emaph / outpaint-controlnet-union

Outpaint an image using controlnet union for SDXL.

Cold

Public
7.6K runs
L40S

Iterate in playground

Run with an API

Playground API Examples README Versions

Input

prompt

string

Shift + Return to add a new line

Default: ""

negative_prompt

string

Shift + Return to add a new line

Things you do not want to see in your image

Default: "ugly"

image

file

The image to expand

left

integer

Outpaint left

Default: 150

right

integer

Outpaint right

Default: 150

top

integer

Outpaint top

Default: 150

bottom

integer

Outpaint bottom

Default: 150

steps

integer

(minimum: 1, maximum: 50)

Number of inference steps

Default: 20

cfg

number

(minimum: 0, maximum: 20)

Guidance scale

Default: 4

output_format

string

Format of the output images

Default: "webp"

output_quality

integer

(minimum: 0, maximum: 100)

Quality of the output images, from 0 to 100. 100 is best quality, 0 is lowest quality.

Default: 80

seed

integer

Set a seed for reproducibility. Random by default.

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

npx create-replicate --model=emaph/outpaint-controlnet-union

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";
import fs from "node:fs";

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

Run emaph/outpaint-controlnet-union using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "emaph/outpaint-controlnet-union:377564d35153c66f8629d9540480813685d114f0552e9a3c9ffe5dd315091e68",
  {
    input: {
      cfg: 4,
      top: 0,
      left: 400,
      image: "https://i.postimg.cc/vmY5BZP3/2024-04-16-21-30-59-9956.png",
      right: 400,
      steps: 20,
      bottom: 0,
      prompt: "ice dragon, wings wide open",
      output_format: "webp",
      output_quality: 80,
      negative_prompt: "ugly"
    }
  }
);

// To access the file URL:
console.log(output[0].url()); //=> "http://example.com"

// To write the file to disk:
fs.writeFile("my-image.png", output[0]);

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 emaph/outpaint-controlnet-union using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "emaph/outpaint-controlnet-union:377564d35153c66f8629d9540480813685d114f0552e9a3c9ffe5dd315091e68",
    input={
        "cfg": 4,
        "top": 0,
        "left": 400,
        "image": "https://i.postimg.cc/vmY5BZP3/2024-04-16-21-30-59-9956.png",
        "right": 400,
        "steps": 20,
        "bottom": 0,
        "prompt": "ice dragon, wings wide open",
        "output_format": "webp",
        "output_quality": 80,
        "negative_prompt": "ugly"
    }
)
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 emaph/outpaint-controlnet-union 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": "emaph/outpaint-controlnet-union:377564d35153c66f8629d9540480813685d114f0552e9a3c9ffe5dd315091e68",
    "input": {
      "cfg": 4,
      "top": 0,
      "left": 400,
      "image": "https://i.postimg.cc/vmY5BZP3/2024-04-16-21-30-59-9956.png",
      "right": 400,
      "steps": 20,
      "bottom": 0,
      "prompt": "ice dragon, wings wide open",
      "output_format": "webp",
      "output_quality": 80,
      "negative_prompt": "ugly"
    }
  }' \
  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/emaph/outpaint-controlnet-union@sha256:377564d35153c66f8629d9540480813685d114f0552e9a3c9ffe5dd315091e68 \
  -i 'cfg=4' \
  -i 'top=0' \
  -i 'left=400' \
  -i 'image="https://i.postimg.cc/vmY5BZP3/2024-04-16-21-30-59-9956.png"' \
  -i 'right=400' \
  -i 'steps=20' \
  -i 'bottom=0' \
  -i 'prompt="ice dragon, wings wide open"' \
  -i 'output_format="webp"' \
  -i 'output_quality=80' \
  -i 'negative_prompt="ugly"'

