Golden retriever in a pickup: photo-to-video test
A golden retriever photo becomes a 15-second pickup-scene video with AI-created speech. See the source photo and the measured 496 × 864 MP4.
Test settings and measured results
- Model
- bytedance/seedance-2-mini
- Mode
- Scene mode · pickup
- Requested output
- 15 seconds · 480p · 9:16
- Measured MP4
- 15.104 seconds · 496 × 864 pixels · H.264 video + AAC audio
- Submission to saved result
- 153.579 seconds in this test; not a delivery-time guarantee.
- Recorded attempts
- 1 completed generation; 1 earlier failed submission(s).
- Historical provider usage
- 57 Kie provider credits in the October 19 test. These are provider units, not customer credits or a current dollar price.
Method and limitations
Both files were generated through the studio, archived to storage and downloaded through the account-protected endpoint. The downloaded bytes matched the archives. Playback decoded frames with no media error, and both files contain an audio track. Exact dialogue, voice similarity and lip-sync quality were not scored; play the result to assess them yourself.
The dog test also recorded an earlier submission that failed because an old upload URL was unreachable. The provider charged zero for that failed submission and the application returned its credits. The successful dog result was created after uploading a reachable photo. That is one observed failure, not an estimated failure rate.
Only these two pet results are documented here. Human generation is available for authorized adults, but there is no equivalent published human case study yet. These October 19 tests used an earlier pricing configuration; current customer prices and credit quotes are shown in Pricing and the studio.
File SHA-256
34cb2b1bee6385f8be7e205e2ff102f37101271a932f73902b1bb389f6c4a4de