Image-to-video workflows move a still into a new context. That is a creative act, but it is also a reuse of someone else's image if you did not make it. The checklist below is not legal advice; it is the practical minimum we use before a shot goes into a published project.
Before you upload a reference image
- Do you own the image, or have a licence that allows derivative and AI-assisted use?
- If a person is identifiable, do you have their permission for this specific use?
- Is there a trademark, logo, product label or artwork in the frame that you should remove?
- Is the intended use commercial, and does your licence distinguish that from personal use?
- Could the finished clip be mistaken for a real event, a real endorsement or a real person speaking?
When in doubt, generate your own still or photograph your own subject. It is faster than clearing rights afterwards, and it gives you a cleaner first frame — which, as the opening frame checklist explains, is the biggest quality lever you have.
Disclosure that costs nothing
A short line in the description or end card — visuals created with AI assistance — keeps the audience on side and satisfies most platform policies. Put it in editable text, not inside the generated frame. Keep it with the project file so a colleague can find it six months later.
Keep a small project record
| Keep | Why |
|---|---|
| Source still and its provenance | Proves what you had the right to use |
| The six-line prompt | Makes the result reproducible |
| Model / settings used | Explains why a later render differs |
| The one failure you fixed | Turns a retry into a reusable rule |
| Disclosure wording | Keeps publishing consistent across channels |
None of this makes the clip worse. In practice, projects with a clean source and a written record iterate faster, because nobody has to guess which still produced which version.