{"id":"https://openalex.org/W7204959438","doi":"https://doi.org/10.48550/arxiv.2608.29904","title":"Off-Manifold Refinement: Guiding Video Generators with a Frozen World Model","display_name":"Off-Manifold Refinement: Guiding Video Generators with a Frozen World Model","publication_year":2026,"publication_date":"2026-08-30","ids":{"openalex":"https://openalex.org/W7204959438","doi":"https://doi.org/10.48550/arxiv.2608.29904"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.29904","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29904","pdf_url":"https://arxiv.org/pdf/2608.29904","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2608.29904","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5045047993","display_name":"Hai Nguyen-Truong","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4210127305","display_name":"Fulbright University Vietnam","ror":"https://ror.org/02ygfdv70","country_code":"VN","type":"education","lineage":["https://openalex.org/I4210127305"]}],"countries":["VN"],"is_corresponding":false,"raw_author_name":"Nguyen-Truong, Hai","raw_affiliation_strings":["Fulbright University Vietnam"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fulbright University Vietnam","institution_ids":["https://openalex.org/I4210127305"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5075040240","display_name":"Tuan-Anh Vu","orcid":"https://orcid.org/0000-0002-8872-0875","observed_orcids":["https://orcid.org/0000-0002-8872-0875","https://orcid.org/0009-0003-7898-7204"]},"institutions":[{"id":"https://openalex.org/I161318765","display_name":"University of California, Los Angeles","ror":"https://ror.org/046rm7j60","country_code":"US","type":"education","lineage":["https://openalex.org/I161318765"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Vu, Tuan-Anh","raw_affiliation_strings":["University of California , Los Angeles"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California , Los Angeles","institution_ids":["https://openalex.org/I161318765"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5135923618","display_name":"Dang Huynh","orcid":"https://orcid.org/0000-0002-2314-4934","observed_orcids":["https://orcid.org/0000-0002-2314-4934"]},"institutions":[{"id":"https://openalex.org/I4210127305","display_name":"Fulbright University Vietnam","ror":"https://ror.org/02ygfdv70","country_code":"VN","type":"education","lineage":["https://openalex.org/I4210127305"]}],"countries":["VN"],"is_corresponding":false,"raw_author_name":"Huynh, Dang","raw_affiliation_strings":["Fulbright University Vietnam"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Fulbright University Vietnam","institution_ids":["https://openalex.org/I4210127305"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.9994999766349792,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.9994999766349792,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.00019999999494757503,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11206","display_name":"Model Reduction and Neural Networks","score":0.0,"subfield":{"id":"https://openalex.org/subfields/3109","display_name":"Statistical and Nonlinear Physics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/video-generation","display_name":"video generation","score":0.8370000123977661},{"id":"https://openalex.org/keywords/physical-consistency","display_name":"physical consistency","score":0.7369999885559082},{"id":"https://openalex.org/keywords/world-models","display_name":"world models","score":0.6259999871253967},{"id":"https://openalex.org/keywords/physical-dynamics","display_name":"physical dynamics","score":0.3310000002384186},{"id":"https://openalex.org/keywords/diffusive-sampling","display_name":"diffusion sampling","score":0.32199999690055847}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7253999710083008},{"id":"https://openalex.org/C205711294","wikidata":"https://www.wikidata.org/wiki/Q176953","display_name":"Rendering (computer graphics)","level":2,"score":0.5267000198364258},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.4246000051498413},{"id":"https://openalex.org/C2780992000","wikidata":"https://www.wikidata.org/wiki/Q17016113","display_name":"Generator (circuit theory)","level":3,"score":0.41920000314712524},{"id":"https://openalex.org/C2780513914","wikidata":"https://www.wikidata.org/wiki/Q18210350","display_name":"Bottleneck","level":2,"score":0.3993000090122223},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.39629998803138733},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3718000054359436},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3691999912261963},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.3684000074863434},{"id":"https://openalex.org/C61445026","wikidata":"https://www.wikidata.org/wiki/Q217608","display_name":"Fixed point","level":2,"score":0.36559998989105225},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.3443000018596649},{"id":"https://openalex.org/C132094186","wikidata":"https://www.wikidata.org/wiki/Q641585","display_name":"Clutter","level":3,"score":0.3409000039100647},{"id":"https://openalex.org/C2780704645","wikidata":"https://www.wikidata.org/wiki/Q9251458","display_name":"Observer (physics)","level":2,"score":0.3294000029563904},{"id":"https://openalex.org/C3020493868","wikidata":"https://www.wikidata.org/wiki/Q55631277","display_name":"Real world data","level":2,"score":0.29820001125335693},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.2872999906539917},{"id":"https://openalex.org/C3770464","wikidata":"https://www.wikidata.org/wiki/Q775963","display_name":"Smoothing","level":2,"score":0.27900001406669617},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.27720001339912415},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.27489998936653137},{"id":"https://openalex.org/C34862557","wikidata":"https://www.wikidata.org/wiki/Q178985","display_name":"Ode","level":2,"score":0.2711000144481659},{"id":"https://openalex.org/C152124472","wikidata":"https://www.wikidata.org/wiki/Q1204361","display_name":"Redundancy (engineering)","level":2,"score":0.2662999927997589},{"id":"https://openalex.org/C48007421","wikidata":"https://www.wikidata.org/wiki/Q676252","display_name":"Motion capture","level":3,"score":0.260699987411499},{"id":"https://openalex.org/C21080849","wikidata":"https://www.wikidata.org/wiki/Q13611879","display_name":"Data point","level":2,"score":0.26030001044273376},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.26030001044273376},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.25999999046325684},{"id":"https://openalex.org/C3261483","wikidata":"https://www.wikidata.org/wiki/Q119565","display_name":"Frame rate","level":2,"score":0.25920000672340393},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.25369998812675476},{"id":"https://openalex.org/C113954288","wikidata":"https://www.wikidata.org/wiki/Q186885","display_name":"Timestamp","level":2,"score":0.2531999945640564},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.2515000104904175},{"id":"https://openalex.org/C163294075","wikidata":"https://www.wikidata.org/wiki/Q581861","display_name":"Noise reduction","level":2,"score":0.2513999938964844}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.29904","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29904","pdf_url":"https://arxiv.org/pdf/2608.29904","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2608.29904","doi":"https://doi.org/10.48550/arxiv.2608.29904","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.29904","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.29904","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.29904","pdf_url":"https://arxiv.org/pdf/2608.29904","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7204959438.pdf","grobid_xml":"https://content.openalex.org/works/W7204959438.grobid-xml"},"referenced_works_count":8,"referenced_works":["https://openalex.org/W2795843265","https://openalex.org/W4400375159","https://openalex.org/W4402387301","https://openalex.org/W4417328281","https://openalex.org/W6966581363","https://openalex.org/W7111062764","https://openalex.org/W7125835849","https://openalex.org/W7143419904"],"related_works":["https://openalex.org/W7162686099","https://openalex.org/W7124449697","https://openalex.org/W7143332438","https://openalex.org/W7212139762","https://openalex.org/W7124303476","https://openalex.org/W7125835849","https://openalex.org/W7133360590","https://openalex.org/W7161726325","https://openalex.org/W7161847472","https://openalex.org/W4415337149"],"abstract_inverted_index":{"Modern":[0],"video":[1,166],"generators":[2],"routinely":[3],"fail":[4],"at":[5,161],"physical":[6,30,36],"dynamics:":[7],"objects":[8],"float,":[9],"trajectories":[10],"violate":[11],"gravity,":[12],"contacts":[13],"vanish.":[14],"Standard":[15],"denoising":[16],"and":[17,49,59,67,75,130,163,168],"flow-matching":[18],"objectives":[19],"fit":[20],"visual":[21],"data":[22,74],"distributions":[23],"but":[24,38],"do":[25],"not":[26],"explicitly":[27],"penalize":[28],"such":[29],"violations.":[31],"Existing":[32],"remedies":[33],"can":[34,120],"improve":[35],"consistency,":[37],"typically":[39],"add":[40],"substantial":[41],"inference":[42],"or":[43],"training":[44],"cost.":[45],"Candidate-selection":[46],"methods":[47],"generate":[48],"score":[50],"multiple":[51],"videos,":[52],"while":[53],"gradient-based":[54],"world-model":[55,89],"guidance":[56],"repeatedly":[57],"decodes":[58],"re-encodes":[60],"intermediate":[61],"estimates.":[62],"Generator-internal":[63],"refinement":[64],"adds":[65],"perturbation":[66],"re-denoising":[68],"loops,":[69],"whereas":[70],"post-training":[71],"requires":[72,208],"curated":[73],"additional":[76],"optimization.":[77],"We":[78],"propose":[79],"Off-Manifold":[80],"Refinement":[81],"(OMR),":[82],"an":[83,111],"inference-time":[84],"method":[85],"that":[86],"instead":[87],"injects":[88],"feedback":[90],"directly":[91],"into":[92],"a":[93,201],"single":[94],"sampling":[95,128],"trajectory.":[96],"During":[97],"scheduled":[98],"middle":[99],"ODE":[100],"steps,":[101],"we":[102],"augment":[103],"the":[104,108,122,126,139,144,149,158,165,169,183,196,211,216],"generator":[105,145,167],"velocity":[106],"with":[107],"gradient":[109,159],"of":[110,219],"adapter-space":[112],"V-JEPA":[113],"2.1":[114],"surprise":[115],"energy.":[116],"This":[117],"external":[118],"correction":[119],"move":[121],"latent":[123],"away":[124],"from":[125,148,187],"uncorrected":[127],"trajectory":[129],"toward":[131],"regions":[132],"ranked":[133],"as":[134],"more":[135],"physically":[136],"plausible":[137],"by":[138],"frozen":[140],"predictor,":[141],"after":[142],"which":[143],"continues":[146],"rendering":[147],"corrected":[150],"state.":[151],"A":[152],"small":[153],"trained":[154],"latent-to-embedding":[155],"adapter":[156],"keeps":[157],"tractable":[160],"inference,":[162],"both":[164],"world":[170],"model":[171],"remain":[172],"frozen.":[173],"On":[174,200],"our":[175],"fixed":[176,203],"400-prompt":[177],"VideoPhy-2":[178],"detailed":[179],"subset,":[180,206],"OMR":[181],"lifts":[182],"joint":[184],"Semantic-Adherence-and-Physical-Commonsense":[185],"metric":[186],"47.0%":[188],"to":[189],"52.0%":[190],"(+5.0pp":[191],"absolute,":[192],"+10.6%":[193],"relative)":[194],"over":[195],"base":[197,212],"Wan2.2-T2V-A14B":[198],"sampler.":[199],"separate":[202],"50-prompt":[204],"efficiency":[205],"it":[207],"$1.71":[209],"\\times$":[210],"runtime":[213],"rather":[214],"than":[215],"multiplicative":[217],"cost":[218],"reward/search":[220],"alternatives.":[221],"Project":[222],"page:":[223],"https://itruonghai.github.io/omr.":[224]},"counts_by_year":[{"year":2026,"cited_by_count":2}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-02T00:00:00"}