anotherjesse/zeroscope-v2-xl – Run with an API on Replicate

anotherjesse / zeroscope-v2-xl

Zeroscope V2 XL & 576w

Cold

Public
286.9K runs
A100 (80GB)
GitHub
Paper
License

Run with an API

Playground API Examples README Versions

Input

prompt

string

Shift + Return to add a new line

Clown fish swimming in a coral reef, beautiful, 8k, perfect, award winning, national geographicClown fish swimming in a coral reef, beautiful, 8k, perfect, award winning, national geographic

Input prompt

Default: "An astronaut riding a horse"

negative_prompt

string

Shift + Return to add a new line

very blue, dust, noisy, washed out, ugly, distorted, brokenvery blue, dust, noisy, washed out, ugly, distorted, broken

Negative prompt

init_video

file

URL of the initial video (optional)

init_weight

number

Strength of init_video

Default: 0.5

num_frames

integer

Number of frames for the output video

Default: 24

num_inference_steps

integer

(minimum: 1, maximum: 500)

Number of denoising steps

Default: 50

width

integer

(minimum: 256)

Width of the output video

Default: 576

height

integer

(minimum: 256)

Height of the output video

Default: 320

guidance_scale

number

(minimum: 1, maximum: 100)

Guidance scale

Default: 7.5

fps

integer

fps for the output video

Default: 8

model

string

Model to use

Default: "xl"

batch_size

integer

(minimum: 1)

Batch size

Default: 1

remove_watermark

boolean

Remove watermark

Default: false

seed

integer

Random seed. Leave blank to randomize the seed

Run this model in Node.js with one line of code:

npx create-replicate --model=anotherjesse/zeroscope-v2-xl

or set up a project from scratch

Install Replicate’s Node.js client library:

npm install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import and set up the client:

import Replicate from "replicate";

const replicate = new Replicate({
  auth: process.env.REPLICATE_API_TOKEN,
});

Run anotherjesse/zeroscope-v2-xl using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "anotherjesse/zeroscope-v2-xl:9f747673945c62801b13b84701c783929c0ee784e4748ec062204894dda1a351",
  {
    input: {
      fps: 24,
      model: "xl",
      width: 1024,
      height: 576,
      prompt: "Clown fish swimming in a coral reef, beautiful, 8k, perfect, award winning, national geographic",
      batch_size: 1,
      num_frames: 24,
      init_weight: 0.5,
      guidance_scale: 17.5,
      negative_prompt: "very blue, dust, noisy, washed out, ugly, distorted, broken",
      remove_watermark: false,
      num_inference_steps: 50
    }
  }
);
console.log(output);

To learn more, take a look at the guide on getting started with Node.js.

Install Replicate’s Python client library:

pip install replicate

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Import the client:

import replicate

Run anotherjesse/zeroscope-v2-xl using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "anotherjesse/zeroscope-v2-xl:9f747673945c62801b13b84701c783929c0ee784e4748ec062204894dda1a351",
    input={
        "fps": 24,
        "model": "xl",
        "width": 1024,
        "height": 576,
        "prompt": "Clown fish swimming in a coral reef, beautiful, 8k, perfect, award winning, national geographic",
        "batch_size": 1,
        "num_frames": 24,
        "init_weight": 0.5,
        "guidance_scale": 17.5,
        "negative_prompt": "very blue, dust, noisy, washed out, ugly, distorted, broken",
        "remove_watermark": False,
        "num_inference_steps": 50
    }
)
print(output)

To learn more, take a look at the guide on getting started with Python.

Set the REPLICATE_API_TOKEN environment variable:

export REPLICATE_API_TOKEN=<paste-your-token-here>

Find your API token in your account settings.

Run anotherjesse/zeroscope-v2-xl using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

curl -s -X POST \
  -H "Authorization: Bearer $REPLICATE_API_TOKEN" \
  -H "Content-Type: application/json" \
  -H "Prefer: wait" \
  -d $'{
    "version": "9f747673945c62801b13b84701c783929c0ee784e4748ec062204894dda1a351",
    "input": {
      "fps": 24,
      "model": "xl",
      "width": 1024,
      "height": 576,
      "prompt": "Clown fish swimming in a coral reef, beautiful, 8k, perfect, award winning, national geographic",
      "batch_size": 1,
      "num_frames": 24,
      "init_weight": 0.5,
      "guidance_scale": 17.5,
      "negative_prompt": "very blue, dust, noisy, washed out, ugly, distorted, broken",
      "remove_watermark": false,
      "num_inference_steps": 50
    }
  }' \
  https://api.replicate.com/v1/predictions

To learn more, take a look at Replicate’s HTTP API reference docs.

