manifest: restore original known-good faceswap checksums

manifest: restore original known-good faceswap model checksums

Restore original known-good codeformer.fp16.mnn (revert 06-28 emap reconversion)

Restore original known-good inswapper_128.fp16.mnn (revert 06-28 emap reconversion)

Restore original known-good arcface_w600k_r50.fp16.mnn (revert 06-28 emap reconversion)

Restore original known-good scrfd_10g.fp16.mnn (revert 06-28 emap reconversion)

manifest: point Aura Swap to reconverted, parity-verified models (emap fix)

Aura Swap: reconvert codeformer.fp16.mnn (emap baked into inswapper; parity-verified vs onnxruntime)

Aura Swap: reconvert inswapper_128.fp16.mnn (emap baked into inswapper; parity-verified vs onnxruntime)

Aura Swap: reconvert arcface_w600k_r50.fp16.mnn (emap baked into inswapper; parity-verified vs onnxruntime)

Aura Swap: reconvert scrfd_10g.fp16.mnn (emap baked into inswapper; parity-verified vs onnxruntime)

Remove GHOST commercial-license face-swap set (replaced by working inswapper/codeformer)

Route faceswap to working models (inswapper/codeformer); remove GHOST commercial set

Restore working face-swap models (scrfd_10g, arcface_w600k_r50, inswapper_128, codeformer)

Remove non-commercial face-swap models (replaced by commercial-licensed set)

Upload manifest.json with huggingface_hub

Upload restoreformer.fp32.mnn with huggingface_hub

Upload ghost_3_256.fp32.mnn with huggingface_hub

Upload arcface_converter_ghost.fp32.mnn with huggingface_hub

Upload arcface_glintr100_aura.fp32.mnn with huggingface_hub

Upload scrfd_10g_aura.fp32.mnn with huggingface_hub

Upload manifest.json with huggingface_hub

Upload codeformer.fp16.mnn with huggingface_hub

Upload inswapper_128.fp16.mnn with huggingface_hub

Upload arcface_w600k_r50.fp16.mnn with huggingface_hub

Upload scrfd_10g.fp16.mnn with huggingface_hub

Reword LiteRT model descriptions: MediaPipe → LiteRT-LM

Fix LiteRT Gemma 4 E2B/E4B: use Android .litertlm instead of maintenance-mode -web.task

manifest: add Gemma 4 E2B/E4B (LiteRT/MediaPipe .task) as second-engine chat models

manifest: remove legacy Root agent model; mark MiniLM embedding as recommended (now drives auto-memory + faster replies)

docs: add Fancy AI Play Store link

Add BiSeNet face-parser (fp16 MNN) for ADetailer face enhance

Update manifest.json

Update manifest.json

Upload tts-en-amy-low.zip

Upload 2 files

Update PRIVACY.md

Update manifest.json

Update manifest.json

Update manifest.json

Update manifest.json

Update manifest.json

Update PRIVACY.md

Update PRIVACY.md

Add privacy policy

Update manifest.json

Update README.md

Update README.md

Update README.md

Upload manifest.json