Choose the best photo for your Bulin 47 AI video

Start with one adult and one visible face

Use a photo of yourself or another adult who has agreed to appear in an AI-generated video. The studio is designed for one lead performer. If two faces are equally prominent, the model has no reliable way to know which one should become the lead. Crop the image to one person before uploading. Do not use a photo of a minor or someone who has not given permission.

Choose a recent photo where the eyes, nose, mouth, hairline and face outline are visible. A front-facing or slightly turned head is easier for a model to read than a distant profile. Avoid sunglasses, masks, motion blur and strong beauty filters: each removes or changes identity cues before generation begins.

Frame the face and some upper body

A tight face crop may give the model strong facial detail but little guidance for shoulders, clothing and posture. A very distant full-body photo has the opposite problem: the face may occupy too few pixels. For this 9:16 performance, a head-and-shoulders or half-body portrait is a useful middle ground. Leave a little space around the head so the face is not cut off at the edge.

Look at the image at normal phone size. If you cannot quickly recognize the person, the video model may struggle too. A visible upper body gives it clues about clothing color and silhouette, although the generated outfit and movement can still change. Hands do not have to be visible, and seeing them does not guarantee accurate hands in the result.

Use even light, then check the file

Soft light from the front or side makes facial features easier to distinguish. A bright window directly behind the subject can leave the face nearly black; harsh shadows can hide one side of it. You do not need studio equipment. A sharp phone photo taken near a window or outdoors in open shade is often easier to inspect than a heavily edited portrait.

The upload accepts JPG, PNG or WebP up to 10 MB. If your phone saved a HEIC or HEIF image, export or convert it to JPG before choosing it in the studio. If the file is larger than 10 MB, resize or compress a copy while keeping the face sharp. Do not send a screenshot of a tiny thumbnail if you still have the original photo.

What the photo can and cannot control

Your portrait guides the lead performer's appearance. The licensed reference clip guides the scene and movement, while the model generates new video frames. It is not a frame-by-frame face swap. Clothing, hands, background details and face shape can drift, especially during fast movement. A better source photo improves the information available to the model; it cannot guarantee exact likeness or a particular outcome.

Before you create, preview the selected image in the studio and check that it is the intended adult. Each confirmed creation makes one video and uses one credit. If you want to understand the full sequence, read the step-by-step Bulin 47 AI video guide.

Quick portrait check

• One consenting adult is clearly visible.
• The face is sharp, well lit and unobstructed.
• The head and upper body have enough room in the frame.
• The file is JPG, PNG or WebP and no larger than 10 MB.
• You understand that the finished video may differ from the photo.

If your photo is accepted but the job later fails or remains pending, see what to check when a Bulin 47 AI video is not working before starting a second task.