We made this 15-second VELUNE candle study from four AI-generated keyframes. The useful starting point was deciding what each shot needed to communicate.
Four shot roles in our 15-second study. The closing shot uses a still image with a local pull-back.
Here is the sequence:
- Introduce the product — 3.4 seconds. Show the whole jar and readable label. This hero image also becomes the reference for the other views.
- Show a detail — 3 seconds. Move close to the flame, wax and ceramic rim. The label is outside this crop; the detail needs to feel like the same object.
- Suggest the scent — 4.4 seconds. Bring in fig and cedar as visual context, while keeping the candle recognisable.
- Give the name time — 4.2 seconds. Return to a simple composition with room for the closing title.
Before animation, compare each still with the hero: jar proportions, rim shape, colour, label placement and lettering. Keep the product description fixed. Add advertising headlines as separate text layers in the edit, so a copy change does not require another image.
For motion, start with one camera instruction per shot: a gentle push-in, or a locked camera with subtle flame movement. Then inspect the result throughout the clip. “Preserve the label” is a request, not a guarantee.
Our closing animation failed twice. We used the approved still with a small local pull-back instead. Its flame stays static; the product and title remain stable. That fallback belongs in the plan too.
Would you keep the 4.2-second closing hold, or move one second back to the hero shot—and why?
This is an original educational example by Frame & Formula, using AI-generated visuals.