delta-lock/ponynai3 – Run with an API on Replicate

delta-lock / ponynai3

Models fine-tuned from Pony-XL series.

Warm

Public
418.1K runs
H100
GitHub
Weights
License

Iterate in playground

Run with an API

Playground API Examples README Versions

Input

model

string

The model to use

Default: "tPonynai3_v7"

vae

string

The VAE to use

Default: "default"

prompt

string

Shift + Return to add a new line

The prompt, uses Compel weighting syntax

Default: "1girl"

image

file

The image for image to image or as the base for inpainting (will be scaled then cropped to the set width and height)

mask

file

The mask for inpainting, white areas will be modified and black preserved (will be scaled then cropped to the set width and height)

loras

string

Shift + Return to add a new line

The LoRAs to use, must be either a string with format "URL:Strength,URL:Strength,..." (strength is optional, default to 1), or a JSON list dumped as a string containing key "url" (required), "strength" (optional, default to 1), and "civitai_token" (optional, for downloading from CivitAI) (NOTICE: Will download the weights, might take a while if the LoRAs are huge or the download is slow, WILL CHARGE WHEN DOWNLOADING)

negative_prompt

string

Shift + Return to add a new line

The negative prompt (for things you don't want), uses Compel weighting syntax

Default: "animal, cat, dog, big breasts"

cfg_scale

number

(minimum: 1, maximum: 50)

CFG scale defines how much attention the model pays to the prompt when generating, set to 1 to disable

Default: 5

guidance_rescale

number

(minimum: 0, maximum: 5)

The amount to rescale CFG generated noise to avoid generating overexposed images, set to 0 or 1 to disable

Default: 0.5

pag_scale

number

(minimum: 0, maximum: 50)

PAG scale is similar to CFG but it literally makes the result better, it's compatible with CFG too, set to 0 to disable

Default: 3

clip_skip

integer

(minimum: 1)

How many CLIP layers to skip, 1 is actually no skip, this is the behavior in A1111 so it's aligned to it

Default: 1

width

integer

(minimum: 1, maximum: 4096)

The width of the image

Default: 1184

height

integer

(minimum: 1, maximum: 4096)

The height of the image

Default: 864

prepend_preprompt

boolean

Prepend preprompt (Prompt: "score_9, score_8_up, score_7_up, " | Negative Prompt: "score_4, score_3, score_2, score_1, worst quality, bad hands, bad feet, ")

Default: true

scheduler

string

The scheduler to use

Default: "Euler a"

steps

integer

(minimum: 1, maximum: 100)

The steps when generating

Default: 35

strength

number

(minimum: 0, maximum: 1)

How much noise to add (for image to image and inpainting only, larger value indicates more noise added to the input image)

Default: 0.7

blur_factor

number

(minimum: 0)

The factor to blur the inpainting mask for smoother transition between masked and unmasked

Default: 5

batch_size

integer

(minimum: 1, maximum: 4)

Number of images to generate (1-4), note if you set this to 4, some high resolution gens might fail because of not enough VRAM

Default: 1

seed

integer

The seed used when generating, set to -1 to use a random seed

Default: -1

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

npx create-replicate --model=delta-lock/ponynai3

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 delta-lock/ponynai3 using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

const output = await replicate.run(
  "delta-lock/ponynai3:848da0d3e5a762b8662592acd1818003a3b4672f513d7250895bd0d96c6a48c9",
  {
    input: {
      vae: "default",
      seed: -1,
      model: "tPonynai3_v7",
      steps: 35,
      width: 1184,
      height: 864,
      prompt: "1girl",
      strength: 0.7,
      cfg_scale: 5,
      clip_skip: 1,
      pag_scale: 3,
      scheduler: "Euler a",
      batch_size: 1,
      blur_factor: 5,
      negative_prompt: "animal, cat, dog, big breasts",
      guidance_rescale: 0.5,
      prepend_preprompt: true
    }
  }
);

// 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 delta-lock/ponynai3 using Replicate’s API. Check out the model's schema for an overview of inputs and outputs.