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/emaph/outpaint-controlnet-union@sha256:377564d35153c66f8629d9540480813685d114f0552e9a3c9ffe5dd315091e68
curl -s -X POST \
  -H "Content-Type: application/json" \
  -d $'{
    "input": {
      "cfg": 4,
      "top": 0,
      "left": 400,
      "image": "https://i.postimg.cc/vmY5BZP3/2024-04-16-21-30-59-9956.png",
      "right": 400,
      "steps": 20,
      "bottom": 0,
      "prompt": "ice dragon, wings wide open",
      "output_format": "webp",
      "output_quality": 80,
      "negative_prompt": "ugly"
    }
  }' \
  http://localhost:5000/predictions

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

Output

{
  "completed_at": "2024-07-28T13:02:19.473227Z",
  "created_at": "2024-07-28T13:00:12.069000Z",
  "data_removed": false,
  "error": null,
  "id": "sbx0zj2vmnrgm0cgz5pt274q5g",
  "input": {
    "top": 0,
    "left": 400,
    "image": "https://i.postimg.cc/vmY5BZP3/2024-04-16-21-30-59-9956.png",
    "right": 400,
    "bottom": 0,
    "prompt": "ice dragon, wings wide open",
    "output_format": "webp",
    "output_quality": 80,
    "negative_prompt": "ugly"
  },
  "logs": "Random seed set to: 2036446667\nChecking inputs\nDownloading https://i.postimg.cc/vmY5BZP3/2024-04-16-21-30-59-9956.png to /tmp/inputs/2024-04-16-21-30-59-9956.png\n✅ /tmp/inputs/2024-04-16-21-30-59-9956.png\n====================================\nChecking weights\n✅ diffusion_pytorch_model_promax.safetensors exists in ComfyUI/models/controlnet\n⏳ Downloading juggernautXL_v8Rundiffusion.safetensors to ComfyUI/models/checkpoints\n✅ juggernautXL_v8Rundiffusion.safetensors downloaded to ComfyUI/models/checkpoints in 16.15s, size: 6776.19MB\n====================================\nRunning workflow\ngot prompt\nExecuting node 4, title: Load Checkpoint, class type: CheckpointLoaderSimple\nmodel_type EPS\nUsing pytorch attention in VAE\nUsing pytorch attention in VAE\nloaded straight to GPU\nRequested to load SDXL\nLoading 1 new model\nExecuting node 6, title: Positive, class type: CLIPTextEncode\nRequested to load SDXLClipModel\nLoading 1 new model\nExecuting node 7, title: Negetive, class type: CLIPTextEncode\nExecuting node 16, title: Load ControlNet Model, class type: ControlNetLoader\nExecuting node 17, title: SetUnionControlNetType, class type: SetUnionControlNetType\nExecuting node 11, title: Load Image, class type: LoadImage\nExecuting node 10, title: Pad Image for Outpainting, class type: ImagePadForOutpaint\nExecuting node 48, title: