Separate the source from the motion
Ask two questions of any preview. What kind of photo was this film made from — and what does the template do that the photo did not already contain? If the preview shows a studio-lit portrait in a plain room, the motion is the only information you are getting.
Four things a preview can tell you
- Camera behaviour — is the frame locked, pushing or orbiting? That carries to your result.
- Motion magnitude — small idle motion versus a full body movement. Magnitude is the main driver of artefacts.
- Duration and pacing — how much happens in the loop tells you what the template expects to fill.
- Wardrobe class — templates are often tuned to a garment type; lingerie models behave differently from streetwear models.
What a preview cannot tell you
It cannot tell you how the template behaves on a face with different bone structure, under different lighting, or at a different crop. Those three variables explain most of the gap between a preview and a personal result.
The cheapest valid test
Pick the template on the shortest available render, apply it to the exact photo you intend to use, and inspect it. Two minutes of your own footage beats twenty preview loops.