nightmareai / cogvideo

Text-to-video generation

Cold

Public
32.8K runs
A100 (80GB)
GitHub
Paper
License

Run with an API

Playground API Examples README Versions

Input

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

npx create-replicate --model=nightmareai/cogvideo

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 nightmareai/cogvideo using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "nightmareai/cogvideo:00b1c7885c5f1d44b51bcb56c378abc8f141eeacf94c1e64998606515fe63a8d",
  {
    input: {
      seed: -1,
      prompt: "a black cat",
      translate: true,
      both_stages: true,
      use_guidance: true
    }
  }
);
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 nightmareai/cogvideo using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "nightmareai/cogvideo:00b1c7885c5f1d44b51bcb56c378abc8f141eeacf94c1e64998606515fe63a8d",
    input={
        "seed": -1,
        "prompt": "a black cat",
        "translate": True,
        "both_stages": True,
        "use_guidance": True
    }
)

# The nightmareai/cogvideo model can stream output as it's running.
# The predict method returns an iterator, and you can iterate over that output.
for item in output:
    # https://replicate.com/nightmareai/cogvideo/api#output-schema
    print(item)

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 nightmareai/cogvideo 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": "00b1c7885c5f1d44b51bcb56c378abc8f141eeacf94c1e64998606515fe63a8d",
    "input": {
      "seed": -1,
      "prompt": "a black cat",
      "translate": true,
      "both_stages": true,
      "use_guidance": true
    }
  }' \
  https://api.replicate.com/v1/predictions

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

Output

{
  "completed_at": "2022-07-26T06:14:48.029302Z",
  "created_at": "2022-07-26T05:44:58.164332Z",
  "data_removed": false,
  "error": null,
  "id": "vi5hhcfyqnft3jhxusetmmaqea",
  "input": {
    "seed": "-1",
    "prompt": "a black cat",
    "translate": true,
    "both_stages": true,
    "use_guidance": true
  },
  "logs": "2022-07-26 06:02:08 DEBUG    Starting new HTTPS connection (1): translate.google.com:443\n2022-07-26 06:02:08 DEBUG    https://translate.google.com:443 \"GET /m?tl=zh-CN&sl=en&q=a+black+cat HTTP/1.1\" 200 None\n2022-07-26 06:02:08 DEBUG    moving stage 2 model to cpu\n2022-07-26 06:02:08 DEBUG    moving stage 1 model to cuda\n2022-07-26 06:02:12 DEBUG    moving in model1 takes time: 3.75\n2022-07-26 06:02:12 INFO     [Generating First Frame with CogView2]Raw text: 一只黑猫 高清摄影\n2022-07-26 06:02:51 INFO     [First Frame]Taken time 39.40\n\n2022-07-26 06:02:51 INFO     [Stage1: Generating Subsequent Frames, Frame Rate 1.0]\nraw text: 一只黑猫 视频\n2022-07-26 06:02:51 DEBUG    Using Guidance In Inference\n2022-07-26 06:02:51 DEBUG    Limit spatial-channel's mem to current frame\n2022-07-26 06:08:17 DEBUG    Attempting to acquire lock 140517844067568 on /root/.icetk_models/ice_image.pt.lock\n2022-07-26 06:08:17 DEBUG    Lock 140517844067568 acquired on /root/.icetk_models/ice_image.pt.lock\n2022-07-26 06:08:17 DEBUG    Attempting to release lock 140517844067568 on /root/.icetk_models/ice_image.pt.lock\n2022-07-26 06:08:17 DEBUG    Lock 140517844067568 released on /root/.icetk_models/ice_image.pt.lock\n\nSave to:  /tmp/tmph7a4dbpv/output/stage1\n2022-07-26 06:08:24 INFO     CogVideo Stage1 completed. Taken time 371.81\n\n2022-07-26 06:08:24 DEBUG    moving stage 1 model to cpu\n2022-07-26 06:08:35 DEBUG    moving stage-2 model to cuda\n2022-07-26 06:08:39 DEBUG    moving in stage-2 model takes time: 4.04\n2022-07-26 06:08:39 INFO     [Stage2: Generating Frames, Frame Rate 2]\nraw text: 一只黑猫\n2022-07-26 06:08:39 DEBUG    Limit spatial-channel's mem to current frame\n2022-07-26 06:10:16 INFO     Duration 2.00, Taken time 96.67\n\n2022-07-26 06:10:16 INFO     [Stage2: Generating Frames, Frame Rate 4]\nraw text: 一只黑猫\n2022-07-26 06:10:16 DEBUG    Limit spatial-channel's mem to current frame",
  "metrics": {
    "predict_time": 760.519747,
    "total_time": 1789.86497
  },
  "output": [
    "https://replicate.delivery/mgxm/f4412e8a-5583-43a9-a5d1-8e0106e15a09/0.gif",
    "https://replicate.delivery/mgxm/6c3a4e36-2332-4280-a506-a4bfd4197f7b/0.gif"
  ],
  "started_at": "2022-07-26T06:02:07.509555Z",
  "status": "succeeded",
  "urls": {
    "get": "https://api.replicate.com/v1/predictions/vi5hhcfyqnft3jhxusetmmaqea",
    "cancel": "https://api.replicate.com/v1/predictions/vi5hhcfyqnft3jhxusetmmaqea/cancel"
  },
  "version": "6d55d6fdc738b66c996c46b2fe2a4ef965e613be4cb44d84664c07f305e92c2d"
}

Generated in

12 minutes 41 seconds

Tweak it Report

This example was created by a different version, nightmareai/cogvideo:6d55d6fd.

Examples

View more examples

Run time and cost

This model costs approximately $2.33 to run on Replicate, or 0 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 28 minutes. The predict time for this model varies significantly based on the inputs.

Readme

CogVideo: Large-scale Pretraining for Text-to-Video Generation via Transformers

Setup time can be long as the container is 63GB.

Image prompts are supported thanks to a contribution from nev

Stage 1 output will be a few frames, stage 2 interpolates a longer video and performs dsr resampling.

When running both stages, stage 1 output will render when ready, stage 2 will follow when complete.

Please see the official CogVideo repo for more information: https://github.com/THUDM/CogVideo