InvertMask, class type: InvertMask\nExecuting node 47, title: Convert Mask to Image, class type: MaskToImage\nExecuting node 49, title: ImageCompositeMasked, class type: ImageCompositeMasked\nExecuting node 15, title: Apply ControlNet (Advanced), class type: ControlNetApplyAdvanced\nExecuting node 39, title: VAE Encode (for Inpainting), class type: VAEEncodeForInpaint\nRequested to load AutoencoderKL\nLoading 1 new model\nExecuting node 3, title: KSampler, class type: KSampler\nRequested to load ControlNet\nLoading 1 new model\n  0%|          | 0/21 [00:00<?, ?it/s]\n  5%|▍         | 1/21 [00:00<00:08,  2.48it/s]\n 10%|▉         | 2/21 [00:00<00:06,  2.80it/s]\n 14%|█▍        | 3/21 [00:01<00:06,  2.93it/s]\n 19%|█▉        | 4/21 [00:01<00:05,  3.00it/s]\n 24%|██▍       | 5/21 [00:01<00:05,  3.03it/s]\n 29%|██▊       | 6/21 [00:02<00:04,  3.05it/s]\n 33%|███▎      | 7/21 [00:02<00:04,  3.07it/s]\n 38%|███▊      | 8/21 [00:02<00:04,  3.08it/s]\n 43%|████▎     | 9/21 [00:02<00:03,  3.08it/s]\n 48%|████▊     | 10/21 [00:03<00:03,  3.09it/s]\n 52%|█████▏    | 11/21 [00:03<00:03,  3.10it/s]\n 57%|█████▋    | 12/21 [00:03<00:02,  3.10it/s]\n 62%|██████▏   | 13/21 [00:04<00:02,  3.11it/s]\n 67%|██████▋   | 14/21 [00:04<00:02,  3.11it/s]\n 71%|███████▏  | 15/21 [00:04<00:01,  3.11it/s]\n 76%|███████▌  | 16/21 [00:05<00:01,  3.11it/s]\n 81%|████████  | 17/21 [00:05<00:01,  3.11it/s]\n 86%|████████▌ | 18/21 [00:05<00:00,  3.11it/s]\n 90%|█████████ | 19/21 [00:06<00:00,  3.11it/s]\n 95%|█████████▌| 20/21 [00:06<00:00,  3.11it/s]\n100%|██████████| 21/21 [00:06<00:00,  3.11it/s]\n100%|██████████| 21/21 [00:06<00:00,  3.07it/s]\nExecuting node 8, title: VAE Decode, class type: VAEDecode\nExecuting node 60, title: Save Image, class type: SaveImage\nPrompt executed in 13.26 seconds\noutputs:  {'60': {'images': [{'filename': 'outpainted_00001_.png', 'subfolder': '', 'type': 'output'}]}}\n====================================\noutpainted_00001_.png",
  "metrics": {
    "predict_time": 31.679283715,
    "total_time": 127.404227
  },
  "output": [
    "https://replicate.delivery/pbxt/krjseJnglBxuaqpAv6G7j7J8xZbhzDHs6d2Gf0MQym2aNfZmA/outpainted_00001_.webp"
  ],
  "started_at": "2024-07-28T13:01:47.793943Z",
  "status": "succeeded",
  "urls": {
    "get": "https://api.replicate.com/v1/predictions/sbx0zj2vmnrgm0cgz5pt274q5g",
    "cancel": "https://api.replicate.com/v1/predictions/sbx0zj2vmnrgm0cgz5pt274q5g/cancel"
  },
  "version": "4d11bfa652f267e27f480471dd8d435cd92c2b99750644a0b38303d13946aae5"
}

