← All articles

Photo tips

Choosing better photos for an AI dance video

Prepare usable reference photos by checking light, framing, obstructions and file formats.

Written and maintained by Alors On Danse AI Video · Updated

Prioritize visible detail

Choose an evenly lit photo with readable facial features and a clear outline. Strong backlighting can hide the face, while heavy beauty filters can remove useful detail. Zoom into the original and inspect the eyes, nose and mouth. Enlarging a blurry image does not recover information that was never captured.

Keep one subject per upload

Do not ask the model to identify your intended person in a group shot. Leave space around the head when cropping, rather than cutting off the hair or chin. A portrait supplies facial information; a fuller view also supplies clothing and appearance cues. Neither guarantees exact preservation. Similar lighting and framing make the two references easier to compare.

Check obstructions and file type

Sunglasses, masks, hands and hair across the face reduce the available reference information. Prefer an unobstructed original over a repeatedly compressed screenshot. Supported files are JPG, PNG and WebP, up to 10 MB each. Export an actual conversion for formats such as HEIC; renaming the extension does not convert the image.

Compare before submitting

Confirm that each image shows the intended subject, contains no unintended private information and is authorized for use. To improve an unsatisfactory attempt, change the more problematic photo while keeping the other as a comparison. Clear inputs give the process better information, but do not guarantee identical faces or movement.

Keep reading