p-image-ideogram

Generate high-quality images from text with strong photorealism, typography, and prompt adherence.

P-Image-Ideogram is Pruna’s optimized text-to-image model built with Ideogram.

How it works

The model takes:

• A prompt describing the image you want to generate.
• A thinking level that controls the balance between generation cost and image quality.
• An aspect ratio, or custom width and height.

It returns a generated image in your chosen output format.

Tips

• Increase thinking when image quality, prompt adherence, or typography matters most. Lower levels are useful for faster, lower-cost iterations.
• Turn prompt_upsampling off when you want the model to follow your prompt exactly as written.
• Use structured JSON prompts for precise compositions. You can specify elements such as hex colors and bounding boxes to control layout and placement.
• Use quotation marks around any text that must appear in the image, and describe its font, size, color, and position.
• Choose custom as the aspect_ratio when setting width and height directly. Custom dimensions are rounded to multiples of 16 and support values up to 2560 pixels.

Example

import replicate

output = replicate.run(
    "prunaai/p-image-ideogram",
    input={
        "prompt": (
            "Editorial beauty campaign for a violet perfume bottle on a "
            "reflective black surface, dramatic studio lighting. "
            "At the top, elegant white serif text reading: "
            "\"MIDNIGHT BLOOM\""
        ),
        "thinking": "high",
        "aspect_ratio": "1:1",
        "prompt_upsampling": True,
        "output_format": "jpg",
        "output_quality": 90,
        "seed": 42,
    },
)

print(output)

Resources

Docs: https://docs.pruna.ai/en/stable/docs_pruna_endpoints/index.html