Nine state-of-the-art models. All run on your device through ONNX Runtime. Indistinguishable from cloud AI in capability — except we never see your photos.
Each model is a pre-quantized ONNX file (50–200 MB). PhotoSphere downloads them on-demand the first time you use a feature. Everything from there is local.
Models fetch from our CDN. SHA-256 verified. Resumable.
Microsoft's runtime, optimized for CPU + GPU. Same engine as Office.
8 GB RAM works. GPU optional. M-series Macs and modern Intel/AMD fly.
No telemetry, no API call, no "anonymized features." Verify with Wireshark.
CLIP (Contrastive Language–Image Pretraining) maps photos and text into the same 512-dimensional space. Photos with similar meaning have similar embeddings. We compute one embedding per photo, store it in an HNSW index, and at query time embed your text query and find the nearest neighbors.
Two-stage pipeline. RetinaFace detects every face in a photo with sub-pixel landmark precision. We crop each face to 112×112 and feed it into ArcFace, which produces a 512-dimensional embedding. Cosine similarity between embeddings clusters the same person across thousands of photos.
YOLOv8-nano runs on every photo at indexing time. 80 COCO classes covers most of what you photograph: people, vehicles, animals, food, sports gear, kitchen items, electronics, furniture. Auto-tags become regular searchable tags — you can edit, merge, or delete them.
LaMa (Large Mask Inpainting) takes a 4-channel input — your photo plus a binary mask of what to remove — and synthesizes plausible content for the masked area using Fourier convolutions. Unlike older inpainting algorithms, it handles large masks beautifully.
U²Net is a salient-object segmentation network. Feed it a photo, get a soft alpha matte that separates the subject from the background. Use it to delete the background, swap skies (combined with masking), or extract subjects for product shots.
Two models in sequence. DBNet locates text regions with bounding polygons. CRNN with CTC decoding reads each region into Unicode text. Output is indexed alongside filename/camera in SQLite FTS5 so receipts, signs, screenshots, and notes become full-text searchable.
Neural Image Assessment scores every photo from 1–10 on aesthetic quality. Combined with a heuristic Laplacian variance for sharpness. Powers "auto-pick the best from a burst" and a "gems" Memory.
Recover detail from motion blur and slight defocus. Generative deblurring trained on synthetic blur kernels. Run on the photos you almost deleted.
One ONNX model per artistic style (Van Gogh, Hokusai, Monet, etc.). Drop a style on a photo. Each style is small (~5 MB) and downloadable separately.
Beach, mountain, urban, forest, indoor, sport, food, document, screenshot. A lightweight classifier auto-labels every photo. Searchable as a tag, mappable to album rules.
| Minimum CPU | x86-64 with AVX2 (any 2015+ Intel/AMD), Apple M1+, ARM64 with NEON |
| Recommended CPU | Apple M-series, Intel Core i5 12th gen+, AMD Ryzen 5000+ |
| Minimum RAM | 4 GB (basic library), 8 GB (with AI features) |
| Recommended RAM | 16 GB for libraries over 25,000 photos |
| GPU acceleration | Optional. CUDA, DirectML, CoreML, ROCm — auto-detected if available |
| Disk for app | 100 MB |
| Disk for AI models | ~1.2 GB if you install all nine. Selective download. |
| Disk for library | Bring your own. Index lives at ~50 MB per 10,000 photos. |
| Network | One-time model download (CDN). After that: zero. |
Apple, Google, and Adobe also offer most of these features. They also see every photo. PhotoSphere doesn't.