Show HN: Tenderness – open-source synthetic data generation for VLM/OCR

Category: ai-ml

Tags: synthetic-data, document-rendering, ocr, layout-analysis, pango, cairo

Score: 6.3/10 (Innovation: 6, Technical: 7, Documentation: 6, Utility: 6)

Tenderness renders text and images into documents (images, SVGs, PDFs) with precise, known layout, eliminating the need for OCR or heuristics. It targets synthetic data generation for vision-language models and layout understanding benchmarks. The library uses Cairo and Pango for fast, deterministic rendering and provides bounding boxes at multiple granularities.

Target audience: ML researchers, data engineers, computer vision engineers

Repository: https://github.com/paperchase-labs/tenderness · Python · Apache-2.0 · 15 stars

View on Hacker News