Name: RealVisXL V3 API: Multi-ControlNet LoRA Image Generator
Brand: magicapi
SKU: api-magicapi-ai-image-editing-and-generation-api-sdxl-based-realvisxl-v3-multi-controlnet-lora
Availability: InStock
Author: API Market

Shashank Agarwal

# 🖼️ Multi-ControlNet Image Generator API (sdxl-based RealVisXL V3 Multi-ControlNet LoRA)

## Overview
RealVisXL V3 Multi-ControlNet LoRA generates images with the RealVisXL V3 model, an SDXL-based checkpoint. It supports text-to-image, image-to-image (img2img), inpainting with a mask, LoRA loading, and up to three ControlNets at once (edges, depth, line art, pose and more).

Send a `prompt`, optionally with an input `image`, a `mask` and up to three ControlNet images. Each request returns a list of image URLs: the processed control image for each active ControlNet, followed by the generated images.

## Features
- **Three modes**: text-to-image with `prompt` only, img2img with `image` and `prompt_strength`, and inpainting with `image` plus `mask` (black areas are kept, white areas are repainted).
- **Up to three ControlNets**: `controlnet_1`, `controlnet_2` and `controlnet_3` each take one of `none`, `edge_canny`, `illusion`, `depth_leres`, `depth_midas`, `soft_edge_pidi`, `soft_edge_hed`, `lineart`, `lineart_anime`, `openpose`, with their own image, conditioning scale (0 to 4) and start and end points (0 to 1).
- **Sizing**: set `width` and `height`, or size the output from the input image, a ControlNet image or the mask with `sizing_strategy`.
- **Up to 4 images per request** with `num_outputs`.
- **Sampling controls**: `scheduler` (7 options), `num_inference_steps` (1 to 500), `guidance_scale` (1 to 50), `seed`, and an optional `base_image_refiner` pass.
- **LoRA**: `lora_weights` and `lora_scale` (0 to 1).

## Use Cases
1. **Pose- or layout-matched images**: keep the pose, depth or outlines of a reference image while changing the subject and style.
2. **Photo variations**: restyle an existing photo with img2img.
3. **Inpainting**: replace part of an image by masking it and describing the new content.
4. **Combined control**: mix, for example, an `openpose` and a `depth_midas` ControlNet to control both pose and scene depth.

## Visual Examples
### Example Input
ControlNet 1 image (`controlnet_1` `soft_edge_hed`):

![Reference image sent as controlnet_1_image in the RealVisXL V3 example request](https://blog.api.market/wp-content/uploads/2026/10/realvisxl-v3-multi-controlnet-lora-input-api-market-1.png)

Prompt: "A detailed photo of an astronaut riding a unicorn through a field of flowers"

Settings: `width` `768`, `height` `768`, `sizing_strategy` `width_height`, `num_outputs` `1`, `scheduler` `K_EULER`, `num_inference_steps` `30`, `guidance_scale` `7.5`, `prompt_strength` `0.8`, `refine` `no_refiner`, `lora_scale` `0.8`, `apply_watermark` `false`, `negative_prompt` `""`, `controlnet_1_conditioning_scale` `0.8`, `controlnet_1_start` `0`, `controlnet_1_end` `1`, `controlnet_2` and `controlnet_3` `none`.

### Example Output
The response listed two images. First, the soft-edge control image made from the ControlNet 1 input:

![Soft edge HED control image produced from the example input](https://blog.api.market/wp-content/uploads/2026/10/realvisxl-v3-multi-controlnet-lora-output-1-gcx0y6fv-api-market.png)

Then the generated image:

![Astronaut riding a unicorn through a field of flowers generated by RealVisXL V3](https://blog.api.market/wp-content/uploads/2026/10/realvisxl-v3-multi-controlnet-lora-output-2-gcx0y6fv-api-market.jpg)

