Pixel Art Synthetic 10K: 9,852 AI-Generated Images with Prompts

Sc077y created a synthetic dataset of 9,852 pixel art images, each paired with a descriptive text prompt. The images are 64x64 pixels in size and use a shared 256-color palette. The dataset was generated using the FLUX.1-dev model with a Retro-Pixel LoRA and was last updated on February 26, 2026.

Use Cases

• Training or fine-tuning text-to-image models based on the described synthetic image-prompt pairs.
• Benchmarking image generation quality for pixel art or low-resolution styles based on the 64x64 grid size.
• Studying the relationship between prompt semantics and visual output in a controlled synthetic environment.
• Developing tools for pixel art asset generation based on the described dataset characteristics.

Strengths

• Contains 9,852 image-prompt pairs, providing a substantial collection for model training.
• Images have a consistent 64x64 pixel resolution and a defined 256-color palette.
• Dataset generation method is specified as using the FLUX.1-dev model with a Retro-Pixel LoRA.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect stylistic bias inherent to the specific generative model and LoRA used.

Provenance

Source

huggingface

Collection Method

Synthetically generated using FLUX.1-dev with a Retro-Pixel LoRA.

Time Range

null

Freshness

Last updated 2026-02-26 23:04:12; freshness should be verified.

Geography

null