https://prompt-aligned.github.io/\n","updatedAt":"2024-01-13T05:13:20.491Z","author":{"_id":"64c2ec0d5b922ce89b45af46","avatarUrl":"/avatars/77c477c0ebc2820c8806d0f0f6ed7b09.svg","fullname":"Andrew Carr","name":"andrew-cartwheel","type":"user","isPro":true,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":3}},"numEdits":0,"identifiedLanguage":{"language":"en","probability":0.22645296156406403},"editors":["andrew-cartwheel"],"editorAvatarUrls":["/avatars/77c477c0ebc2820c8806d0f0f6ed7b09.svg"],"reactions":[],"isReport":false}},{"id":"65a54abb3e3a00a9c15ad569","createdAt":"2024-01-15T15:09:47.000Z","type":"comment","data":{"edited":true,"hidden":true,"hiddenBy":"","latest":{"raw":"This comment has been hidden","html":"This comment has been hidden","updatedAt":"2024-01-15T16:17:44.490Z"},"numEdits":0,"editors":[],"editorAvatarUrls":[],"reactions":[]}}],"primaryEmailConfirmed":false,"paper":{"id":"2401.06105","authors":[{"_id":"65a0bd799185dcca3061a974","user":{"_id":"60defb17bb3e1aaa18b005bb","avatarUrl":"/avatars/086264e93d6f7ba2eab5cc1b218df4a5.svg","isPro":false,"fullname":"Moab Arar","user":"moabarar","type":"user"},"name":"Moab Arar","status":"admin_assigned","statusLastChangedAt":"2024-01-12T09:40:45.421Z","hidden":false},{"_id":"65a0bd799185dcca3061a975","user":{"_id":"630551fd80bc5e03dad21ea6","avatarUrl":"/avatars/849e41404df698cc89c68939de45ec9a.svg","isPro":false,"fullname":"Andrey Voynov","user":"avoin","type":"user"},"name":"Andrey Voynov","status":"admin_assigned","statusLastChangedAt":"2024-01-12T09:40:51.688Z","hidden":false},{"_id":"65a0bd799185dcca3061a976","user":{"_id":"63071a86fd156a1d624a4c46","avatarUrl":"/avatars/bfc2a2169c080304124eb7f9ab04306d.svg","isPro":false,"fullname":"Amir Hertz","user":"amirhertz","type":"user"},"name":"Amir Hertz","status":"admin_assigned","statusLastChangedAt":"2024-01-12T09:40:57.655Z","hidden":false},{"_id":"65a0bd799185dcca3061a977","user":{"_id":"62f6ab4fc3372328414c8689","avatarUrl":"/avatars/4f728a5b70c9fe4a64e80e2b643ca620.svg","isPro":false,"fullname":"Omri Avrahami","user":"omriav","type":"user"},"name":"Omri Avrahami","status":"admin_assigned","statusLastChangedAt":"2024-01-12T09:41:04.417Z","hidden":false},{"_id":"65a0bd799185dcca3061a978","user":{"_id":"6335ae887fcbeb615b0f9924","avatarUrl":"/avatars/60fd9305ccc56f17d6007f781c54fe2b.svg","isPro":false,"fullname":"Shlomi Fruchter","user":"shlomif","type":"user"},"name":"Shlomi Fruchter","status":"admin_assigned","statusLastChangedAt":"2024-01-12T10:01:53.899Z","hidden":false},{"_id":"65a0bd799185dcca3061a979","name":"Yael Pritch","hidden":false},{"_id":"65a0bd799185dcca3061a97a","user":{"_id":"628507161949ebcae8e24ec3","avatarUrl":"/avatars/008ecb3daa4c8187b5f339f1176b3c39.svg","isPro":false,"fullname":"Daniel Cohen-Or","user":"cohenor","type":"user"},"name":"Daniel Cohen-Or","status":"admin_assigned","statusLastChangedAt":"2024-01-12T10:02:04.337Z","hidden":false},{"_id":"65a0bd799185dcca3061a97b","name":"Ariel Shamir","hidden":false}],"publishedAt":"2024-01-11T18:35:33.000Z","submittedOnDailyAt":"2024-01-12T01:48:04.903Z","title":"PALP: Prompt Aligned Personalization of Text-to-Image Models","submittedOnDailyBy":{"_id":"60f1abe7544c2adfd699860c","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674929746905-60f1abe7544c2adfd699860c.jpeg","isPro":false,"fullname":"AK","user":"akhaliq","type":"user"},"summary":"Content creators often aim to create personalized images using personal\nsubjects that go beyond the capabilities of conventional text-to-image models.\nAdditionally, they may want the resulting image to encompass a specific\nlocation, style, ambiance, and more. Existing personalization methods may\ncompromise personalization ability or the alignment to complex textual prompts.\nThis trade-off can impede the fulfillment of user prompts and subject fidelity.