THE PHOTO GUIDE · UPDATED OCTOBER 13, 2026

Your photo is the starting point.

A fixed video template removes a lot of setup, but it cannot make every source portrait equally useful. A clear face gives the video model a stronger identity reference. This guide helps you choose a photo before you spend time or money on a render.

You can prepare and preview your photo here now. Check the studio for generation availability, and none of the scene illustrations on this site are results generated from a visitor's uploaded photo.

What makes a useful portrait?

• One clear face. Choose a single-subject photo rather than a crowd or a face occupying a tiny corner.
• Even lighting. Soft daylight or a well-lit room makes facial details more readable than harsh shadows.
• Visible features. Keep eyes, brows, nose, and mouth unobstructed. Remove sunglasses or face coverings.
• A simple expression. A neutral expression or relaxed smile avoids adding unnecessary ambiguity.
• Enough detail. Start with the original image rather than a repeatedly compressed screenshot.

What the current upload preview checks.

The browser accepts JPG, PNG, and WebP up to 10 MB, checks that the image can be decoded, and requires 301–5999 pixels on each side and an aspect ratio strictly between 0.4 and 2.5. That is only a basic input check. It does not detect how many faces are present or predict likeness quality. A sharper original image is generally a better starting point than the minimum accepted size.

Select original virtual adult characters only. Previews stay in page memory until you click Generate, which sends the selected cast directly to the video provider. You can add one lead and up to five individual background portraits.

What the reference video controls.

In the planned workflow, the fixed reference clip supplies the party setting and the lead's actions. Your photo supplies identity. You will not be choosing between the people in our illustration gallery: those are original concept images showing the creative direction, not selectable templates.

Whether clothing, facial expressions, background details, and camera motion stay consistent depends on the model and the tested template. We will not guarantee an exact frame-by-frame copy from a single portrait.

Where AI video can struggle.

A hand passing over a face, a silver cup near the mouth, a fast turn, or another guest moving across the lead can make identity preservation harder. That is why the full video needs review, rather than judging one attractive frame.

If a result drifts, changing the portrait may help, but it is not always the cause. The reference action, occlusion, camera movement, and model settings can also affect the outcome.

How music fits into the result.

The model is instructed to use the reference clip’s background music and sound, retaining rhythm and audiovisual synchronization. You receive its output as returned, without separate audio replacement. Review music quality and completeness before sharing.