Jagrat Patel, operating as TheJagStudio, has released Tessera-small, a diffusion language model designed for generating 16x16 pixel art. The model treats each pixel as a token and uses a bidirectional transformer for parallel unmasking of pixels, conditioned on a frozen CLIP text encoder. It was trained on the Pixel Art Atlas 1.4M dataset using a single NVIDIA RTX Pro 6000 Blackwell workstation over approximately 5.5 hours. Tessera-small is available under a Creative Commons Attribution-NonCommercial 4.0 license, permitting free use for education, research, and personal non-commercial projects, with a separate license required for commercial applications. AI
IMPACT Enables new creative possibilities for pixel art generation, particularly for non-commercial and educational use.
RANK_REASON Release of a new, specialized generative model with a clear license and usage terms. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Hugging Face Trending Models →
- AdamW
- Angaïs
- Creative Commons Attribution-NonCommercial 4.0 International
- huggingface_hub
- Jagrat Patel
- muon
- Nvidia RTX Pro 6000 Blackwell Workstation Edition
- .openai
- Pixel Art Atlas 1.4M
- tessera
- TesseraPipeline
- Tessera Pixel Studio
- TheJagStudio
- thejagstudio/tessera-small
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →