Text-to-video, 256×256, 16 frames @ 8 fps (about 4 s of scene). Same architecture as ZYVIDEO 1.0 (built on ZYI-256 Flash), trained further to move more: only clips with real motion, frames sampled at 4 fps, and an extra loss weight on what changes between frames. Average motion of the samples (mean frame difference, 0-255): 10.59 with 1.0 → 10.77 with 1.1.
| Parameters | 490M (343M spatial from ZYI-256 Flash + 147M temporal) |
| Latent | DC-AE f32c32 per frame — 16 × 8 × 8 × 32 (1024 tokens) |
| Text | flan-T5-base (frozen), English prompts |
| Objective | rectified flow, logit-normal t with shift 2.0, + 1× loss on frame-to-frame differences |
| Data | OpenVid-1M (streamed, filter "motion"), 182,869 unique clips |
| Training | +16,285 steps × batch 32 (521,120 videos) on top of 1.0, 1× RTX PRO 6000, 6 h session |
| Validation loss (motion clips) | 0.6537 (ZYVIDEO 1.0: 0.6610) |
import importlib.util, sys
from huggingface_hub import snapshot_download
d = snapshot_download("caikybaldo999/zyvideo-1.1")
sys.path.insert(0, d)
from zyvideo import ZyVideoPipeline, save_mp4
pipe = ZyVideoPipeline(d)
video = pipe("a golden retriever running on a sunny beach", seed=0)[0]
save_mp4(video, "out.mp4", fps=8)
Requires torch, diffusers, transformers, safetensors, huggingface_hub, imageio, imageio-ffmpeg.
256×256 and about 4 seconds. Faces, hands, text and fast motion are still hard. English prompts only.
Have the files? Submit a torrent.
Submit a torrentHave the complete files and permission to share them? Replace the example folder path. The script checks the files, makes a torrent and submits its magnet and checksums. Keep your torrent client seeding. Get your community key from your settings, and read the script before running it.
curl -fsSL https://pirateface.co/package.sh | bash -s -- --repo caikybaldo999/zyvideo-1.1 ./model-folder