{"id":"https://openalex.org/W7165377043","doi":"https://doi.org/10.48550/arxiv.2606.20364","title":"Judging to Improve: A De-biased VLM-as-3D-Judge Protocol for Single-Image 3D Generation","display_name":"Judging to Improve: A De-biased VLM-as-3D-Judge Protocol for Single-Image 3D Generation","publication_year":2026,"publication_date":"2026-06-18","ids":{"openalex":"https://openalex.org/W7165377043","doi":"https://doi.org/10.48550/arxiv.2606.20364"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.20364","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20364","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,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.20364","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5138402369","display_name":"Ali Asaria","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Asaria, Ali","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138467146","display_name":"Tony Salomone","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Salomone, Tony","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138496190","display_name":"Deep Gandhi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gandhi, Deep","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"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/T10481","display_name":"Computer Graphics and Visualization Techniques","score":0.2838999927043915,"subfield":{"id":"https://openalex.org/subfields/1704","display_name":"Computer Graphics and Computer-Aided Design"},"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/T10481","display_name":"Computer Graphics and Visualization Techniques","score":0.2838999927043915,"subfield":{"id":"https://openalex.org/subfields/1704","display_name":"Computer Graphics and Computer-Aided Design"},"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/T10719","display_name":"3D Shape Modeling and Analysis","score":0.1454000025987625,"subfield":{"id":"https://openalex.org/subfields/2206","display_name":"Computational Mechanics"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.12559999525547028,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/ranking","display_name":"Ranking (information retrieval)","score":0.5724999904632568},{"id":"https://openalex.org/keywords/matching","display_name":"Matching (statistics)","score":0.47870001196861267},{"id":"https://openalex.org/keywords/protocol","display_name":"Protocol (science)","score":0.4603999853134155},{"id":"https://openalex.org/keywords/hardening","display_name":"Hardening (computing)","score":0.3330000042915344},{"id":"https://openalex.org/keywords/probabilistic-logic","display_name":"Probabilistic logic","score":0.32350000739097595},{"id":"https://openalex.org/keywords/conditioning","display_name":"Conditioning","score":0.322299987077713},{"id":"https://openalex.org/keywords/calibration","display_name":"Calibration","score":0.2939999997615814}],"concepts":[{"id":"https://openalex.org/C189430467","wikidata":"https://www.wikidata.org/wiki/Q7293293","display_name":"Ranking (information retrieval)","level":2,"score":0.5724999904632568},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5019999742507935},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.47870001196861267},{"id":"https://openalex.org/C2780385302","wikidata":"https://www.wikidata.org/wiki/Q367158","display_name":"Protocol (science)","level":3,"score":0.4603999853134155},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.39739999175071716},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3564999997615814},{"id":"https://openalex.org/C44255700","wikidata":"https://www.wikidata.org/wiki/Q978423","display_name":"Hardening (computing)","level":3,"score":0.3330000042915344},{"id":"https://openalex.org/C49937458","wikidata":"https://www.wikidata.org/wiki/Q2599292","display_name":"Probabilistic logic","level":2,"score":0.32350000739097595},{"id":"https://openalex.org/C45262634","wikidata":"https://www.wikidata.org/wiki/Q5159291","display_name":"Conditioning","level":2,"score":0.322299987077713},{"id":"https://openalex.org/C200601418","wikidata":"https://www.wikidata.org/wiki/Q2193887","display_name":"Reliability engineering","level":1,"score":0.30730000138282776},{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.2939999997615814},{"id":"https://openalex.org/C146249460","wikidata":"https://www.wikidata.org/wiki/Q2914991","display_name":"Fixation (population genetics)","level":3,"score":0.29170000553131104},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.2870999872684479},{"id":"https://openalex.org/C4554734","wikidata":"https://www.wikidata.org/wiki/Q593744","display_name":"Knowledge