## Input Parameters
| Field | Type | Required | Default | Allowed values | Description |
| --- | --- | --- | --- | --- | --- |
| `prompt` | string | No | `An astronaut riding a rainbow unicorn` |  | Input prompt |
| `negative_prompt` | string | No | `""` |  | Negative prompt |
| `image` | string (URL) | No | none |  | Input image for img2img or inpaint mode |
| `mask` | string (URL) | No | none |  | Input mask for inpaint mode. Black areas will be preserved, white areas will be inpainted |
| `width` | integer | No | `768` |  | Width of output image |
| `height` | integer | No | `768` |  | Height of output image |
| `sizing_strategy` | string | No | `width_height` | `width_height`, `input_image`, `controlnet_1_image`, `controlnet_2_image`, `controlnet_3_image`, `mask_image` | How to size the output: use width and height, or resize based on the input image, a ControlNet image or the mask |
| `num_outputs` | integer | No | `1` | 1 to 4 | Number of images to output |
| `scheduler` | string | No | `K_EULER` | `DDIM`, `DPMSolverMultistep`, `HeunDiscrete`, `KarrasDPM`, `K_EULER_ANCESTRAL`, `K_EULER`, `PNDM` | Scheduler |
| `num_inference_steps` | integer | No | `30` | 1 to 500 | Number of denoising steps |
| `guidance_scale` | number | No | `7.5` | 1 to 50 | Scale for classifier-free guidance |
| `prompt_strength` | number | No | `0.8` | 0 to 1 | Prompt strength when using img2img or inpaint. 1.0 corresponds to full destruction of information in the image |
| `seed` | integer | No | none |  | Random seed. Leave blank to randomize the seed |
| `refine` | string | No | `no_refiner` | `no_refiner`, `base_image_refiner` | Which refine style to use |
| `refine_steps` | integer | No | none |  | For `base_image_refiner`, the number of steps to refine; defaults to `num_inference_steps` |
| `apply_watermark` | boolean | No | `false` |  | Applies a watermark so downstream applications can tell the image was generated. Leave it off if you have other provisions for generating or deploying images safely |
| `lora_scale` | number | No | `0.6` | 0 to 1 | LoRA additive scale. Only applicable on trained models |
| `lora_weights` | string | No | none |  | LoRA weights to use. Leave blank to use the default weights |
| `disable_safety_checker` | boolean | No | `false` |  | Disable safety checker for generated images |
| `controlnet_1` | string | No | `none` | `none`, `edge_canny`, `illusion`, `depth_leres`, `depth_midas`, `soft_edge_pidi`, `soft_edge_hed`, `lineart`, `lineart_anime`, `openpose` | First ControlNet |
| `controlnet_1_image` | string (URL) | No | none |  | Input image for the first ControlNet |
| `controlnet_1_conditioning_scale` | number | No | `0.75` | 0 to 4 | How strong the ControlNet conditioning is |
| `controlnet_1_start` | number | No | `0` | 0 to 1 | When ControlNet conditioning starts |
| `controlnet_1_end` | number | No | `1` | 0 to 1 | When ControlNet conditioning ends |
| `controlnet_2` | string | No | `none` | same values as `controlnet_1` | Second ControlNet |
| `controlnet_2_image` | string (URL) | No | none |  | Input image for the second ControlNet |
| `controlnet_2_conditioning_scale` | number | No | `0.75` | 0 to 4 | How strong the ControlNet conditioning is |
| `controlnet_2_start` | number | No | `0` | 0 to 1 | When ControlNet conditioning starts |
| `controlnet_2_end` | number | No | `1` | 0 to 1 | When ControlNet conditioning ends |
| `controlnet_3` | string | No | `none` | same values as `controlnet_1` | Third ControlNet |
| `controlnet_3_image` | string (URL) | No | none |  | Input image for the third ControlNet |
| `controlnet_3_conditioning_scale` | number | No | `0.75` | 0 to 4 | How strong the ControlNet conditioning is |
| `controlnet_3_start` | number | No | `0` | 0 to 1 | When ControlNet conditioning starts |
| `controlnet_3_end` | number | No | `1` | 0 to 1 | When ControlNet conditioning ends |

