Taplio - Build a growth routine on LinkedIn
Mini toolsPATINA
PATINA is a material-estimation model built by fal for extracting full PBR material sets from single images. fal says it is based on a modified FLUX.2 [klein] backbone with an added DINOv2-based adapter for material prediction, and that each map type was trained separately across 512, 768, and 1024 resolution stages. It outputs basecolor, roughness, metalness, displacement, and normal maps, and also powers broader material-generation and extraction endpoints that can create seamless tiling PBR materials up to 8K.
New
Image
Gen 4
Overview
PATINA is fal’s material-map generation model for turning rendered or photographed surfaces into usable PBR texture maps. It predicts basecolor, normal, roughness, metalness, and height maps from an input image, aiming to convert visually appealing images into assets that can be used in traditional CGI and rendering workflows.
About Features and Labels
Features and Labels is an Artificial Intelligence company that specializes in customizing, deploying, and scaling models on Serverless GPUs using the world's first Python Cloud.
View Company ProfileTools using PATINA
No tools found for this model yet.
Last updated: April 16, 2026
