zurk/hunyuan-video-8bit 🔢📝 → 🖼️

▶️ 239 runs 📅 Dec 2024 ⚙️ Cog 0.13.6 🔗 GitHub

text-to-video

About

Hunyuan Video 8bit model API for video generation

Example Output

Prompt:

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

Output

Performance Metrics

382.78s Prediction Time

656.25s Total Time

All Input Parameters

{
  "seed": 43,
  "width": 960,
  "height": 544,
  "prompt": "A cat walks on the grass, realistic style.",
  "flow_shift": 7,
  "video_length": 65,
  "negative_prompt": "Ugly",
  "num_inference_steps": 40,
  "embedded_guidance_scale": 6
}

Input Parameters

seed Type: integer: Random seed
width Type: integerDefault: 960Range: 128 - 1920: Video width
height Type: integerDefault: 544Range: 128 - 1080: Video height
prompt Type: stringDefault: A cat walks on the grass, realistic style.: Input prompt for video generation
flow_shift Type: numberDefault: 7Range: 0 - 20: Flow shift parameter
video_length Type: integerDefault: 65Range: 16 - 256: Number of frames
negative_prompt Type: stringDefault: Ugly: Negative prompt
num_inference_steps Type: integerDefault: 40Range: 1 - 100: Number of inference steps
embedded_guidance_scale Type: numberDefault: 6Range: 1 - 20: Embedded guidance scale

Output Schema

Output

Type: string • Format: uri

Example Execution Logs

2024-12-09 10:56:51,578-INFO-cog_predictor-seed=43
2024-12-09 10:56:59,013-INFO-ComfyUI-HunyuanVideoWrapper.utils-Input (height, width, video_length) = (544, 960, 65)
2024-12-09 10:57:01,472-INFO-ComfyUI-HunyuanVideoWrapper.hyvideo.diffusion.pipelines.pipeline_hunyuan_video-Sampling 65 frames in 17 latents at 960x544 with 40 inference steps
  0%|          | 0/40 [00:00<?, ?it/s]
  2%|▎         | 1/40 [00:10<06:47, 10.45s/it]
  5%|▌         | 2/40 [00:19<05:55,  9.36s/it]
  8%|▊         | 3/40 [00:27<05:34,  9.05s/it]
 10%|█         | 4/40 [00:36<05:20,  8.90s/it]
 12%|█▎        | 5/40 [00:45<05:08,  8.82s/it]
 15%|█▌        | 6/40 [00:53<04:58,  8.79s/it]
 18%|█▊        | 7/40 [01:02<04:49,  8.77s/it]
 20%|██        | 8/40 [01:11<04:40,  8.75s/it]
 22%|██▎       | 9/40 [01:20<04:31,  8.75s/it]
 25%|██▌       | 10/40 [01:28<04:22,  8.75s/it]
 28%|██▊       | 11/40 [01:37<04:13,  8.75s/it]
 30%|███       | 12/40 [01:46<04:04,  8.75s/it]
 32%|███▎      | 13/40 [01:55<03:56,  8.75s/it]
 35%|███▌      | 14/40 [02:03<03:47,  8.75s/it]
 38%|███▊      | 15/40 [02:12<03:38,  8.75s/it]
 40%|████      | 16/40 [02:21<03:29,  8.75s/it]
 42%|████▎     | 17/40 [02:29<03:21,  8.75s/it]
 45%|████▌     | 18/40 [02:38<03:12,  8.75s/it]
 48%|████▊     | 19/40 [02:47<03:03,  8.75s/it]
 50%|█████     | 20/40 [02:56<02:54,  8.75s/it]
 52%|█████▎    | 21/40 [03:04<02:46,  8.75s/it]
 55%|█████▌    | 22/40 [03:13<02:37,  8.75s/it]
 57%|█████▊    | 23/40 [03:22<02:28,  8.75s/it]
 60%|██████    | 24/40 [03:31<02:19,  8.75s/it]
 62%|██████▎   | 25/40 [03:39<02:11,  8.75s/it]
 65%|██████▌   | 26/40 [03:48<02:02,  8.75s/it]
 68%|██████▊   | 27/40 [03:57<01:53,  8.75s/it]
 70%|███████   | 28/40 [04:06<01:44,  8.75s/it]
 72%|███████▎  | 29/40 [04:14<01:36,  8.76s/it]
 75%|███████▌  | 30/40 [04:23<01:27,  8.76s/it]
 78%|███████▊  | 31/40 [04:32<01:18,  8.76s/it]
 80%|████████  | 32/40 [04:41<01:10,  8.76s/it]
 82%|████████▎ | 33/40 [04:50<01:01,  8.76s/it]
 85%|████████▌ | 34/40 [04:58<00:52,  8.75s/it]
 88%|████████▊ | 35/40 [05:07<00:43,  8.75s/it]
 90%|█████████ | 36/40 [05:16<00:35,  8.77s/it]
 92%|█████████▎| 37/40 [05:25<00:26,  8.76s/it]
 95%|█████████▌| 38/40 [05:33<00:17,  8.76s/it]
 98%|█████████▊| 39/40 [05:42<00:08,  8.75s/it]
100%|██████████| 40/40 [05:51<00:00,  8.76s/it]
100%|██████████| 40/40 [05:51<00:00,  8.78s/it]
2024-12-09 11:02:52,817-INFO-ComfyUI-HunyuanVideoWrapper.utils-Allocated memory: memory=11.985 GB
2024-12-09 11:02:52,817-INFO-ComfyUI-HunyuanVideoWrapper.utils-Max allocated memory: max_memory=17.704 GB
2024-12-09 11:02:52,817-INFO-ComfyUI-HunyuanVideoWrapper.utils-Max reserved memory: max_reserved=19.467 GB

Version Details

Version ID: 392ce81c5d87a6791d8487644b5835983ef1a197d6d77856333997d234ca582b
Version Created: December 9, 2024

Run on Replicate →