Log inSign up
FloyoAI profile banner
user avatar

FloyoAI

@FloyoAI
Where creative professionals, studios & creative teams collaborate, build, run, & scale complex visual AI pipelines.
floyo.ai
Joined November 2024
34
Following
603
Followers
RepliesArticlesMedia
  • user avatar
    FloyoAI
    @FloyoAI
    Aug 14
    6 models dropped in the last few weeks. Here's what each one is actually for. Swipe through the breakdown 🧵 MiniMax H3 · Seedance 2.5 · Qwen Image 3 · FLUX 3 · Reve 2.1 · JoyAI All six run on Floyo. Workflow links in comments. @MiniMax_AI, @BytedanceTalk, @Alibaba_Qwen,
    Image
  • user avatar
    FloyoAI
    @FloyoAI
    Aug 13
    Featuring creator: KatoriKonoe ✨ KatoriKonoe, a creator on Floyo, built Crystal-3K Advanced, a double-pass workflow for creating detailed, high-resolution visuals. It takes your base generations to 2048x3072 in about 35 seconds. See the workflow in action in the video below.
    Image
    00:00
  • user avatar
    FloyoAI
    @FloyoAI
    Aug 12
    We put MiniMax H3, Wan 2.2, and LTX 2.3 side by side again. Same prompts, same setup. H3 still wins on raw quality. Better motion, tighter prompt adherence, and it actually held up at 20 seconds. Asked for "dribble → crossover → shoot" and it followed the full sequence. The
    Image
    00:00
  • user avatar
    FloyoAI
    @FloyoAI
    Aug 10
    A rough sketch with a few labeled boxes comes back as a photorealistic room. Every object still where it was drawn. Most image models read a sketch as a suggestion and rearrange your layout. The Krea 2 Identity Edit LoRA holds the spatial relationships in place, so the
    Image
    00:00
  • user avatar
    FloyoAI
    @FloyoAI
    Aug 7
    We compared three open-weight video models on the same prompts: MiniMax H3, LTX 2.3, and Wan 2.2. MiniMax H3 has the best motion quality of the three. Camera movements feel natural instead of mechanical, especially in tracking and FPV shots. Hair, cloth, and physics all behave
    Image
    00:00