Choose the original camera file when you can. Screenshots, social-media downloads, and repeatedly forwarded images can lose fine details around the eyes, mouth, and hairline. Zoom into the face: if it is already a blur, the AI will have to invent some of the missing information.

A larger file is not automatically a better reference. A sharp face in a modestly sized photo is usually more useful than a large photo where the person is tiny or out of focus.

Soft daylight from a window or an overcast outdoor setting makes facial structure easier to see. Avoid sunglasses, deep shadows across the eyes, strong beauty filters, and extreme wide-angle selfies. Keep ears, hairline, and the whole outline of the face in the image.

Choose an expression you actually want to recognize in the finished portrait. A formal template can change clothing and setting, but it should not be expected to reconstruct a completely hidden face or an unknown expression.

Check the subject count on the template detail page before uploading. One clear shared photograph can work for a couple; separate face photos may be better when one person is too small or turned away in the shared shot. The studio accepts up to four reference photos.

For a family, identify a clear reference for every person. Similar lighting and comparable sharpness make it easier to review the resulting faces together. More uploaded photos do not increase the number of people a template supports.

Inspect each face at a useful viewing size. Check eye shape, age, hair, expression, and any distinctive features. AI artwork is an interpretation, and a polished-looking scene can still contain an inaccurate likeness.

Only continue to a canvas and frame when you are comfortable with the actual preview. If the source photo is too unclear to support the result you want, select a better reference rather than relying on the print to fix the artwork.