H3 with Pika's 6-step LoRA makes a 5 s 1344x768 clip with sound in 86 s on my RTX 5090.…
This is a AI post classified by AGK Intelligence as AI video (a model release), kept by the AI Radar because it carries real work, not commentary.
🔊 SOUND ON 🔊 H3 with Pika's 6-step LoRA makes a 5 s 1344x768 clip with sound in 86 s on my RTX 5090. Base H3 needs 269 s. Four video models, same five prompts, seed 42, all with sound. The video plays prompt 2 with sound, then prompts 4 and 5 side by side. RTX 5090 (32 GB), 5 s clip, median of 5: H3 + Pika-H3-Flex: 86.2 s H3 base: 268.7 s Kandinsky 6 Pro: 237.1 s LTX-2.5: 27.8 s H3 and Pika: ComfyUI 0.44.0, the Comfy-Org templates, Pika 6 steps, base 20 steps LTX-2.5: 22B distilled, two stages, my existing ComfyUI workflow Kandinsky 6 Pro: its own pipeline, pro-distill, block offload, beau
Posted by Niklas (2.8k followers) 1 days ago · 9 likes · 652 views · view the original post on X. Kept by the AI Radar as AI video.
More AI work like this
- Ad teams don't all start from the same place, and we built Boreal-H3 with that in mind. — @Creatify_Labs
- holy shit i can make gassy bowser at home — @cheatyyyy
- This AI video proves that Marvel isn't the only one that can make quality movies — @ZentrixHQ
- Free FLUX 3 Max generations are available on fal through November 9. — @fal
- 有点炸裂，Cascadeur把AI动画的token计费直接干没了🤯 — @AYi_AInotes
- Want to recast EVERY character in a video? — @eachlabs
- wtffff we're VECTORIZING now😭 this is not AI slop, it's literally just letting you edit… — @ForwardEditor
- Your ideas don't wait. Why should your videos? — @wavespeed_ai
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 65.3k posts from 5k X accounts over the last 14 days, 10.4k tools, 19 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 17:44 UTC. Full method.