Generated in

31.7 seconds

Tweak it ShareReport View full prediction

Random seed set to: 2036446667
Checking inputs
Downloading https://i.postimg.cc/vmY5BZP3/2024-04-16-21-30-59-9956.png to /tmp/inputs/2024-04-16-21-30-59-9956.png
✅ /tmp/inputs/2024-04-16-21-30-59-9956.png
====================================
Checking weights
✅ diffusion_pytorch_model_promax.safetensors exists in ComfyUI/models/controlnet
⏳ Downloading juggernautXL_v8Rundiffusion.safetensors to ComfyUI/models/checkpoints
✅ juggernautXL_v8Rundiffusion.safetensors downloaded to ComfyUI/models/checkpoints in 16.15s, size: 6776.19MB
====================================
Running workflow
got prompt
Executing node 4, title: Load Checkpoint, class type: CheckpointLoaderSimple
model_type EPS
Using pytorch attention in VAE
Using pytorch attention in VAE
loaded straight to GPU
Requested to load SDXL
Loading 1 new model
Executing node 6, title: Positive, class type: CLIPTextEncode
Requested to load SDXLClipModel
Loading 1 new model
Executing node 7, title: Negetive, class type: CLIPTextEncode
Executing node 16, title: Load ControlNet Model, class type: ControlNetLoader
Executing node 17, title: SetUnionControlNetType, class type: SetUnionControlNetType
Executing node 11, title: Load Image, class type: LoadImage
Executing node 10, title: Pad Image for Outpainting, class type: ImagePadForOutpaint
Executing node 48, title: InvertMask, class type: InvertMask
Executing node 47, title: Convert Mask to Image, class type: MaskToImage
Executing node 49, title: ImageCompositeMasked, class type: ImageCompositeMasked
Executing node 15, title: Apply ControlNet (Advanced), class type: ControlNetApplyAdvanced
Executing node 39, title: VAE Encode (for Inpainting), class type: VAEEncodeForInpaint
Requested to load AutoencoderKL
Loading 1 new model
Executing node 3, title: KSampler, class type: KSampler
Requested to load ControlNet
Loading 1 new model
  0%|          | 0/21 [00:00<?, ?it/s]
  5%|▍         | 1/21 [00:00<00:08,  2.48it/s]
 10%|▉         | 2/21 [00:00<00:06,  2.80it/s]
 14%|█▍        | 3/21 [00:01<00:06,  2.93it/s]
 19%|█▉        | 4/21 [00:01<00:05,  3.00it/s]
 24%|██▍       | 5/21 [00:01<00:05,  3.03it/s]
 29%|██▊       | 6/21 [00:02<00:04,  3.05it/s]
 33%|███▎      | 7/21 [00:02<00:04,  3.07it/s]
 38%|███▊      | 8/21 [00:02<00:04,  3.08it/s]
 43%|████▎     | 9/21 [00:02<00:03,  3.08it/s]
 48%|████▊     | 10/21 [00:03<00:03,  3.09it/s]
 52%|█████▏    | 11/21 [00:03<00:03,  3.10it/s]
 57%|█████▋    | 12/21 [00:03<00:02,  3.10it/s]
 62%|██████▏   | 13/21 [00:04<00:02,  3.11it/s]
 67%|██████▋   | 14/21 [00:04<00:02,  3.11it/s]
 71%|███████▏  | 15/21 [00:04<00:01,  3.11it/s]
 76%|███████▌  | 16/21 [00:05<00:01,  3.11it/s]
 81%|████████  | 17/21 [00:05<00:01,  3.11it/s]
 86%|████████▌ | 18/21 [00:05<00:00,  3.11it/s]
 90%|█████████ | 19/21 [00:06<00:00,  3.11it/s]
 95%|█████████▌| 20/21 [00:06<00:00,  3.11it/s]
100%|██████████| 21/21 [00:06<00:00,  3.11it/s]
100%|██████████| 21/21 [00:06<00:00,  3.07it/s]
Executing node 8, title: VAE Decode, class type: VAEDecode
Executing node 60, title: Save Image, class type: SaveImage
Prompt executed in 13.26 seconds
outputs:  {'60': {'images': [{'filename': 'outpainted_00001_.png', 'subfolder': '', 'type': 'output'}]}}
====================================
outpainted_00001_.png

This output was created using a different version of the model, emaph/outpaint-controlnet-union:4d11bfa6.

Examples

View more examples

Run time and cost

This model costs approximately $0.089 to run on Replicate, or 11 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 L40S GPU hardware. Predictions typically complete within 92 seconds. The predict time for this model varies significantly based on the inputs.

Readme

An image must be provided as input. Uses controlnet++ for image outpainting.

ControlNet++: All-in-one ControlNet for image generations and editing!

ProMax Model has released!! 12 control + 5 advanced editing, just try it!!!

Advantages about the model

Use bucket training like novelai, can generate high resolutions images of any aspect ratio
Use large amount of high quality data(over 10000000 images), the dataset covers a diversity of situation
Use re-captioned prompt like DALLE.3, use CogVLM to generate detailed description, good prompt following ability
Use many useful tricks during training. Including but not limited to date augmentation, mutiple loss, multi resolution
Use almost the same parameter compared with original ControlNet. No obvious increase in network parameter or computation.
Support 10+ control conditions, no obvious performance drop on any single condition compared with training independently
Support multi condition generation, condition fusion is learned during training. No need to set hyperparameter or design prompts.
Compatible with other opensource SDXL models, such as BluePencilXL, CounterfeitXL. Compatible with other Lora models.

Inference scripts and more details can found: https://github.com/xinsir6/ControlNetPlus/tree/main

emaph / outpaint-controlnet-union

Input

Output

Examples

Run time and cost

Readme

ControlNet++: All-in-one ControlNet for image generations and editing!

ProMax Model has released!! 12 control + 5 advanced editing, just try it!!!

Advantages about the model

Image Outpainting

Logs (sbx0zj2vmnrgm0cgz5pt274q5g)