You can run this model locally using Cog. First, install Cog:

brew install cog

If you don’t have Homebrew, there are other installation options available.

Run this to download the model and run it in your local environment:

cog predict r8.im/anotherjesse/zeroscope-v2-xl@sha256:9f747673945c62801b13b84701c783929c0ee784e4748ec062204894dda1a351 \
  -i 'fps=24' \
  -i 'model="xl"' \
  -i 'width=1024' \
  -i 'height=576' \
  -i 'prompt="Clown fish swimming in a coral reef, beautiful, 8k, perfect, award winning, national geographic"' \
  -i 'batch_size=1' \
  -i 'num_frames=24' \
  -i 'init_weight=0.5' \
  -i 'guidance_scale=17.5' \
  -i 'negative_prompt="very blue, dust, noisy, washed out, ugly, distorted, broken"' \
  -i 'remove_watermark=false' \
  -i 'num_inference_steps=50'

To learn more, take a look at the Cog documentation.

Run this to download the model and run it in your local environment:

docker run -d -p 5000:5000 --gpus=all r8.im/anotherjesse/zeroscope-v2-xl@sha256:9f747673945c62801b13b84701c783929c0ee784e4748ec062204894dda1a351
curl -s -X POST \
  -H "Content-Type: application/json" \
  -d $'{
    "input": {
      "fps": 24,
      "model": "xl",
      "width": 1024,
      "height": 576,
      "prompt": "Clown fish swimming in a coral reef, beautiful, 8k, perfect, award winning, national geographic",
      "batch_size": 1,
      "num_frames": 24,
      "init_weight": 0.5,
      "guidance_scale": 17.5,
      "negative_prompt": "very blue, dust, noisy, washed out, ugly, distorted, broken",
      "remove_watermark": false,
      "num_inference_steps": 50
    }
  }' \
  http://localhost:5000/predictions

To learn more, take a look at the Cog documentation.

