Wan 3.0

Turn photos, clips and sound into scenes worth sharing.

Enter your idea to generate
Wan 3.0

Key Features of Wan 3.0

Create scenes from text and references, set the opening and ending frames, or edit an existing video with native sound.

Story direction

Give the story room to unfold

Build a product reveal, social ad or multi-shot story in a single 2–30 second generation. Describe the action, camera movement and timing so each beat leads into the next. Native audio is generated with the picture.

Multimodal references

Make references do different jobs

Use a portrait for the character, a clip for movement and an audio sample for rhythm or voice. Explain what each reference should guide. This is useful for product campaigns, recurring characters and actions that are hard to describe with words alone.

First and last frames

Choose how the shot opens and closes

Animate an opening composition or supply both endpoints for a planned transition. A closed product box can lead into a reveal, or two storyboard images can guide a continuous shot. Use the frame workflow when the starting and ending compositions matter.

Native video editing

Change a scene or continue its action

Upload a source clip and describe the changes you want in clothing, background, lighting or visual treatment. Choose Video Edit to work on the existing scene, or Video Extend to generate what happens after its ending. The original clip goes directly to the model.

Native audio

Direct the picture and sound together

Describe dialogue, music, atmosphere and sound effects alongside the visible action. Audio references can guide the scene as well. Sound is enabled by default. Use the audio switch when you want a silent result. Review the picture and soundtrack together before choosing a take.

Wan 3.0 specs and credit cost

The settings you can choose when you generate with Wan 3.0 on LumiYing, and what each option costs. The model's maker may offer other settings on its own platforms.

Wan 3.0 generation settings on LumiYing
OutputVideo
Resolution480p, 720p, and 1080p
Length2–30 seconds
Aspect ratiosAuto, 16:9, 9:16, 4:3, 3:4, and 1:1
Ways to guide itText prompt, Reference files, First and last frame, Extend a video, and Edit a video
Reference limits10 images, 5 videos, and 5 audio clips
Credits per second of video
ResolutionStandardWith video reference
480p44
720p7.57.5
1080p14.514.5

Example: a 5-second 480p Wan 3.0 video costs 20 credits.

With a video reference, credits are charged for the generated length plus the total length of the input videos.

See credit packs and plans

How to Use Wan 3.0

01

Choose a starting point

Select Wan 3.0. Begin with a prompt, add image, video or audio references, or switch to first and last frames. For an existing clip, choose Video Edit or Video Extend.

02

Direct the action and output

Describe the subject, motion, camera and sound. Choose 480P, 720P or 1080P and a 2–30 second output. With reference video, the input and output must fit within 30 seconds in total.

03

Check credits and compare takes

Review the displayed credit estimate, generate, then check movement, continuity and sound. Refine one part of the prompt or references at a time so you can see what changed.

FAQ About Wan 3.0

We've answered the most frequently asked questions

Prime offers faster end-to-end generation with the same core creative capabilities. Both support the same resolutions, reference types, frame control and native editing workflows here. Standard costs fewer credits per second. Prime is useful when faster iteration matters.

Use up to 10 images, 5 video clips and 5 audio clips. Each video or audio clip must be 1–15 seconds. Videos can total up to 15 seconds, and audio has its own 15-second total. Images, video and audio can be combined in reference mode.

Choose 2–30 seconds at 480P, 720P or 1080P. Available ratios are 16:9, 9:16, 4:3, 3:4, 1:1 and adaptive. Video input plus output cannot exceed 30 seconds. Frame-based generation and native editing or continuation follow the source composition.

References guide appearance, subjects, settings and motion while allowing a new composition. First and last frames define the endpoints of the shot. Pick the workflow that fits your goal. Frame inputs and mixed reference media are separate modes.

Yes. Video Edit accepts one 2–15 second source clip and follows its duration while applying your requested changes. Video Extend uses a source clip and your continuation direction to generate the next scene. These workflows use the model’s native video inputs.

Yes. Native sound is enabled by default, including dialogue, music, ambience and effects described in the prompt. Audio references are supported, and you can switch generated audio off. The audio switch does not change the per-second credit rate.

The per-second rates are 480P 4 / 720P 7.5 / 1080P 14.5 credits. Ordinary generation and video-reference workflows use the same rates. Reference-video billing includes input plus output seconds. Native editing uses twice the source duration rounded up to whole seconds. Check the full estimate before generating.

Start with a 2-second text or image generation at 480P for 8 credits. Test a single action and a simple camera move first. Add video references when you need their movement, since input-video seconds contribute to the total credit estimate.

Put your references into motion.

Choose a product, character or source clip and direct the next scene with Wan 3.0.