output = replicate.run(
    "delta-lock/ponynai3:848da0d3e5a762b8662592acd1818003a3b4672f513d7250895bd0d96c6a48c9",
    input={
        "vae": "default",
        "seed": -1,
        "model": "tPonynai3_v7",
        "steps": 35,
        "width": 1184,
        "height": 864,
        "prompt": "1girl",
        "strength": 0.7,
        "cfg_scale": 5,
        "clip_skip": 1,
        "pag_scale": 3,
        "scheduler": "Euler a",
        "batch_size": 1,
        "blur_factor": 5,
        "negative_prompt": "animal, cat, dog, big breasts",
        "guidance_rescale": 0.5,
        "prepend_preprompt": True
    }
)
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 delta-lock/ponynai3 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": "delta-lock/ponynai3:848da0d3e5a762b8662592acd1818003a3b4672f513d7250895bd0d96c6a48c9",
    "input": {
      "vae": "default",
      "seed": -1,
      "model": "tPonynai3_v7",
      "steps": 35,
      "width": 1184,
      "height": 864,
      "prompt": "1girl",
      "strength": 0.7,
      "cfg_scale": 5,
      "clip_skip": 1,
      "pag_scale": 3,
      "scheduler": "Euler a",
      "batch_size": 1,
      "blur_factor": 5,
      "negative_prompt": "animal, cat, dog, big breasts",
      "guidance_rescale": 0.5,
      "prepend_preprompt": true
    }
  }' \
  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/delta-lock/ponynai3@sha256:848da0d3e5a762b8662592acd1818003a3b4672f513d7250895bd0d96c6a48c9 \
  -i 'vae="default"' \
  -i 'seed=-1' \
  -i 'model="tPonynai3_v7"' \
  -i 'steps=35' \
  -i 'width=1184' \
  -i 'height=864' \
  -i 'prompt="1girl"' \
  -i 'strength=0.7' \
  -i 'cfg_scale=5' \
  -i 'clip_skip=1' \
  -i 'pag_scale=3' \
  -i 'scheduler="Euler a"' \
  -i 'batch_size=1' \
  -i 'blur_factor=5' \
  -i 'negative_prompt="animal, cat, dog, big breasts"' \
  -i 'guidance_rescale=0.5' \
  -i 'prepend_preprompt=true'

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/delta-lock/ponynai3@sha256:848da0d3e5a762b8662592acd1818003a3b4672f513d7250895bd0d96c6a48c9
curl -s -X POST \
  -H "Content-Type: application/json" \
  -d $'{
    "input": {
      "vae": "default",
      "seed": -1,
      "model": "tPonynai3_v7",
      "steps": 35,
      "width": 1184,
      "height": 864,
      "prompt": "1girl",
      "strength": 0.7,
      "cfg_scale": 5,
      "clip_skip": 1,
      "pag_scale": 3,
      "scheduler": "Euler a",
      "batch_size": 1,
      "blur_factor": 5,
      "negative_prompt": "animal, cat, dog, big breasts",
      "guidance_rescale": 0.5,
      "prepend_preprompt": true
    }
  }' \
  http://localhost:5000/predictions

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

Output

{
  "completed_at": "2025-05-11T10:55:47.000294Z",
  "created_at": "2025-05-11T10:53:38.905000Z",
  "data_removed": false,
  "error": null,
  "id": "gc73rwn2v5rmc0cpqvxad1ep18",
  "input": {
    "vae": "default",
    "seed": -1,
    "model": "tPonynai3_v7",
    "steps": 35,
    "width": 1184,
    "height": 864,
    "prompt": "1girl",
    "strength": 0.7,
    "cfg_scale": 5,
    "clip_skip": 1,
    "pag_scale": 3,
    "scheduler": "Euler a",
    "batch_size": 1,
    "blur_factor": 5,
    "negative_prompt": "animal, cat, dog, big breasts",
    "guidance_rescale": 0.5,
    "prepend_preprompt": true
  },
  "logs": "Using text to image mode.\nUsing seed: 421887767\n  0%|          | 0/35 [00:00<?, ?it/s]\n  3%|▎         | 1/35 [00:00<00:08,  4.21it/s]\n  9%|▊         | 3/35 [00:00<00:04,  7.09it/s]\n 11%|█▏        | 4/35 [00:00<00:04,  7.01it/s]\n 14%|█▍        | 5/35 [00:00<00:04,  6.97it/s]\n 17%|█▋        | 6/35 [00:00<00:04,  6.93it/s]\n 20%|██        | 7/35 [00:01<00:04,  6.89it/s]\n 23%|██▎       | 8/35 [00:01<00:03,  6.87it/s]\n 26%|██▌       | 9/35 [00:01<00:03,  6.86it/s]\n 29%|██▊       | 10/35 [00:01<00:03,  6.85it/s]\n 31%|███▏      | 11/35 [00:01<00:03,  6.84it/s]\n 34%|███▍      | 12/35 [00:01<00:03,  6.82it/s]\n 37%|███▋      | 13/35 [00:01<00:03,  6.82it/s]\n 40%|████      | 14/35 [00:02<00:03,  6.83it/s]\n 43%|████▎     | 15/35 [00:02<00:02,  6.81it/s]\n 46%|████▌     | 16/35 [00:02<00:02,  6.80it/s]\n 49%|████▊     | 17/35 [00:02<00:02,  6.80it/s]\n 51%|█████▏    | 18/35 [00:02<00:02,  6.80it/s]\n 54%|█████▍    | 19/35 [00:02<00:02,  6.80it/s]\n 57%|█████▋    | 20/35 [00:02<00:02,  6.80it/s]\n 60%|██████    | 21/35 [00:03<00:02,  6.79it/s]\n 63%|██████▎   | 22/35 [00:03<00:01,  6.78it/s]\n 66%|██████▌   | 23/35 [00:03<00:01,  6.79it/s]\n 69%|██████▊   | 24/35 [00:03<00:01,  6.78it/s]\n 71%|███████▏  | 25/35 [00:03<00:01,  6.78it/s]\n 74%|███████▍  | 26/35 [00:03<00:01,  6.77it/s]\n 77%|███████▋  | 27/35 [00:03<00:01,  6.77it/s]\n 80%|████████  | 28/35 [00:04<00:01,  6.77it/s]\n 83%|████████▎ | 29/35 [00:04<00:00,  6.76it/s]\n 86%|████████▌ | 30/35 [00:04<00:00,  6.76it/s]\n 89%|████████▊ | 31/35 [00:04<00:00,  6.75it/s]\n 91%|█████████▏| 32/35 [00:04<00:00,  6.75it/s]\n 94%|█████████▍| 33/35 [00:04<00:00,  6.74it/s]\n 97%|█████████▋| 34/35 [00:05<00:00,  6.74it/s]\n100%|██████████| 35/35 [00:05<00:00,  6.74it/s]\n100%|██████████| 35/35 [00:05<00:00,  6.77it/s]",
  "metrics": {
    "predict_time": 7.55949215,
    "total_time": 128.095294
  },
  "output": [
    "https://replicate.delivery/xezq/Z0fpC39b1awejU1b0CqGnleieJnyd0p1ZHN7WW0KINJLDNuSB/0.png"
  ],
  "started_at": "2025-05-11T10:55:39.440801Z",
  "status": "succeeded",
  "urls": {
    "stream": "https://stream.replicate.com/v1/files/bcwr-j3tpiviwrgh635l2dfeusegomwllkku43hbxoaf7zamfkrnbdcua",
    "get": "https://api.replicate.com/v1/predictions/gc73rwn2v5rmc0cpqvxad1ep18",
    "cancel": "https://api.replicate.com/v1/predictions/gc73rwn2v5rmc0cpqvxad1ep18/cancel"
  },
  "version": "848da0d3e5a762b8662592acd1818003a3b4672f513d7250895bd0d96c6a48c9"
}

