{"paper":{"title":"InstantID: Zero-shot Identity-Preserving Generation in Seconds","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"InstantID generates high-fidelity personalized images from one face photo in seconds without fine-tuning.","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Anthony Chen, Haofan Wang, Huaxia Li, Qixun Wang, Xu Bai, Xu Tang, Yao Hu, Zekui Qin","submitted_at":"2024-01-15T07:50:18Z","abstract_excerpt":"There has been significant progress in personalized image synthesis with methods such as Textual Inversion, DreamBooth, and LoRA. Yet, their real-world applicability is hindered by high storage demands, lengthy fine-tuning processes, and the need for multiple reference images. Conversely, existing ID embedding-based methods, while requiring only a single forward inference, face challenges: they either necessitate extensive fine-tuning across numerous model parameters, lack compatibility with community pre-trained models, or fail to maintain high face fidelity. Addressing these limitations, we "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our plug-and-play module adeptly handles image personalization in various styles using just a single facial image, while ensuring high fidelity.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the IdentityNet design, by imposing strong semantic and weak spatial conditions on facial and landmark images integrated with textual prompts, will deliver high face fidelity across styles without fine-tuning or multiple references.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"InstantID enables zero-shot identity-preserving image generation from one facial image via a novel IdentityNet that combines strong semantic and weak spatial conditioning with text prompts in diffusion models.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"InstantID generates high-fidelity personalized images from one face photo in seconds without fine-tuning.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"c7c55b2b29e48134074dbb9842a109140b90530e246b06d2351cf5ae3084f219"},"source":{"id":"2401.07519","kind":"arxiv","version":2},"verdict":{"id":"ece50b4f-89f0-4340-beb7-5021bdf36772","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T20:57:42.525913Z","strongest_claim":"Our plug-and-play module adeptly handles image personalization in various styles using just a single facial image, while ensuring high fidelity.","one_line_summary":"InstantID enables zero-shot identity-preserving image generation from one facial image via a novel IdentityNet that combines strong semantic and weak spatial conditioning with text prompts in diffusion models.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the IdentityNet design, by imposing strong semantic and weak spatial conditions on facial and landmark images integrated with textual prompts, will deliver high face fidelity across styles without fine-tuning or multiple references.","pith_extraction_headline":"InstantID generates high-fidelity personalized images from one face photo in seconds without fine-tuning."},"references":{"count":28,"sample":[{"doi":"","year":2022,"title":"eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers","work_id":"2cd7b629-ab37-4ce5-b51e-aa4d99547468","ref_index":1,"cited_arxiv_id":"2211.01324","is_internal_anchor":true},{"doi":"","year":2023,"title":"arXiv preprint arXiv:2307.09481 (2023)","work_id":"b5906af7-b3b8-4517-a1d2-9687f00960ea","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2021,"title":"In: 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021","work_id":"8e700a39-a905-4502-b336-4e1a69b5bf6a","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.48550/arxiv.2208.01618","year":2022,"title":"An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion","work_id":"ca618c21-3ba6-448e-bd86-bcecff3cdeb5","ref_index":4,"cited_arxiv_id":"2208.01618","is_internal_anchor":true},{"doi":"","year":2023,"title":"Designing an encoder for fast personalization of text-to-image models","work_id":"38db57b1-fb65-4b24-9152-ee4147c37fb4","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":28,"snapshot_sha256":"d825eb2dc763f997431cb7886d5e136835bba665b743088a5786fdf3f3995597","internal_anchors":7},"formal_canon":{"evidence_count":1,"snapshot_sha256":"b3b5502522b8ff5bf8e0b933470cd7db20d5632d257acc3bd0214da84b616474"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}