How to Choose Photos for AI Performance Videos

A useful source photo makes a person's features easy to see. Before starting a photo-based performance video, spend a few minutes checking the two images together. These checks help you prepare clear inputs without promising a particular generation result.

Use soft, even lighting

Choose a photo taken near a window or in another evenly lit setting. Avoid strong shadows across the eyes and harsh backlighting that leaves the face dark. Natural skin detail is more useful than a heavy beauty filter.

Keep the face visible

A front-facing or slightly turned head is a practical starting point. Avoid sunglasses, hands covering the face, motion blur, and tightly cropped foreheads or chins. If possible, use one clearly identifiable person per photo so there is no ambiguity about the intended subject.

Check image quality and framing

Use the original image when possible instead of a compressed screenshot. Make sure the face remains clear when viewed at a comfortable size. Cropping can remove distractions, but keep the full head visible and do not enlarge a tiny image expecting it to recover missing detail.

Follow the file or URL requirements shown in the workspace. If it asks for an image URL, use a link that actually serves the intended image and that you have permission to share. Avoid exposing private account links or access tokens.

Keep the pair consistent

Compare the images side by side. Similar lighting and clear faces make it easier to evaluate the pair. Name or organize the files by their intended left and right positions to avoid swapping the cast accidentally.

Check consent and intended use

Ask both people whether they agree to the edit and where it will be shared. Permission to send a photo privately is not necessarily permission to publish it or use it in an advertisement. Only use source media, music, and other assets for which you hold the necessary rights.

Final checklist

• Both faces are visible and in focus.
• Lighting is even enough to see the main facial features.
• There is one intended subject in each image.
• Left and right positions are clear.
• The input meets the requirements currently shown in the product.
• The people and rights holders have given the permissions needed for the intended use.

After processing, review the complete video for visual errors and label AI-edited content when needed to avoid confusion. For input or task questions, email support@hotellobbyai.info. New to the concept? Read What Is Hotel Lobby AI?.