jzhang38/fast-hunyuan-video 🔢📝 → 🖼️

▶️ 465 runs 📅 Dec 2024 ⚙️ Cog 0.13.6 🔗 GitHub ⚖️ License

text-to-video

About

Fast Hunyuan Video by Hao AI Lab

Example Output

Prompt:

"A cat walks on the grass, realistic style."

Output

Performance Metrics

233.80s Prediction Time

338.56s Total Time

All Input Parameters

{
  "fps": 24,
  "seed": 0,
  "width": 1280,
  "height": 720,
  "prompt": "A cat walks on the grass, realistic style.",
  "flow_shift": 17,
  "num_frames": 125,
  "guidance_scale": 1,
  "negative_prompt": "",
  "embedded_cfg_scale": 6,
  "num_inference_steps": 6
}

Input Parameters

fps Type: integerDefault: 24Range: 1 - 60: Frames per second of output video
seed Type: integerDefault: 0: 0 for Random seed. Set for reproducible generation
width Type: integerDefault: 1280Range: 256 - ∞: Width of output video
height Type: integerDefault: 720Range: 256 - ∞: Height of output video
prompt Type: stringDefault: A cat walks on the grass, realistic style.: Text prompt for video generation
flow_shift Type: integerDefault: 17Range: 1 - 20: Flow shift parameter
num_frames Type: integerDefault: 125Range: 16 - ∞: Number of frames to generate
guidance_scale Type: numberDefault: 1Range: 0.1 - 10: Classifier free guidance scale
negative_prompt Type: stringDefault:: Text prompt to specify what you don't want in the video.
embedded_cfg_scale Type: numberDefault: 6Range: 0.1 - 10: Embedded classifier free guidance scale
num_inference_steps Type: integerDefault: 6Range: 1 - 50: Number of denoising steps

Output Schema

Output

Type: string • Format: uri

Example Execution Logs

Using seed: 26705
2024-12-18 23:38:43.793 | INFO     | fastvideo.models.hunyuan.inference:predict:444 - Input (height, width, video_length) = (720, 1280, 125)
2024-12-18 23:38:43.793 | DEBUG    | fastvideo.models.hunyuan.inference:predict:502 -
height: 720
width: 1280
video_length: 125
prompt: ['A cat walks on the grass, realistic style.']
neg_prompt: ['Aerial view, aerial view, overexposed, low quality, deformation, a poor composition, bad hands, bad teeth, bad eyes, bad limbs, distortion']
seed: 26705
infer_steps: 6
num_videos_per_prompt: 1
guidance_scale: 1.0
n_tokens: 460800
flow_shift: 17
embedded_guidance_scale: 6.0
  0%|          | 0/6 [00:00<?, ?it/s]
 17%|█▋        | 1/6 [00:33<02:47, 33.48s/it]
 33%|███▎      | 2/6 [01:06<02:12, 33.19s/it]
 50%|█████     | 3/6 [01:39<01:39, 33.32s/it]
 67%|██████▋   | 4/6 [02:13<01:06, 33.38s/it]
 83%|████████▎ | 5/6 [02:46<00:33, 33.43s/it]
100%|██████████| 6/6 [03:20<00:00, 33.44s/it]
100%|██████████| 6/6 [03:20<00:00, 33.40s/it]
2024-12-18 23:42:36.345 | INFO     | fastvideo.models.hunyuan.inference:predict:530 - Success, time: 232.55122566223145
huggingface/tokenizers: The current process just got forked, after parallelism has already been used. Disabling parallelism to avoid deadlocks...
To disable this warning, you can either:
- Avoid using `tokenizers` before the fork if possible
- Explicitly set the environment variable TOKENIZERS_PARALLELISM=(true | false)

Version Details

Version ID: 6915992f696ccc626a53bb3e7e8941af911e9121df705d5f2cfde4a2a58f8330
Version Created: December 18, 2024

Run on Replicate →