\nWe propose a new approach focusing on personalization methods for a\nsingle prompt to address this issue. We term our approach prompt-aligned\npersonalization. While this may seem restrictive, our method excels in\nimproving text alignment, enabling the creation of images with complex and\nintricate prompts, which may pose a challenge for current techniques. In\nparticular, our method keeps the personalized model aligned with a target\nprompt using an additional score distillation sampling term. We demonstrate the\nversatility of our method in multi- and single-shot settings and further show\nthat it can compose multiple subjects or use inspiration from reference images,\nsuch as artworks. We compare our approach quantitatively and qualitatively with\nexisting baselines and state-of-the-art techniques.","upvotes":50,"discussionId":"65a0bd7c9185dcca3061aa79","ai_summary":"Prompt-aligned personalization improves text alignment and enhances personalized image creation with complex prompts by using score distillation sampling.","ai_keywords":["prompt-aligned personalization","score distillation sampling"]},"canReadDatabase":false,"canManagePapers":false,"canSubmit":false,"hasHfLevelAccess":false,"upvoted":false,"upvoters":[{"_id":"620783f24e28382272337ba4","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/620783f24e28382272337ba4/zkUveQPNiDfYjgGhuFErj.jpeg","isPro":false,"fullname":"GuoLiangTang","user":"Tommy930","type":"user"},{"_id":"637cb133bf9c9748e3be984e","avatarUrl":"/avatars/ce4859cd0de226b02e8f53fb0cc8e494.svg","isPro":false,"fullname":"Tausif Iqbal","user":"TieIncred","type":"user"},{"_id":"6538119803519fddb4a17e10","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6538119803519fddb4a17e10/ffJMkdx-rM7VvLTCM6ri_.jpeg","isPro":false,"fullname":"samusenps","user":"samusenps","type":"user"},{"_id":"635cada2c017767a629db012","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1667018139063-noauth.jpeg","isPro":false,"fullname":"Ojasvi Singh Yadav","user":"ojasvisingh786","type":"user"},{"_id":"638f308fc4444c6ca870b60a","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/638f308fc4444c6ca870b60a/Q11NK-8-JbiilJ-vk2LAR.png","isPro":true,"fullname":"Linoy Tsaban","user":"linoyts","type":"user"},{"_id":"60c8d264224e250fb0178f77","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/60c8d264224e250fb0178f77/i8fbkBVcoFeJRmkQ9kYAE.png","isPro":true,"fullname":"Adam Lee","user":"Abecid","type":"user"},{"_id":"6423e376774cc34079728933","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/6423e376774cc34079728933/uMPBasPAGe4bgmNqabr8m.png","isPro":false,"fullname":"H3c","user":"h3clikejava","type":"user"},{"_id":"648eb1eb59c4e5c87dc116e0","avatarUrl":"/avatars/c636cea39c2c0937f01398c94ead5dad.svg","isPro":false,"fullname":"fdsqefsgergd","user":"T-representer","type":"user"},{"_id":"63470b9f3ea42ee2cb4f3279","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/Xv8-IxM4GYM91IUOkRnCG.png","isPro":false,"fullname":"NG","user":"SirRa1zel","type":"user"},{"_id":"62a9afdd0472c0b7f94c491c","avatarUrl":"/avatars/7ca2d750fb67cc848dc07a9161bfa9dd.svg","isPro":false,"fullname":"Martin Viewegger","user":"Viewegger","type":"user"},{"_id":"64747f7e33192631bacd8831","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/64747f7e33192631bacd8831/dstkZJ4sHJSeqLesV5cOC.jpeg","isPro":false,"fullname":"Taufiq Dwi Purnomo","user":"taufiqdp","type":"user"},{"_id":"6572aa5849719ff0da8a9c8c","avatarUrl":"/avatars/516d1d3aaa0439ab738977787ce9c7d4.svg","isPro":false,"fullname":"CLEMENT L","user":"LVXXX","type":"user"}],"acceptLanguages":["*"],"dailyPaperRank":2}">
Prompt-aligned personalization improves text alignment and enhances personalized image creation with complex prompts by using score distillation sampling.
AI-generated summary
Content creators often aim to create personalized images using personal
subjects that go beyond the capabilities of conventional text-to-image models.
Additionally, they may want the resulting image to encompass a specific
location, style, ambiance, and more. Existing personalization methods may
compromise personalization ability or the alignment to complex textual prompts.
This trade-off can impede the fulfillment of user prompts and subject fidelity.
We propose a new approach focusing on personalization methods for a
single prompt to address this issue. We term our approach prompt-aligned
personalization. While this may seem restrictive, our method excels in
improving text alignment, enabling the creation of images with complex and
intricate prompts, which may pose a challenge for current techniques. In
particular, our method keeps the personalized model aligned with a target
prompt using an additional score distillation sampling term. We demonstrate the
versatility of our method in multi- and single-shot settings and further show
that it can compose multiple subjects or use inspiration from reference images,
such as artworks. We compare our approach quantitatively and qualitatively with
existing baselines and state-of-the-art techniques.