, Seedance 2.0 Mini, Kling 3.0 Std, Kling 3.0 Pro, Veo 3.1 Lite, Veo 3.1 Fast, WAN 2.7, Sora 2, LTX-2 Pro and Grok Imagine Video. Set the length, resolution and aspect where the model allows, add a negative prompt, sound or an end frame on the models that take them, and run up to four copies with automatic retries.
Build the shot as in Singles: a persona, products, clothing and a background, each from its photos. Pick one of 16 reference-to-video models and describe the scene; the clip keeps their look and belongs to the shot's lead. Each model takes 3 to 15 photos, depending on the model, and Start says when a shot sends more than the one you picked takes.
Yes. Pick one of your clips or upload one, then one of 63 models: edit it from a prompt, extend it, sync its lips to an audio track you upload, dub it into another language, add sound effects and music, reframe it to a new shape, upscale it, or take out its background or an object. Most of them bill per second of the clip or audio that goes in, so the price shows once they are picked, and the new clip belongs to the same persona as the old one.
Build the shot as in Singles and pick one of 10 effects and template models. Some put the lead's photo into a ready-made effect, from wings to Halloween looks, or an anime style. The outfit try-on dresses the lead in the shot's clothing, the ad models make a short ad with the shot's product, and the rest make a clip from a prompt with the shot's photos as references. The Kling effects bill per clip, priced by the effect; the rest bill per second.
Write the scene, the motion and the camera, and pick one of 36 text-to-video models. No image goes in, so the clip belongs to no persona; it lands in Review with your other clips. Length, resolution, aspect and sound follow what each model takes, and the price shows before you start.
Upload a reference video (mp4, mov or webm). First an image model puts your persona or product into its first frame, and you pick the frame you like or reroll it. Then Kling motion control animates that frame by following the movement in the source video, on one of 4 Kling models, keeping the original sound if you want it. Or tick Skip frame generation and the first reference photo is frame 1.
It is billed by the length of the source clip, at least 3 seconds and at most 30, plus the first-frame image unless you skip it.
Pick a photo of the persona, upload a voice-over (mp3, wav or m4a) and choose one of 13 talking-avatar models. The clip runs as long as your audio. There is no text-to-speech: you record or source the audio yourself.
Every video model is priced per second at the provider's list rate, and the total shows on Start before you run it, on your own WaveSpeed key. The model price pages list the rate for every model and resolution.
Yes. Add a product, then place it in a scene with a background, on its own or with one of your personas, who can wear a garment from your clothing library. A persona and a product in the same shot need Creator or above, because Solo holds one persona or product. Turn the best stills into video clips, or make a talking avatar clip from a photo and your own voice-over file.
It depends on the model. Image to video runs from 1 to 30 seconds, reference to video from 1 to 30, text to video from 1 to 30, motion control bills up to 30 seconds of the source, and a talking clip follows your audio, up to 10 minutes on the longest models.
No. There is no text-to-speech. Upload a voice-over you recorded or have the rights to, and the clip is animated to match it.