stable diffusion

Txt2Img API Model Upgrade

Souvik Datta

May 17, 2024 • 3 min read

We’ve transitioned from Stable Diffusion v1.5 to the Pix-Art-Sigma model.

PixArt-Σ is a cutting-edge Diffusion Transformer model (DiT) that directly generates images in stunning 4K resolution. This upgrade brings a remarkable leap in quality, offering images with higher fidelity and better alignment with text prompts compared to its predecessor, PixArt-α.

What Can You Expect?

Massive improvements in image quality over the old Text to Image API based on Stable Diffusion v1.5!

Check out the comparisons below to see the dramatic enhancements.

New is Way Better

Prompt	Old SDv1.5	New Pix-Art-Sigma
Surreal portrait of Anna Sawai, detailed, colorful
Ghibli-style house, courtyard, Mediterranean plants
Temptress of the Crystal, aurora borealis, ethereal
Mountain village, sunset, golden hour, acrylic painting

See How Steps Affect the Image Generation Process

Steps	Old SDv1.5	Pix-Art-Sigma
15
20
30
40
50

Try the New Txt2Img API

Try out our new Txt2Img API to see the improvements firsthand. The integration is straightforward:

import requests

url = "https://api.monsterapi.ai/v1/generate/txt2img"

payload = {
      "samples": 1,
      "seed": 42,
      "steps": 30,
      "prompt": "Dog portrait, watercolor style inspired by an existing photograph, emulating an octane render quality, 8K resolution, aiming for photorealism akin to trending ArtStation pieces, bathed in soft, volumetric, cinematic lighting with chiaroscuro effect, reminiscent of award-winning photography, with the feel of a masterpiece oil painting with influences of Raphael, Caravaggio, Greg Rutkowski, Beeple."
  }

headers = {
    "accept": "application/json",
    "content-type": "application/json",
    "authorization": "Bearer YOUR_MONSTER_API_KEY"
}

response = requests.post(url, json=payload, headers=headers)

print(response.text)

Additional Resources -

Txt2Img Documentation