Questions before you download.

Do I need to upload photos to the cloud during culling?

No. Pixel Cafe's core culling pipeline runs locally on your machine. Blur screening, face analysis, burst clustering and adaptive ranking all use your GPU or CPU. Cloud sync is optional and only enabled when you want collaboration or multi-device workflows.

Does it support RAW files?

Yes. Pixel Cafe extracts previews to accelerate RAW handling and supports major camera brands including Sony, Canon, Nikon, Fujifilm and Sigma. HEIC, JPEG and TIFF are also supported.

What stages are included in the AI pipeline?

Stage 1 sharpness and exposure, Stage 2 face analysis, Stage 3 clustering for bursts and similar compositions, Stage 4 adaptive pairwise ranking and Stage 5 optional VLM aesthetic scoring.

Will Pixel Cafe collect my photos for training?

No. Your photos are not uploaded, collected or used for model training. Local-first processing is a core product promise. Sync only transfers content you explicitly choose.

Which operating systems and hardware are supported?

macOS 12+ on Apple Silicon or Intel, and Windows 10+ on x64 with optional CUDA acceleration. Apple M-series machines benefit from Metal inference acceleration.

How many samples are needed before personal ranking starts working?

The default threshold is 1,000 star or reject samples. Before that, Pixel Cafe uses global baseline weights. Once the threshold is reached, personal ranking is enabled automatically and can still be rolled back.