Fair-fight video test: HappyHorse 1.1 beats Kling 3.0 on character consistency under identical prompts
kimmonismus · x · 2026-09-04
The author argues most AI video model "comparisons" are invalid: different prompts, tweaked settings, edited-out failures. He tested HappyHorse 1.1 and Kling 3.0 properly — same prompt, same reference images, same duration, no edits — across five dimensions:
- Lip-sync & speech: judged side-by-side on mouth-word matching, timing, and close-up expressions.
- Character & scene consistency: HappyHorse 1.1 was noticeably more consistent. In a crash scene, HappyHorse kept the injured character reacting — raised hand, visible blood, a living expression — while Kling 3.0's character lay motionless with no reaction.
- Complex motion: continuous sports/dance/action shots to expose whether a model understands weight, momentum, and balance.
- Camera control: identical timestamped storyboards (push in, orbit, crane up, pull out) to see which model loses the subject mid-move — the difference between a usable shot and a fifth regeneration.
- Price & workflow: prices only matter when compared like-for-like on output, duration, quality, and resolution.
He invites others to rerun the test and share results.
Related event: Blogger Slams Unfair AI Video Model Comparisons, Advocates Rigorous Testing(2 posts)→
More from Multimodal
- Early user: Seedance 2.5 delivers effortlessly cinematic shots, a favorite gen AI model — heypearlai · 2026-09-04
- Minimax-h3-Turbo ships FL2V Turbo 4-step v1.2 at 768p — Any_Fee5299 · 2026-09-04
- Local Dune skit on a 4090: character sheet plus voice reference workflow — r0ni · 2026-09-04
- Using Qwen3-VL-4B to auto-expand prompts for Minimax H3 video generation — CountFloyd_ · 2026-09-04
- Z-Image Base prompting experiment: natural-language scene blocks beat tag lists — Maleficent-Bowl-4841 · 2026-09-04
- Video generation isn't solved: fal h3 costs $22/hour vs $0.24 mobile gamer spend, 100x gap remains — IndraVahan · 2026-09-04