vantilator2000/pbr-texture-gen

Generate seamless PBR texture maps from text prompts with Segmind SSD-1B

Public
18 runs
Playground API Examples README Versions

Examples

View more

Run time and cost

This model runs on Nvidia T4 GPU hardware. We don't yet have enough runs of this model to provide performance information.

Readme

PBR Texture Generator

Generate seamless PBR (Physically Based Rendering) texture maps from text descriptions.

Output

This model generates 4 texture maps from a single prompt:

Map Description
Diffuse Base color/albedo texture
Normal Surface detail for lighting
Roughness Surface smoothness (black=smooth, white=rough)
AO Ambient occlusion for soft shadows

Usage

import replicate

output = replicate.run(
    "vantilator2000/pbr-texture-gen",
    input={
        "prompt": "seamless red brick wall texture, weathered, detailed",
        "resolution": 1024,
        "tiling_strength": 0.5,
        "num_steps": 8,
        "seed": 42
    }
)

# output.diffuse, output.normal, output.roughness, output.ao

Parameters

Parameter Default Description
prompt - Text description of the texture
resolution 512 Output size (512 or 1024)
tiling_strength 0.5 Seamless tiling blend (0-1)
num_steps 8 Inference steps (1-30, higher=better quality)
seed -1 Random seed (-1 for random)

Example Prompts

  • seamless dark wood texture, oak, detailed grain
  • seamless marble texture, white with grey veins
  • seamless grass texture, top-down view, lawn
  • seamless concrete texture, weathered, cracks
  • seamless metal texture, brushed steel
  • seamless fabric texture, blue denim, detailed

Tips

  1. Include “seamless” in your prompt for better tiling
  2. Use “top-down view” for floor/ground textures
  3. Increase num_steps (12-20) for higher quality
  4. Set tiling_strength to 0.7+ for perfect seamless edges

Use Cases

  • Game development (Unity, Unreal Engine)
  • 3D rendering (Blender, Maya, 3ds Max)
  • Architectural visualization
  • Product design
  • Digital art

Model

Powered by Segmind SSD-1B - a distilled version of SDXL optimized for speed and quality.

Model created
Model updated