## Pricing
- **Pay As You Go (PAYG) pricing**: you only pay for what you use.
- **Price**: $0.015 per API unit.
- **1 API unit = 1 successful request with up to 13 seconds of processing time.** A longer request counts one more unit per started 13 seconds of processing, so more `num_outputs`, larger sizes, more `num_inference_steps`, extra ControlNets or the refiner can cost more.
- The example request above took about 5 seconds of processing, so it was billed as 1 unit and costs $0.015.
- **Failed or cancelled predictions are not billed.**
- **GET requests to check prediction status** are free and don't count toward API unit usage.
- **No monthly subscription fee.**

## API Documentation

### Base URL
All API requests go to: `https://prod.api.market/api/v1/magicapi/ai-image-editing-and-generation-api-sdxl-based-realvisxl-v3-multi-controlnet-lora`

### Authentication
Send your api.market API key in the `x-api-market-key` header:
```
x-api-market-key: YOUR_API_MARKET_KEY
```

### Asynchronous Processing
**IMPORTANT**: This API is asynchronous. The flow is:
1. Send a POST request to `/predictions`. A successful request returns **201 Created** with a prediction `id` and a `status` of `starting` or `processing`.
2. Poll GET `/predictions/{id}` with that ID.
3. Keep polling while `status` is `starting` or `processing`.
4. Stop when `status` is `succeeded` (read `output`), `failed` or `canceled` (read `error`).

A polling interval of 1 second and an overall timeout of 600 seconds (10 minutes) work well.

### Endpoints

#### Create a Prediction
```bash
curl -X POST "https://prod.api.market/api/v1/magicapi/ai-image-editing-and-generation-api-sdxl-based-realvisxl-v3-multi-controlnet-lora/predictions" \
  -H "Content-Type: application/json" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY" \
  -d '{
    "version": "90a4a3604cd637cb9f1a2bdae1cfa9ed869362ca028814cdce310a78e27daade",
    "input": {
      "width": 768,
      "height": 768,
      "prompt": "A detailed photo of an astronaut riding a unicorn through a field of flowers",
      "refine": "no_refiner",
      "scheduler": "K_EULER",
      "lora_scale": 0.8,
      "num_outputs": 1,
      "controlnet_1": "soft_edge_hed",
      "controlnet_2": "none",
      "controlnet_3": "none",
      "guidance_scale": 7.5,
      "apply_watermark": false,
      "negative_prompt": "",
      "prompt_strength": 0.8,
      "sizing_strategy": "width_height",
      "controlnet_1_end": 1,
      "controlnet_2_end": 1,
      "controlnet_3_end": 1,
      "controlnet_1_image": "https://blog.api.market/wp-content/uploads/2026/10/realvisxl-v3-multi-controlnet-lora-input-api-market-1.png",
      "controlnet_1_start": 0,
      "controlnet_2_start": 0,
      "controlnet_3_start": 0,
      "num_inference_steps": 30,
      "controlnet_1_conditioning_scale": 0.8,
      "controlnet_2_conditioning_scale": 0.8,
      "controlnet_3_conditioning_scale": 0.75
    }
  }'
```

Response (201 Created):
```json
{
  "id": "gcx0y6fvc1rmw0d1ccxtmgrk8w",
  "version": "90a4a3604cd637cb9f1a2bdae1cfa9ed869362ca028814cdce310a78e27daade",
  "status": "starting",
  "created_at": "2026-10-22T10:50:37.969Z"
}
```

#### Check Prediction Status
```bash
curl -X GET "https://prod.api.market/api/v1/magicapi/ai-image-editing-and-generation-api-sdxl-based-realvisxl-v3-multi-controlnet-lora/predictions/{id}" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY"
```

Response when the prediction has finished (200 OK):
```json
{
  "id": "gcx0y6fvc1rmw0d1ccxtmgrk8w",
  "status": "succeeded",
  "output": [
    "https://blog.api.market/wp-content/uploads/2026/10/realvisxl-v3-multi-controlnet-lora-output-1-gcx0y6fv-api-market.png",
    "https://blog.api.market/wp-content/uploads/2026/10/realvisxl-v3-multi-controlnet-lora-output-2-gcx0y6fv-api-market.jpg"
  ],
  "error": null,
  "created_at": "2026-10-22T10:50:37.969Z",
  "started_at": "2026-10-22T10:51:35.923Z",
  "completed_at": "2026-10-22T10:51:40.877Z"
}
```

`output` is a list of image URLs: first one processed control image per active ControlNet, then the generated images (`num_outputs` of them).

