Show HN: DeepFaceLab – Free AI Face Swap Online (deepfacelab.app)

DeepFaceLab has launched a browser-based, free (with optional premium tiers) AI face‑swap service that runs entirely in the cloud, removing the need for local high‑end hardware or downloads. The platform automates a three‑stage pipeline—frame extraction and face landmark detection, model training on thousands of facial examples, and per‑frame model merging—with real‑time previews, automatic alignment, quality filtering, histogram color matching, artifact removal and temporal smoothing to ensure consistent, flicker‑free results. Advanced features include multi‑face replacement, head/hair/shoulder swaps, lip‑sync and expression transfer, plus speed‑vs‑quality models, GPU‑accelerated TensorFlow training, multi‑threaded tasks and distributed cloud processing that can produce short clips in tens of seconds under favorable conditions. For the AI/ML community and creators this is significant because it packages deep‑learning face synthesis tools (landmark mapping, hierarchical feature extraction, continuous optimization) into an accessible, scalable web service, lowering the barrier for indie VFX, localization, education and research. Technically it highlights practical production‑grade additions—automatic QA, face similarity sorting for efficient training, color/lighting correction, and frame interpolation—that improve fidelity and throughput. The release also has broader implications: easier dataset generation and experimentation for computer vision researchers, while raising familiar ethical and misuse considerations around realistic identity transfer that the community will need to address.