How to Use Video to Video AI
Video to Video AI creates a new version of existing footage. Start with one clear shot, separate what should stay from what should change, then inspect the whole output. This guide covers source preparation, four prompt patterns, and practical fixes. The cover is a workflow illustration.
Choose the workflow that matches your source
The input should match the job. Editing a clip, borrowing its movement, and animating a still image are different starting points.
Prepare the source and remove prompt conflicts
Before adding more instructions, check whether the source clearly shows the result you want to preserve.
Choose a continuous shot
Prefer one readable action without hard cuts, heavy blur, or a subject disappearing behind foreground objects. Trim longer footage into separate shots in your video editor before upload.
Use a preserve-and-change formula
Write: source + details to preserve + one requested change + visual constraints. For example, keep the walking direction and camera position; change the scene to dusk with warm window light.
Give supporting images one purpose
If you add an image reference, say whether it guides style, a prop, or appearance. Avoid asking it to redefine the camera or movement already supplied by the source clip.
Reduce flicker and texture changes
Choose a stable source and simplify the requested texture. Remove conflicting lighting descriptions and avoid combining several styles. Compare the same moments in the source and edited clip.
Diagnose shape or identity drift
Check where the face, hand, or product becomes hidden or blurred. Use a clearer source segment, name the essential details, and reduce the scope of the transformation.
Separate camera and subject movement
If the source pans left while your prompt asks for a rightward orbit, the directions conflict. Keep the original camera instruction explicit and change only the visual treatment first.
A four-step video editing workflow
Keep the same source and a record of each prompt so that successive versions are comparable.
Inspect and upload the source
Watch the clip once before uploading. Identify its action, camera movement, framing, and the detail that must remain recognizable. Open Video to Video and add that single source clip.
Write one observable edit
Use the source's displayed @ label. List two or three details to retain and one requested change. Avoid vague instructions such as make everything better or change the whole scene but keep everything identical.
Confirm the editing mode
Review the selected model and accepted input requirements. In this source-editing mode, duration and framing follow the input handling of the model; do not plan around controls that the editor does not expose.
Review and change one variable
Compare the opening, middle, and ending, then play the full clip for timing and flicker. If the output fails, change the source segment or one prompt instruction, not both at once.
Four Video to Video prompt templates
Replace @Video1 with the label shown for your upload. These are starting instructions to adapt, not tested output guarantees.
Subtle relighting
Edit @Video1. Keep the dancer's movement, terracotta outfit, camera position, and studio layout. Change only the lighting to soft moonlight with a warm rim. Keep skin tones natural and shadows consistent throughout the shot.
Use when movement already works and the main goal is a different mood.
Painted video restyle
Restyle @Video1 as hand-painted gouache animation with a sea-teal and warm ochre palette. Retain the traveler's walking direction, rust coat, shoulder bag, coastal path, and camera framing. Use consistent brush texture across the clip.
Specify one medium and a small palette instead of mixing unrelated style names.
Environment change
Edit @Video1. Keep the cobalt teapot, walnut handle, spout shape, stone plinth, and original camera movement. Replace the studio background with a misty woodland at dawn. Match the new light to the teapot and keep it grounded on the plinth.
Inspect edges, reflections, and contact shadows rather than only the background.
Product variation
Edit @Video1. Keep the shoe's silhouette, sole, stitching, position, and the original slow camera move. Change only the upper fabric from gray to muted forest green. Retain natural material texture and do not add text or new accessories.
A narrow edit makes unintended geometry changes easier to spot.
Choose the right AI video workflow
Move from preparation to generation, compare a simpler input method, or explore related workflows.
Video to Video AI Generator
Transform an existing video with AI. Upload a source clip, describe what to keep or change, and explore new styles, lighting, and environments.
How to Make a Photo Dance with AI
Learn how to make a photo dance with AI: choose a clear image, describe a short routine or add a motion reference, and use three practical dance prompts.
How to Extend a Video with AI
Learn how to extend videos with AI: prepare a source clip, write a continuation prompt, choose settings, and review the result with practical examples.
How to Use Reference to Video AI
Learn how to use Reference to Video AI with image, video, and audio assets, clear @ roles, four workflow steps, copyable prompts, and troubleshooting.
Image to Video AI Generator
Animate one image when you only need a starting frame and motion prompt.
Text to Video AI Generator
Turn a written prompt into an AI video. Describe one scene, choose a video model, aspect ratio, resolution, and duration, then carry your prompt into LumiYing to generate and refine the clip.
AI Models
Review the image and video models currently available in LumiYing.
Pricing
View the available plans for different creation workflows.
Video to Video workflow questions
How do I change the style of a video?
Upload one clear source clip, specify the medium and palette you want, and name the action and framing to retain. Generate, compare the full clip, and refine one instruction.
What makes a good source video?
One continuous shot with a readable subject and controlled movement is easier to direct. Avoid heavy blur, abrupt cuts, and long occlusions when important details must remain stable.
What should a video editing prompt include?
Identify the source, describe what to preserve, request one visible change, and state the most important constraints. Keep camera movement separate from the subject's action.
Why does the edited video flicker?
Unstable source detail, complicated textures, or conflicting style and lighting instructions can contribute. Simplify the edit, use a cleaner source, and compare matching moments to locate the problem.
Why does a face or product change shape?
Blur, occlusion, and large transformations make details harder to retain. Try a clearer segment, reduce competing references, and narrow the edit. Exact identity or geometry is not guaranteed.
Can I edit a long video in one generation?
Use the current duration limit shown in the uploader. For longer work, prepare separate shots, edit and review them individually, then assemble the accepted clips in your video editor.
When should I use Reference to Video instead?
Choose it when the source clip should supply only motion or camera guidance for a new scene. Use Video to Video when the existing footage itself is the shot you want to edit.
Try one clear change on your own footage
Open the tool with a prepared clip and a short preserve-and-change prompt.