WHAT IT DOES
Features
in detail.
- Processing an existing video
- One replacement face per task
- Photograph as an appearance reference
- Automatic selection of the person in frame
Start with a suitable pair of files
DeepFace Video has a narrow task: it takes an existing video and transfers a face from a photograph into it. It does not create a story from a description or shoot a new video from scratch. The main choice therefore happens before processing. Choose a scene whose source video already contains suitable movement, clothing, background and actions.
Choose a photograph with a distinct face that is not heavily obscured by hair or accessories. The video benefits from a clear angle and a sufficiently large face. Abrupt turns, hands covering the face and severe blur complicate transfer. Choose a segment with a sharp face and even lighting.
Check whom the service will select
YesAI's current description specifies one replaced face per task. The service selects the first person to appear; if several appear together, it uses the leftmost. This is not manual selection of any character through text. Before sending, watch the video's beginning and check that this rule matches your plan.
'Replace the face of the person on the right' does not override automatic selection. If the desired character does not fit the rules, find another source segment or prepare suitable material beforehand. This service form does not allow manual replacement-area selection or replacing several faces at once.
Evaluate the result in motion
One successful frame is not enough. Watch the entire video, especially head turns, lighting changes and partial face occlusion. Examine the hairline, jaw, expressions and transitions around the face. YesAI's description says the tool preserves the source duration, frame rate and sound; after processing, check that the finished file suits your intended use.
You can plan the idea here, but DeepFace Video in YesAI does not use text as its operational prompt. The button opens the tool, where video and photo uploads are required. Your scene note helps prepare sources and stays here. You select and send the files in YesAI.
Suitable tasks
- Creative video featuring you
- Coherent video portrait
- Experimenting with a staged scene
- Preparing material with suitable source movement
Things to consider
- An MP4 up to 50MB and a JPG, JPEG or PNG photograph up to 5MB are required.
- When several people are present, the person is selected automatically under YesAI's rules.
- The text field does not replace source files or control generation of a new video.
- When working with another person's appearance, use material with their consent.
FROM AN EXAMPLE TO YOUR OWN IDEA
How to prepare for processing
Prepare files for one of the workflows.
Checking the source video
Prepare your own short staged video: one person appears first, their face visible from the front without occlusion. Check the beginning, middle and end for abrupt turns and blur.
Choosing a photograph
Choose a photograph of your face with even lighting, good sharpness and no occlusion. Compare its angle with the video's main frames. Ensure formats and file sizes meet YesAI's requirements.
Workflow
- 01
Choose a video where the desired person fits the automatic selection rule.
- 02
Open DeepFace Video in YesAI and upload the video and photograph.
- 03
Check the face in motion and the finished file's integrity.
A FEW MORE ANSWERS
Questions about DeepFace Video
Can I start face replacement with text alone?
No. An existing MP4 and a face photograph are needed. A scene idea helps preparation but is not a prompt for this tool.
How is a person chosen when several are present?
YesAI selects automatically: the first person to appear, or the leftmost if several appear together. Text instructions do not change this rule.
How many faces are replaced at once?
The current service description specifies one face per task. Choose the source video with that limit in mind.
Which files should I prepare?
Use an MP4 video up to 50MB and a face photograph in JPG, JPEG or PNG up to 5MB. Before uploading, check that the face is clearly visible in both.
