Sana by Nvidia

sana

Sana, a text-to-image framework that can efficiently generate images up to 4096 × 4096 resolution

H100 80GB

Fast Inference

REST API

Try in Console API Docs Examples

Model Information

Response Time~1 sec

StatusActive

Version

0.0.1

Updated6 days ago

Live Demo

Average runtime: ~1 seconds

Input

Configure model parameters

Negative Prompt

Specify things to not see in the output

Enter your negative prompt

Prompt

Input prompt

a cyberpunk cat with a neon sign that says "Sana"

Width

Width of output image

1024

Height

Height of output image

1024

Num Inference Steps

Number of denoising steps

Guidance Scale

Classifier-free guidance scale

Pag Guidance Scale

PAG Guidance scale

Seed

Random seed. Leave blank to randomize the seed

Enter your seed

Output

View generated results

Result

Preview, share or download your results with a single click.

Cost is calculated based on execution time.The model is charged at $0.0018 per second. With a $1 budget, you can run this model approximately 555 times, assuming an average execution time of 1 seconds per run.

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": "sana",
            "version": "0.0.1",
            "input": {
  "seed": null,
  "width": "1024",
  "height": "1024",
  "prompt": "a cyberpunk cat with a neon sign that says \"Sana\"",
  "guidance_scale": "5",
  "negative_prompt": "your negative prompt here",
  "pag_guidance_scale": "2",
  "num_inference_steps": "18"
}
        }
    )
    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: ~1 seconds
Rate limit: 60 requests/minute
Concurrent requests: 10 maximum
Use long-polling to check prediction status until completion

Overview

Sana by Nvidia is designed for generating high-quality images based on detailed textual prompts. With a focus on flexibility and precision, it supports advanced customization through adjustable parameters. Whether you're creating artistic visuals, concept art, or professional imagery, this model provides the model to bring your ideas to life.

Technical Specifications

Text-to-Image Capability: Generates realistic and artistic images from detailed textual descriptions.
Negative Prompting: Allows precise control over unwanted elements in the output.
Configurable Parameters: Provides extensive options to fine-tune outputs based on user preferences.

Key Considerations

Resolution and Performance: Higher resolutions (width and height) increase processing time; balance quality with performance needs.
Prompt Length: Overly long prompts may dilute the model’s focus. Stick to succinct, targeted descriptions.
Guidance Scale Balance: Excessive values for guidance_scale or pag_guidance_scale might lead to unnatural or overemphasized elements.
Seed for Reproducibility: Use the same seed value to regenerate identical results.

Tips & Tricks

Refine Your Prompt: Test variations of your description to discover the best phrasing for your desired output.
Negative Prompt Efficiency: Use negative_prompt to filter out undesired elements and focus on key details.
Guidance Scale: Start with moderate values between 8–12 for balanced outputs. For more creative or artistic results, experiment with higher values like 15–18. Use lower values (e.g., 5–7) for a subtler influence on the output. Avoid extreme values unless specific effects are desired, as they may lead to unnatural results.
Inference Steps: For quick previews, use values between 10–20 to get a sense of the output without long processing times. For detailed and high-quality outputs, use 30–50 steps. Avoid going beyond 60, as the improvements often diminish while processing time increases significantly.
Seed Control: Reuse specific seed values to reproduce consistent results for iterative projects.
Pag Guidance Scale: Use values between 10–14 to subtly enhance the structure or style of the output. For stronger stylistic influence, increase to 15–18, and for a lighter touch, experiment with 6–9. Avoid values below 5, as they may not have a noticeable impact on the results.Combining guidance_scale at 10–12 and pag_guidance_scale at 12–15 often provides a harmonious balance between adherence to the prompt and artistic styling.

Capabilities

Creates stunning, high-resolution images from textual descriptions.
Supports detailed customization through multiple adjustable parameters.
Enables repeatable results using the seed parameter.

What can I use for?

Artistic Creations: Generate concept art, illustrations, or unique designs.
Professional Projects: Design marketing visuals, product mockups, or presentation materials.
Creative Exploration: Experiment with prompts to explore new artistic styles and ideas.

Things to be aware of

Detailed Scenes: Describe intricate settings (e.g., "a bustling city at night with neon signs and rain-soaked streets").
Negative Refinements: Use negative_prompt to avoid unwanted elements (e.g., "no haze, no people").
High-Quality Outputs: Increase num_inference_steps for sharper, more polished images.
Consistent Themes: Reuse seed values to maintain a consistent style across multiple outputs.
Creative Styles: Experiment with guidance_scale to explore different levels of prompt adherence and artistic influence.

Limitations

Abstract Concepts: May struggle to interpret highly abstract or ambiguous prompts.
Processing Time: High-resolution images or extensive steps can lead to longer generation times.
Prompt Sensitivity: Minor changes in wording can significantly impact results.

Output Format: PNG

Related AI Models

Fooocus

fooocus-api

Text to Image

Omni Zero Couple

omni-zero-couples

Text to Image

Flux 1.1 Pro

flux-1-1-pro

Text to Image

Imagen 3

imagen-3

Text to Image