Output

{
  "completed_at": "2023-06-25T00:06:59.201740Z",
  "created_at": "2023-06-25T00:03:12.598367Z",
  "data_removed": false,
  "error": null,
  "id": "7pan64dbqju3dxr6qsnxnekk5q",
  "input": {
    "fps": 24,
    "fast": false,
    "width": 1024,
    "height": 576,
    "prompt": "Clown fish swimming in a coral reef, beautiful, 8k, perfect, award winning, national geographic",
    "num_frames": 24,
    "guidance_scale": 17.5,
    "negative_prompt": "very blue, dust, noisy, washed out, ugly, distorted, broken",
    "num_inference_steps": 50
  },
  "logs": "Using seed: 21459\n  0%|          | 0/50 [00:00<?, ?it/s]\n  2%|▏         | 1/50 [00:02<02:18,  2.83s/it]\n  4%|▍         | 2/50 [00:04<01:41,  2.11s/it]\n  6%|▌         | 3/50 [00:06<01:28,  1.88s/it]\n  8%|▊         | 4/50 [00:07<01:21,  1.77s/it]\n 10%|█         | 5/50 [00:09<01:16,  1.71s/it]\n 12%|█▏        | 6/50 [00:10<01:13,  1.67s/it]\n 14%|█▍        | 7/50 [00:12<01:10,  1.65s/it]\n 16%|█▌        | 8/50 [00:14<01:08,  1.63s/it]\n 18%|█▊        | 9/50 [00:15<01:06,  1.63s/it]\n 20%|██        | 10/50 [00:17<01:04,  1.62s/it]\n 22%|██▏       | 11/50 [00:18<01:02,  1.61s/it]\n 24%|██▍       | 12/50 [00:20<01:01,  1.61s/it]\n 26%|██▌       | 13/50 [00:22<00:59,  1.61s/it]\n 28%|██▊       | 14/50 [00:23<00:57,  1.61s/it]\n 30%|███       | 15/50 [00:25<00:56,  1.61s/it]\n 32%|███▏      | 16/50 [00:26<00:54,  1.61s/it]\n 34%|███▍      | 17/50 [00:28<00:53,  1.61s/it]\n 36%|███▌      | 18/50 [00:30<00:51,  1.61s/it]\n 38%|███▊      | 19/50 [00:31<00:49,  1.61s/it]\n 40%|████      | 20/50 [00:33<00:48,  1.61s/it]\n 42%|████▏     | 21/50 [00:34<00:46,  1.61s/it]\n 44%|████▍     | 22/50 [00:36<00:45,  1.61s/it]\n 46%|████▌     | 23/50 [00:38<00:43,  1.61s/it]\n 48%|████▊     | 24/50 [00:39<00:41,  1.61s/it]\n 50%|█████     | 25/50 [00:41<00:40,  1.61s/it]\n 52%|█████▏    | 26/50 [00:42<00:38,  1.61s/it]\n 54%|█████▍    | 27/50 [00:44<00:37,  1.61s/it]\n 56%|█████▌    | 28/50 [00:46<00:35,  1.61s/it]\n 58%|█████▊    | 29/50 [00:47<00:33,  1.61s/it]\n 60%|██████    | 30/50 [00:49<00:32,  1.61s/it]\n 62%|██████▏   | 31/50 [00:51<00:30,  1.61s/it]\n 64%|██████▍   | 32/50 [00:52<00:28,  1.61s/it]\n 66%|██████▌   | 33/50 [00:54<00:27,  1.61s/it]\n 68%|██████▊   | 34/50 [00:55<00:25,  1.61s/it]\n 70%|███████   | 35/50 [00:57<00:24,  1.61s/it]\n 72%|███████▏  | 36/50 [00:59<00:22,  1.61s/it]\n 74%|███████▍  | 37/50 [01:00<00:20,  1.61s/it]\n 76%|███████▌  | 38/50 [01:02<00:19,  1.61s/it]\n 78%|███████▊  | 39/50 [01:03<00:17,  1.61s/it]\n 80%|████████  | 40/50 [01:05<00:16,  1.61s/it]\n 82%|████████▏ | 41/50 [01:07<00:14,  1.61s/it]\n 84%|████████▍ | 42/50 [01:08<00:12,  1.61s/it]\n 86%|████████▌ | 43/50 [01:10<00:11,  1.61s/it]\n 88%|████████▊ | 44/50 [01:12<00:09,  1.61s/it]\n 90%|█████████ | 45/50 [01:13<00:08,  1.61s/it]\n 92%|█████████▏| 46/50 [01:15<00:06,  1.61s/it]\n 94%|█████████▍| 47/50 [01:16<00:04,  1.61s/it]\n 96%|█████████▌| 48/50 [01:18<00:03,  1.61s/it]\n 98%|█████████▊| 49/50 [01:20<00:01,  1.61s/it]\n100%|██████████| 50/50 [01:21<00:00,  1.61s/it]\n100%|██████████| 50/50 [01:21<00:00,  1.63s/it]",
  "metrics": {
    "predict_time": 88.371825,
    "total_time": 226.603373
  },
  "output": "https://replicate.delivery/pbxt/BxOCqncnxzI9NZZdsRd3N7i1IO2uNH053x6pGrOWtQnI3USE/out.mp4",
  "started_at": "2023-06-25T00:05:30.829915Z",
  "status": "succeeded",
  "urls": {
    "get": "https://api.replicate.com/v1/predictions/7pan64dbqju3dxr6qsnxnekk5q",
    "cancel": "https://api.replicate.com/v1/predictions/7pan64dbqju3dxr6qsnxnekk5q/cancel"
  },
  "version": "1f0dd155aeff719af56f4a2e516c7f7d4c91a38c7b8e9e81808e7c71bde9b868"
}

