Post
42
We’re releasing Lunara Art Eval, an art evaluation dataset for studying aesthetic quality, emotional resonance and content integrity in image generation.
It contains 8,000 generated images across 1,000 shared prompts, covering diverse artistic styles. The eight models are Qwen-Image, AuraFlow, GPT-Image-1 Mini, Stable Diffusion 3.5 Large Turbo, HiDream-I1 Fast, FLUX.2-Klein-4B, Z-Image-Turbo and Moonworks Lunara.
Each image includes scores on the three dimensions, allowing comparisons across models using the same prompts and evaluator.
Dataset: moonworks/lunara-art-eval
Paper: https://arxiv.org/abs/2609.22272
The paper also introduces Lunara’s Diffusion Mixture Transformer architecture, with fewer than 10B active parameters, and CAT, an active-learning-inspired training algorithm that iteratively updates the training distribution.
We’re a new lab focused on model architectures and training algorithms, working closely with artists and creators. Our earlier public datasets are:
- [Lunara Aesthetic I]( moonworks/lunara-aesthetic): art images with human-refined prompts and region, style and topic labels.
- [Lunara Aesthetic II]( moonworks/lunara-aesthetic-image-variations): original images paired with contextual variations.
We’d appreciate feedback from people working on image generation and evaluation, particularly on where automated scores with your judgment.
It contains 8,000 generated images across 1,000 shared prompts, covering diverse artistic styles. The eight models are Qwen-Image, AuraFlow, GPT-Image-1 Mini, Stable Diffusion 3.5 Large Turbo, HiDream-I1 Fast, FLUX.2-Klein-4B, Z-Image-Turbo and Moonworks Lunara.
Each image includes scores on the three dimensions, allowing comparisons across models using the same prompts and evaluator.
Dataset: moonworks/lunara-art-eval
Paper: https://arxiv.org/abs/2609.22272
The paper also introduces Lunara’s Diffusion Mixture Transformer architecture, with fewer than 10B active parameters, and CAT, an active-learning-inspired training algorithm that iteratively updates the training distribution.
We’re a new lab focused on model architectures and training algorithms, working closely with artists and creators. Our earlier public datasets are:
- [Lunara Aesthetic I]( moonworks/lunara-aesthetic): art images with human-refined prompts and region, style and topic labels.
- [Lunara Aesthetic II]( moonworks/lunara-aesthetic-image-variations): original images paired with contextual variations.
We’d appreciate feedback from people working on image generation and evaluation, particularly on where automated scores with your judgment.