### Status Codes
- **201 Created**: prediction created (POST `/predictions`). This is the normal success response.
- **200 OK**: prediction details returned (GET `/predictions/{id}`).
- **400 Bad Request**: invalid input, for example a `controlnet_1` value that is not in the list or a `num_outputs` above 4.
- **401 Unauthorized**: missing or invalid API key.
- **404 Not Found**: no prediction with that ID.
- **429 Too Many Requests**: rate limit exceeded.
- **500 Internal Server Error** or **503 Service Unavailable**: temporary server problem; retry later.

## Rate Limits
All plans allow 10 requests per second. Requests above that return `429 Too Many Requests`; back off before retrying and honour any `Retry-After` header. Polling a prediction with GET is free but still counts toward the per-second limit, so poll every 1 to 5 seconds rather than in a tight loop.

## Error Handling
- Check the HTTP status code of every request and read the response body for the error message.
- A prediction can also end with `status` set to `failed`; the `error` field then explains why. Failed predictions are not billed.
- Retry 429, 500 and 503 responses with exponential backoff. Fix the input before retrying a 400.

## Implementation Guide
### Example Code

#### JavaScript
```javascript
const apiKey = "YOUR_API_MARKET_KEY";
const baseUrl = "https://prod.api.market/api/v1/magicapi/ai-image-editing-and-generation-api-sdxl-based-realvisxl-v3-multi-controlnet-lora";

async function createAndWaitForPrediction(input) {
  const response = await fetch(`${baseUrl}/predictions`, {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "x-api-market-key": apiKey
    },
    body: JSON.stringify({ version: "90a4a3604cd637cb9f1a2bdae1cfa9ed869362ca028814cdce310a78e27daade", input })
  });

  if (response.status !== 201) {
    throw new Error(`API error: ${response.status} ${await response.text()}`);
  }

  const prediction = await response.json();
  return pollForResults(prediction.id);
}

const delay = (ms) => new Promise((resolve) => setTimeout(resolve, ms));

async function pollForResults(predictionId) {
  const maxTimeout = 600000;
  const pollInterval = 1000;
  const startTime = Date.now();

  while (Date.now() - startTime < maxTimeout) {
    const response = await fetch(`${baseUrl}/predictions/${predictionId}`, {
      headers: { "x-api-market-key": apiKey }
    });

    if (!response.ok) {
      throw new Error(`API error: ${response.status} ${await response.text()}`);
    }

    const prediction = await response.json();

    if (prediction.status === "succeeded") {
      return prediction.output;
    }

    if (prediction.status === "failed" || prediction.status === "canceled") {
      throw new Error(`Prediction ${prediction.status}: ${prediction.error || "Unknown error"}`);
    }

    await delay(pollInterval);
  }

  throw new Error("Prediction timed out after 10 minutes");
}

createAndWaitForPrediction({
  width: 768,
  height: 768,
  prompt: "A detailed photo of an astronaut riding a unicorn through a field of flowers",
  refine: "no_refiner",
  scheduler: "K_EULER",
  lora_scale: 0.8,
  num_outputs: 1,
  controlnet_1: "soft_edge_hed",
  controlnet_2: "none",
  controlnet_3: "none",
  guidance_scale: 7.5,
  apply_watermark: false,
  negative_prompt: "",
  prompt_strength: 0.8,
  sizing_strategy: "width_height",
  controlnet_1_end: 1,
  controlnet_2_end: 1,
  controlnet_3_end: 1,
  controlnet_1_image: "https://blog.api.market/wp-content/uploads/2026/10/realvisxl-v3-multi-controlnet-lora-input-api-market-1.png",
  controlnet_1_start: 0,
  controlnet_2_start: 0,
  controlnet_3_start: 0,
  num_inference_steps: 30,
  controlnet_1_conditioning_scale: 0.8,
  controlnet_2_conditioning_scale: 0.8,
  controlnet_3_conditioning_scale: 0.75
}).then((output) => console.log(output));
```