base","level":2,"score":0.2858000099658966},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.2768000066280365},{"id":"https://openalex.org/C76178495","wikidata":"https://www.wikidata.org/wiki/Q4808784","display_name":"Asset (computer security)","level":2,"score":0.2741999924182892},{"id":"https://openalex.org/C42058472","wikidata":"https://www.wikidata.org/wiki/Q810214","display_name":"Base (topology)","level":2,"score":0.27230000495910645},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.26080000400543213},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.25440001487731934}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.20364","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20364","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,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.20364","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.20364","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,"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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"A":[0],"companion":[1],"study":[2],"established":[3],"a":[4,30,58,83,183,242],"de-biased,":[5],"cross-model":[6],"VLM-as-3D-judge":[7],"that":[8,26,113,232],"reliably":[9],"ranks":[10],"single-image-to-3D":[11],"mesh":[12],"quality":[13],"where":[14,53],"cheap":[15,247],"geometry":[16],"and":[17,41,64,99,110,132,138,182,226,262],"CLIP":[18],"proxies":[19],"fall":[20],"short.":[21],"This":[22],"paper":[23],"asks:":[24],"can":[25],"judge's":[27],"preferences":[28],"specialize":[29],"strong":[31,150,243],"open":[32],"generator,":[33],"TRELLIS,":[34],"on":[35,259],"one":[36],"asset":[37],"class":[38],"(furniture),":[39],"cheaply":[40],"without":[42],"human":[43],"labels?":[44],"Taking":[45],"the":[46,54,62,81,149,186,200,206,217,224,230,263],"judge":[47,60,85,92,264],"from":[48,89,135],"ranking":[49,70],"to":[50,94],"optimization":[51],"is":[52,76,212,229,249,266],"work":[55],"lives.":[56],"Pushing":[57],"VLM":[59],"into":[61],"training":[63,84],"evaluation":[65,91],"loop":[66],"exposes":[67],"failure":[68,103],"modes":[69,104],"never":[71],"triggered,":[72],"so":[73,167],"our":[74,146],"contribution":[75],"an":[77,90,129],"optimization-grade":[78],"hardening":[79],"of":[80],"judge:":[82],"(Qwen2.5-VL-7B)":[86],"held":[87],"distinct":[88],"(InternVL3-8B)":[93],"break":[95],"circularity;":[96],"position-bias":[97],"correction;":[98],"fixes":[100],"for":[101],"three":[102],"(image":[105],"overload,":[106],"geometry-hiding":[107],"splat":[108],"renders,":[109],"reference-free":[111],"judging":[112],"rewards":[114],"clean-but-wrong":[115],"outputs),":[116],"with":[117,140,199,246],"calibration":[118],"evidence":[119],"(clear-gap":[120],"win-rate":[121,208],"0.83-1.0;":[122],"base-vs-base":[123],"~0.5).":[124],"Using":[125],"this":[126],"protocol":[127,265],"as":[128],"independent":[130],"evaluator,":[131],"working":[133],"only":[134],"public":[136,260],"models":[137],"data":[139],"lightweight":[141,257],"parameter-efficient":[142],"adaptation,":[143],"we":[144],"find":[145],"methods":[147],"match":[148],"base":[151,157,245],"rather":[152],"than":[153,256],"exceed":[154],"it.":[155],"Independent":[156],"samples":[158],"carry":[159],"essentially":[160],"no":[161,203],"learnable":[162],"preference":[163],"(0.94":[164],"order-flip":[165],"rate),":[166],"signal":[168],"must":[169],"be":[170],"engineered":[171],"by":[172],"quality-contrastive":[173],"construction.":[174],"Across":[175],"six":[176],"adaptation":[177,248],"methods,":[178],"two":[179],"input":[180],"regimes,":[181],"severity":[184],"sweep,":[185],"most":[187],"targeted":[188],"-":[189,195],"conditioner":[190],"repair":[191,228],"under":[192],"severe":[193],"degradation":[194],"reaches":[196],"parity":[197],"(0.50)":[198],"base,":[201],"while":[202],"method":[204],"clears":[205],">=65%":[207],"target.":[209],"The":[210],"result":[211],"mechanistic:":[213],"clean":[214],"inputs":[215],"saturate":[216],"judge,":[218],"flow-DIT":[219],"fine-tuning":[220],"washes":[221],"out":[222],"through":[223],"sampler,":[225],"conditioning":[227],"locus":[231],"moves":[233],"geometry.":[234],"Win-rates":[235],"are":[236],"directional":[237],"at":[238],"n=8":[239],"objects.":[240],"Matching":[241],"public-data":[244],"itself":[250],"informative:":[251],"exceeding":[252],"it":[253],"needs":[254],"more":[255],"PEFT":[258],"data,":[261],"reusable.":[267]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-20T00:00:00"}