Before you begin
- A source video you can use and available credits.
- A supported clip range; the selected model shows its current limits.
What you’ll achieve:
Model choices
For unrestricted AI video object editing, choose a compatible Less restrictive model such as Wan 3.0 Prime Edit. Native written-direction models do not require subject detection; detection-based models have a different selection flow.
Upload your source video
Open Video Editor and choose Upload video or a saved clip. Wait for preparation and duration to finish. Choose a short range with a clear view of the subject so you can judge the change against the original.

Choose Add an object
Open What would you like to change? and choose Add an object. Identify a range where the target area is visible. This example uses a native text-directed edit, so adding an object does not require a detected subject mask.

Choose the editing model
Open Change model and compare Less restrictive models for an unrestricted editing workflow, or switch to Moderated if that suits your project. This example uses a native written-direction model. Check reference support, duration and resolution before continuing.

Explain the object and placement
Describe the item, its material and where it should appear. For example: add a small backpack on the floor beside the explorer; keep the person and camera movement. Add a reference image only if the selected model accepts it, and use its reference name in your direction when supported.

Review settings, price and the result
Open Settings to check the selected range, resolution, audio and whether to return the selected section or full video. Close Settings, then select Review price. Confirm the quoted edit when ready; inspect the completed result in My edits and download it while keeping your original source.

A useful tip
Describe contact with the scene: on the table, held in the left hand, or beside the chair. That is more useful than an object name without placement.
If something goes wrong
If an object appears in the wrong place, make a focused correction to its position and scale. A detection-based editing engine exposes different selection controls; read its requirements after switching models.
Read first, create when you’re ready
Reading this guide is free. The linked tool shows its own requirements and current credit price. Screenshots use demonstration data; review your own inputs and the quoted cost before submitting paid work.
Reviewed 2026-10-18 · Tong Academy