Generated in

7.6 seconds

Tweak it ShareReport

Using text to image mode.
Using seed: 421887767
  0%|          | 0/35 [00:00<?, ?it/s]
  3%|▎         | 1/35 [00:00<00:08,  4.21it/s]
  9%|▊         | 3/35 [00:00<00:04,  7.09it/s]
 11%|█▏        | 4/35 [00:00<00:04,  7.01it/s]
 14%|█▍        | 5/35 [00:00<00:04,  6.97it/s]
 17%|█▋        | 6/35 [00:00<00:04,  6.93it/s]
 20%|██        | 7/35 [00:01<00:04,  6.89it/s]
 23%|██▎       | 8/35 [00:01<00:03,  6.87it/s]
 26%|██▌       | 9/35 [00:01<00:03,  6.86it/s]
 29%|██▊       | 10/35 [00:01<00:03,  6.85it/s]
 31%|███▏      | 11/35 [00:01<00:03,  6.84it/s]
 34%|███▍      | 12/35 [00:01<00:03,  6.82it/s]
 37%|███▋      | 13/35 [00:01<00:03,  6.82it/s]
 40%|████      | 14/35 [00:02<00:03,  6.83it/s]
 43%|████▎     | 15/35 [00:02<00:02,  6.81it/s]
 46%|████▌     | 16/35 [00:02<00:02,  6.80it/s]
 49%|████▊     | 17/35 [00:02<00:02,  6.80it/s]
 51%|█████▏    | 18/35 [00:02<00:02,  6.80it/s]
 54%|█████▍    | 19/35 [00:02<00:02,  6.80it/s]
 57%|█████▋    | 20/35 [00:02<00:02,  6.80it/s]
 60%|██████    | 21/35 [00:03<00:02,  6.79it/s]
 63%|██████▎   | 22/35 [00:03<00:01,  6.78it/s]
 66%|██████▌   | 23/35 [00:03<00:01,  6.79it/s]
 69%|██████▊   | 24/35 [00:03<00:01,  6.78it/s]
 71%|███████▏  | 25/35 [00:03<00:01,  6.78it/s]
 74%|███████▍  | 26/35 [00:03<00:01,  6.77it/s]
 77%|███████▋  | 27/35 [00:03<00:01,  6.77it/s]
 80%|████████  | 28/35 [00:04<00:01,  6.77it/s]
 83%|████████▎ | 29/35 [00:04<00:00,  6.76it/s]
 86%|████████▌ | 30/35 [00:04<00:00,  6.76it/s]
 89%|████████▊ | 31/35 [00:04<00:00,  6.75it/s]
 91%|█████████▏| 32/35 [00:04<00:00,  6.75it/s]
 94%|█████████▍| 33/35 [00:04<00:00,  6.74it/s]
 97%|█████████▋| 34/35 [00:05<00:00,  6.74it/s]
100%|██████████| 35/35 [00:05<00:00,  6.74it/s]
100%|██████████| 35/35 [00:05<00:00,  6.77it/s]

Examples

View more examples

Run time and cost

This model costs approximately $0.073 to run on Replicate, or 13 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 H100 GPU hardware. Predictions typically complete within 49 seconds. The predict time for this model varies significantly based on the inputs.

Readme

NOTE: In favour of NoobAI-XL/Illustrious-XL series models which have better quality, this repo will most likely NOT receive any updates anymore. Please switch to delta-lock/noobai-xl if you can.