#### Python
```python
import time
import requests

API_KEY = "YOUR_API_MARKET_KEY"
BASE_URL = "https://prod.api.market/api/v1/magicapi/ai-image-editing-and-generation-api-sdxl-based-realvisxl-v3-multi-controlnet-lora"


def create_and_wait_for_prediction(input_data):
    headers = {"Content-Type": "application/json", "x-api-market-key": API_KEY}
    response = requests.post(
        f"{BASE_URL}/predictions",
        headers=headers,
        json={"version": "90a4a3604cd637cb9f1a2bdae1cfa9ed869362ca028814cdce310a78e27daade", "input": input_data},
    )

    if response.status_code != 201:
        raise Exception(f"API error: {response.status_code} {response.text}")

    return poll_for_results(response.json()["id"])


def poll_for_results(prediction_id):
    max_timeout = 600
    poll_interval = 1
    headers = {"x-api-market-key": API_KEY}
    start_time = time.time()

    while time.time() - start_time < max_timeout:
        response = requests.get(f"{BASE_URL}/predictions/{prediction_id}", headers=headers)

        if response.status_code != 200:
            raise Exception(f"API error: {response.status_code} {response.text}")

        prediction = response.json()

        if prediction["status"] == "succeeded":
            return prediction["output"]

        if prediction["status"] in ("failed", "canceled"):
            raise Exception(f"Prediction {prediction['status']}: {prediction.get('error') or 'Unknown error'}")

        time.sleep(poll_interval)

    raise Exception("Prediction timed out after 10 minutes")


print(create_and_wait_for_prediction({
    "width": 768,
    "height": 768,
    "prompt": "A detailed photo of an astronaut riding a unicorn through a field of flowers",
    "refine": "no_refiner",
    "scheduler": "K_EULER",
    "lora_scale": 0.8,
    "num_outputs": 1,
    "controlnet_1": "soft_edge_hed",
    "controlnet_2": "none",
    "controlnet_3": "none",
    "guidance_scale": 7.5,
    "apply_watermark": False,
    "negative_prompt": "",
    "prompt_strength": 0.8,
    "sizing_strategy": "width_height",
    "controlnet_1_end": 1,
    "controlnet_2_end": 1,
    "controlnet_3_end": 1,
    "controlnet_1_image": "https://blog.api.market/wp-content/uploads/2026/10/realvisxl-v3-multi-controlnet-lora-input-api-market-1.png",
    "controlnet_1_start": 0,
    "controlnet_2_start": 0,
    "controlnet_3_start": 0,
    "num_inference_steps": 30,
    "controlnet_1_conditioning_scale": 0.8,
    "controlnet_2_conditioning_scale": 0.8,
    "controlnet_3_conditioning_scale": 0.75,
}))
```

## FAQ
- **Why does the output contain more images than `num_outputs`?**
  - Each active ControlNet adds its processed control image (for example the soft-edge map) at the start of the list. The generated images follow.
- **How do I inpaint part of an image?**
  - Send the photo as `image` and a `mask` where white marks the area to repaint and black the area to keep, then describe the new content in `prompt`. `prompt_strength` sets how much the masked area changes.
- **Can I use ControlNets without an input image?**
  - Yes. Set `controlnet_1` to a type such as `openpose` and pass `controlnet_1_image`; leave `image` empty for text-to-image with control.
- **How much does one request cost?**
  - $0.015 for each request with up to 13 seconds of processing. The example took about 5 seconds.
- **Is the 201 response from POST an error?**
  - No. 201 Created means the prediction was accepted. Use the returned `id` to poll GET `/predictions/{id}` until `status` is `succeeded`, `failed` or `canceled`.
- **What if my prediction fails?**
  - Failed predictions are not billed. Read the `error` field, adjust the input and try again.

## Contact Us
If you have any questions, reach out to us through the chat on the bottom right of the page. You can also email our support team at hello@magicapi.com.


Text-to-image, img2img and inpainting with RealVisXL V3, plus LoRA and up to three ControlNets (canny, depth, line art, pose). Pay as you go, no monthly fee.