Ultimate SD Upscale

ultimate-sd-upscale

Ultimate SD Upscale is a Face Enhancer AI model that improves facial details and enhances image quality.

A100 80GB

Fast Inference

REST API

Try in Console API Docs Examples

Model Information

Response Time~40 sec

StatusActive

Version

0.0.1

Updated28 days ago

API Reference

View Full Documentation

Prerequisites

Create an API Key from the Eachlabs Console
Install the required dependencies for your chosen language (e.g., requests for Python)

API Integration Steps

1. Create a Prediction

Send a POST request to create a new prediction. This will return a prediction ID that you'll use to check the result. The request should include your model inputs and API key.

import requests
import time

API_KEY = "YOUR_API_KEY"  # Replace with your API key
HEADERS = {
    "X-API-Key": API_KEY,
    "Content-Type": "application/json"
}

def create_prediction():
    response = requests.post(
        "https://api.eachlabs.ai/v1/prediction/",
        headers=HEADERS,
        json={
            "model": "ultimate-sd-upscale",
            "version": "0.0.1",
            "input": {
  "seam_fix_padding": "16",
  "seed": null,
  "denoise": "0.2",
  "mask_blur": "8",
  "tile_width": "512",
  "upscale_by": "2",
  "tile_height": "512",
  "mode_type": "Linear",
  "upscaler": "4x-UltraSharp",
  "steps": "20",
  "cfg": "8",
  "scheduler": "normal",
  "tile_padding": "32",
  "seam_fix_width": "64",
  "seam_fix_denoise": "1",
  "seam_fix_mask_blur": "8",
  "controlnet_strength": "1",
  "force_uniform_tiles": "True",
  "use_controlnet_tile": "True",
  "seam_fix_mode": "None",
  "image": "your_file.image/jpeg",
  "negative_prompt": "your negative prompt here",
  "positive_prompt": "your positive prompt here",
  "sampler_name": "euler"
}
        }
    )
    prediction = response.json()
    
    if prediction["status"] != "success":
        raise Exception(f"Prediction failed: {prediction}")
    
    return prediction["predictionID"]

2. Get Prediction Result

Poll the prediction endpoint with the prediction ID until the result is ready. The API uses long-polling, so you'll need to repeatedly check until you receive a success status.

def get_prediction(prediction_id):
    while True:
        result = requests.get(
            f"https://api.eachlabs.ai/v1/prediction/{prediction_id}",
            headers=HEADERS
        ).json()
        
        if result["status"] == "success":
            return result
        elif result["status"] == "error":
            raise Exception(f"Prediction failed: {result}")
        
        time.sleep(1)  # Wait before polling again

3. Complete Example

Here's a complete example that puts it all together, including error handling and result processing. This shows how to create a prediction and wait for the result in a production environment.

try:
    # Create prediction
    prediction_id = create_prediction()
    print(f"Prediction created: {prediction_id}")
    
    # Get result
    result = get_prediction(prediction_id)
    print(f"Output URL: {result['output']}")
    print(f"Processing time: {result['metrics']['predict_time']}s")
except Exception as e:
    print(f"Error: {e}")

Additional Information

The API uses a two-step process: create prediction and poll for results
Response time: ~40 seconds
Rate limit: 60 requests/minute
Concurrent requests: 10 maximum
Use long-polling to check prediction status until completion

Overview

Ultimate SD Upscale is designed for high-quality image upscaling while maintaining fine details and improving overall image fidelity. With advanced options for tile-based upscaling, ControlNet integration, and seamless blending, it allows users to upscale images with precision, making it ideal for creative projects requiring enhanced visuals.

Technical Specifications

Upscaling Methods: Includes multiple upscaling algorithms like 4x_NMKD-Siax_200k and RealESRGAN_x4plus, each optimized for different use cases (e.g., anime, sharp details).

Tile Handling: Supports customizable tile sizes and padding to ensure memory efficiency and seamless image blending.

ControlNet Integration: Adds structure-preserving capabilities, adjustable through the controlnet_strength parameter.

Flexible Sampling and Scheduling: Offers advanced sampling techniques (e.g., dpmpp_2m, euler_ancestral) and scheduling options (karras, exponential) for fine control over the upscaling process.

Key Considerations

Resource Usage: Larger images or higher tile overlap settings may increase processing time and resource consumption.

Tile Seams: Ensure appropriate seam-fix settings to avoid visible tile boundaries in the final output.

Prompt Optimization: Misaligned prompts can lead to unintended artifacts. Carefully curate positive and negative prompts.

ControlNet Activation: ControlNet can introduce noise if improperly calibrated. Adjust the strength based on the input image and desired output.

Tips & Tricks

Input-Specific Adjustments

Upscaler:
- Use 4x_NMKD-Siax_200k for highly detailed imagery.
- Choose RealESRGAN_x4plus_anime_6B for anime-style images.
- Select 4x-UltraSharp for photos requiring enhanced sharpness.
Sampler Name:
- dpmpp_2m and dpmpp_2m_sde are recommended for smooth and balanced outputs.
- euler_ancestral provides a softer, artistic finish.
Scheduler:
- Use karras for natural-looking results with gradual transitions.
- exponential is suited for sharper, high-contrast images.
Mode Type:
- Select Linear for consistent upscaling across the image.
- Use Chess for complex, tiled designs to minimize memory usage.
Tile Dimensions and Padding:
- Start with a tile width and height of 512x512 for most cases.
- Increase padding to 32 for better blending at tile edges.

General Settings

Positive and Negative Prompts:
- Be explicit in positive prompts to emphasize desired features (e.g., "highly detailed, sharp textures").
- Use negative prompts to avoid unwanted artifacts (e.g., "no blurriness, no distorted edges").
ControlNet:
- For intricate details, set controlnet_strength between 0.5 and 0.7.
- Activate Use ControlNet Tile for highly structured or grid-like images.
Denoise and CFG Scale:
- Keep denoise between 0.4 and 0.6 to preserve image structure while reducing unwanted noise.
- Set cfg to 7.0 for balanced creativity and prompt adherence.
Seam Fix Options:
- Use Half Tile + Intersections for seamless transitions in tiled outputs.
- Adjust seam_fix_denoise to 0.3 and seam_fix_mask_blur to 8 for better blending.

Capabilities

Upscales images up to 4x their original resolution.

Supports tile-based processing for high-resolution images.

Integrates seamlessly with ControlNet for structure-aware upscaling.

Flexible parameter options for fine-tuning outputs.

What can I use for?

Enhance image quality for digital art, photography, or animations.

Upscale anime and other stylized visuals while preserving intricate details.

Generate high-resolution assets for professional and creative projects.

Things to be aware of

Use Linear mode and 4x-UltraSharp for realistic photographs requiring sharpness.

Test tiled upscaling with a tile size of 256x256 and seam fix enabled for large images.

Activate ControlNet for architectural or grid-like images, adjusting controlnet_strength to preserve details.

Limitations

Processing Time: Larger images or high tile overlap may lead to slower processing.

Tile Artifacts: Improper settings may leave visible seams or artifacts in tiled outputs.

Prompt Dependency: Results are highly dependent on prompt accuracy and clarity.

ControlNet Overuse: Excessive ControlNet strength may distort image structure.

Output Format:PNG

Related AI Models

Image Upscaler by Each AI

each-upscaler

Image to Image

SDXL Controlnet Lora

sdxl-controlnet-lora

Image to Image

Couple Image Generation

eachlabs-couple

Image to Image

Flux Depth Dev

flux-depth-dev

Image to Image