Create Jean Phil AI videos from a photo
One authorized photo, a movement reference and a realistic quality check.
Prepare one image
Use your own photo, a consenting adult’s photo, or an authorized fictional character. Choose a clear face with space around the body. Jean Phil uses one image; no second portrait or ending frame is required.
Set up the generation
Choose Character image in the studio, upload the image and select a movement template. Preview the reference, then review duration, quality and format. The reference guides motion; it is not a promise to reproduce the clip exactly.
Check cost and identity
For a 5-second video, 576P costs 45 credits and 768P costs 75. Inspect the finished face, hands and clothing before sharing. Our real tests produced playable videos but changed the input appearance noticeably. Label shared results as AI-generated.
Follow the controls in order
Sign in, choose Character image, select your authorized JPG, PNG or WebP, then choose Solo room dance or another available template. The photo thumbnail is a local preview until you submit. Expand Video settings: the reference length and your output length are separate. A reference labeled 11s does not prevent choosing a 5s output. Check the consent box only when its statement is true. The final button shows the credit cost before submission.
What the two published tests actually show
Both tests used the same AI-created fictional adult input and the Solo room dance reference. They are two outputs of one template, not proof of several independently tested templates. The 768P test cost 75 credits and returned 768 × 1344; the 576P test cost 45 credits and returned 576 × 1024. These are observed pixel dimensions, not a promise that every quality/format combination returns those dimensions.
Why the hairstyle and clothing changed
The current photo-mode preset asks to preserve facial identity while also adding a blond bob, curled moustache and houndstooth suit. Those instructions deliberately change appearance. It also gives the video reference responsibility for motion, framing and setting. This conflict is one reason to separate style transformation from identity preservation; it does not establish the sole cause of facial drift. An identity-first prompt is being prepared for a controlled test, and has not yet been validated.
After submission: wait, review, download
On October 20 the provider accepted the task at 08:06:50 UTC, started rendering at 08:13:52 and completed it at 08:18:01. That was 7m 02s in queue plus 4m 09s rendering. Polling paused during the run; resuming the existing task recovered the result without a second generation. Once complete, play the full clip, inspect the face and hands, and use Download video. Keep the task ID if you need support.
The actual settings panel
One template, two real outputs
Both used this AI-created fictional adult input. These tests demonstrate delivery, not accurate identity replacement or a resolution benchmark.
| Stage | Time | Elapsed |
|---|---|---|
| Accepted → rendering | 08:06:50 → 08:13:52 | 7m 02s |
| Rendering → complete | 08:13:52 → 08:18:01 | 4m 09s |
| Total | 08:06:50 → 08:18:01 | 11m 11s |