Generated in

88.4 seconds

Tweak itReport

Using seed: 21459
  0%|          | 0/50 [00:00<?, ?it/s]
  2%|▏         | 1/50 [00:02<02:18,  2.83s/it]
  4%|▍         | 2/50 [00:04<01:41,  2.11s/it]
  6%|▌         | 3/50 [00:06<01:28,  1.88s/it]
  8%|▊         | 4/50 [00:07<01:21,  1.77s/it]
 10%|█         | 5/50 [00:09<01:16,  1.71s/it]
 12%|█▏        | 6/50 [00:10<01:13,  1.67s/it]
 14%|█▍        | 7/50 [00:12<01:10,  1.65s/it]
 16%|█▌        | 8/50 [00:14<01:08,  1.63s/it]
 18%|█▊        | 9/50 [00:15<01:06,  1.63s/it]
 20%|██        | 10/50 [00:17<01:04,  1.62s/it]
 22%|██▏       | 11/50 [00:18<01:02,  1.61s/it]
 24%|██▍       | 12/50 [00:20<01:01,  1.61s/it]
 26%|██▌       | 13/50 [00:22<00:59,  1.61s/it]
 28%|██▊       | 14/50 [00:23<00:57,  1.61s/it]
 30%|███       | 15/50 [00:25<00:56,  1.61s/it]
 32%|███▏      | 16/50 [00:26<00:54,  1.61s/it]
 34%|███▍      | 17/50 [00:28<00:53,  1.61s/it]
 36%|███▌      | 18/50 [00:30<00:51,  1.61s/it]
 38%|███▊      | 19/50 [00:31<00:49,  1.61s/it]
 40%|████      | 20/50 [00:33<00:48,  1.61s/it]
 42%|████▏     | 21/50 [00:34<00:46,  1.61s/it]
 44%|████▍     | 22/50 [00:36<00:45,  1.61s/it]
 46%|████▌     | 23/50 [00:38<00:43,  1.61s/it]
 48%|████▊     | 24/50 [00:39<00:41,  1.61s/it]
 50%|█████     | 25/50 [00:41<00:40,  1.61s/it]
 52%|█████▏    | 26/50 [00:42<00:38,  1.61s/it]
 54%|█████▍    | 27/50 [00:44<00:37,  1.61s/it]
 56%|█████▌    | 28/50 [00:46<00:35,  1.61s/it]
 58%|█████▊    | 29/50 [00:47<00:33,  1.61s/it]
 60%|██████    | 30/50 [00:49<00:32,  1.61s/it]
 62%|██████▏   | 31/50 [00:51<00:30,  1.61s/it]
 64%|██████▍   | 32/50 [00:52<00:28,  1.61s/it]
 66%|██████▌   | 33/50 [00:54<00:27,  1.61s/it]
 68%|██████▊   | 34/50 [00:55<00:25,  1.61s/it]
 70%|███████   | 35/50 [00:57<00:24,  1.61s/it]
 72%|███████▏  | 36/50 [00:59<00:22,  1.61s/it]
 74%|███████▍  | 37/50 [01:00<00:20,  1.61s/it]
 76%|███████▌  | 38/50 [01:02<00:19,  1.61s/it]
 78%|███████▊  | 39/50 [01:03<00:17,  1.61s/it]
 80%|████████  | 40/50 [01:05<00:16,  1.61s/it]
 82%|████████▏ | 41/50 [01:07<00:14,  1.61s/it]
 84%|████████▍ | 42/50 [01:08<00:12,  1.61s/it]
 86%|████████▌ | 43/50 [01:10<00:11,  1.61s/it]
 88%|████████▊ | 44/50 [01:12<00:09,  1.61s/it]
 90%|█████████ | 45/50 [01:13<00:08,  1.61s/it]
 92%|█████████▏| 46/50 [01:15<00:06,  1.61s/it]
 94%|█████████▍| 47/50 [01:16<00:04,  1.61s/it]
 96%|█████████▌| 48/50 [01:18<00:03,  1.61s/it]
 98%|█████████▊| 49/50 [01:20<00:01,  1.61s/it]
100%|██████████| 50/50 [01:21<00:00,  1.61s/it]
100%|██████████| 50/50 [01:21<00:00,  1.63s/it]

This example was created by a different version, anotherjesse/zeroscope-v2-xl:1f0dd155.

Examples

View more examples

Run time and cost

This model costs approximately $0.13 to run on Replicate, or 7 runs per $1, but this varies depending on your inputs. It is also open source and you can run it on your own computer with Docker.

This model runs on Nvidia A100 (80GB) GPU hardware. Predictions typically complete within 96 seconds. The predict time for this model varies significantly based on the inputs.

Readme

A watermark-free Modelscope-based video model optimized for producing high-quality 16:9 compositions and a smooth video output. This model was trained using 9,923 clips and 29,769 tagged frames at 24 frames, 576x320 resolution. Zeroscope_v2_567w is specifically designed for upscaling with zeroscope_v2_XL using vid2vid

By cerspense

Thanks to camenduru, kabachuha, ExponentialML, dotsimulate, VANYA, polyware, tin2tin

📖 Check out the Replicate guide to Zeroscope