Wan 3.0 AI video generator

Choose text to describe a new scene, or image mode to provide visual material. Image ingredients accept up to ten pictures; frames mode uses a first image and an optional last image.

Use ingredients for subject and setting guidance, or frames when the planned beginning and ending matter. This workflow accepts images rather than video, audio or document references.

Select a fixed duration from 2 to 30 seconds and 480P, 720P or 1080P in the workspace. Choose enough time for the action to be readable without packing several scenes into one clip.

Duration is chosen before submission; automatic duration is unavailable here. Changing the length or resolution changes the quote, so check it again before creating.

Use a supported ratio shown in the current input mode and plan the composition for the destination. Give the subject room to move and leave space for any titles you will add later.

Check your reference crop as well as the output setting. A wide source with important details at the edges can become difficult to use in a vertical edit.

Wan 3.0 provides an audio option. Select it when sound belongs in the shot and describe the relevant atmosphere or sound event alongside the visible action.

Watch and listen to the completed clip before using it. Check timing, unwanted sounds and any dialogue, then make the final mix in your editing tool.

Choose frames mode, add the first image and add a last image if you need to guide the ending. Explain the movement that connects the two visual states.

Keep identity, lighting and composition reasonably compatible between the images. An ending reference gives direction; inspect the actual final frame rather than assuming an exact match.

Name the subject, one action, the setting and a useful camera direction. For example, ask a traveller to turn toward a station window while the camera moves slowly closer.

With images attached, focus on what changes and what should remain recognisable. Avoid conflicting camera moves or several sequential scenes in one short request.

Open Wan 3.0 Prime to explore its separate interpretation with similar input controls. Keep the brief and chosen settings comparable, then check Prime’s own quote.

HappyHorse 1.1 offers text or one image for 3–15-second clips; Reference accepts up to nine images. Choose according to the inputs and controls your scene needs, rather than the model name alone.

Use clear images with a visible subject and deliberate crop. For ingredients, assign distinct roles such as character, clothing and location, and remove unrelated or conflicting pictures.

For frames mode, check both endpoints at full size. Prepare or edit a source image in the image workspace if important details are missing before adding it to video creation.

Read the current credit quote after choosing the model, duration and resolution. Confirm the final settings before submitting instead of assuming that an earlier quote still applies.

If a task is pending or fails, open its record in Assets and inspect the status and available message. Keep the task ID for support and check the original record before submitting another task.

Watch the whole clip, including its ending. Inspect faces, hands, product details, text, movement and audio where selected; a good opening frame alone is not enough.

Confirm that you may use the input material and the intended content. Finish cropping, titles, sound and timing in your editor, then review the exported version for its publishing channel.