[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-380580-105":3,"detail-sidebar-cat-1-en-105":80,"doc-detail-380580-en":126},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":73,"head_meta":75,"extra_data":77,"updated_unix":79},105,"en","culturevidbench-benchmarking-cultural-understanding-in-text-to-video-generation","CultureVidBench - Benchmarking Cultural Understanding in Text-to-Video Generation","","Text-to-video (T2V) generation has improved rapidly, yet cultural context understanding remains insufficiently measured. Existing benchmarks emphasize perceptual quality, physical plausibility, and text-video alignment, without testing whether outputs reflect culturally specific objects, actions, rituals, visible text, or audio cues. CultureVidBench is a benchmark with 1,000 curated prompts across 12 countries, 6 continents, 8 regions, and 14 cultural aspects in three categories.",{"@graph":14,"@context":72},[15,34,55],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/template/","Template",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/template/presentations/","Presentations",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/template/culturevidbench-benchmarking-cultural-understanding-in-text-to-video-generation/380580/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/culturevidbench-benchmarking-cultural-understanding-in-text-to-video-generation/380580.png","ImageObject",442,249,{"name":42,"@type":43},"kopisore","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-19","2026-09-24",true,{"@type":52,"interactionType":53,"userInteractionCount":33},"InteractionCounter",{"@type":54},"ViewAction",{"@type":56,"mainEntity":57},"FAQPage",[58,64,68],{"name":59,"@type":60,"acceptedAnswer":61},"What gap does CultureVidBench address in current T2V evaluation?","Question",{"text":62,"@type":63},"It fills the lack of direct assessment for culturally faithful content, beyond perceptual quality, physical plausibility, and general text-video alignment.","Answer",{"name":65,"@type":60,"acceptedAnswer":66},"What does CultureVidBench include in its evaluation scope?",{"text":67,"@type":63},"It provides 1,000 curated prompts covering 12 countries, 6 continents, 8 cultural regions, and 14 cultural aspects organized into material culture, social practice & performance, and ritual & ceremony.",{"name":69,"@type":60,"acceptedAnswer":70},"How are T2V models evaluated on CultureVidBench?",{"text":71,"@type":63},"Through human user studies and MLLM-based automatic assessment across cultural faithfulness, multimodal cultural rendering, semantic adherence, and perceptual quality.","https://schema.org",{"og:url":32,"og:type":74,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":76,"canonical":32},"index,follow",{"doc_id":78,"site_id":7},380580,1792403943,{"code":4,"msg":81,"data":82},"success",[83,87,92,97,102,107,112,117,122],{"id":84,"doc_module":22,"doc_module_name":25,"category_name":29,"show_sort_weight":85,"slug":86},11,90,"presentations",{"id":88,"doc_module":22,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},12,"Resumes",80,"resumes",{"id":93,"doc_module":22,"doc_module_name":25,"category_name":94,"show_sort_weight":95,"slug":96},14,"Invoices",70,"invoices",{"id":98,"doc_module":22,"doc_module_name":25,"category_name":99,"show_sort_weight":100,"slug":101},15,"Posters",60,"posters",{"id":103,"doc_module":22,"doc_module_name":25,"category_name":104,"show_sort_weight":105,"slug":106},16,"Social Media",50,"social-media",{"id":108,"doc_module":22,"doc_module_name":25,"category_name":109,"show_sort_weight":110,"slug":111},17,"Forms",40,"forms",{"id":113,"doc_module":22,"doc_module_name":25,"category_name":114,"show_sort_weight":115,"slug":116},18,"Letters",30,"letters",{"id":118,"doc_module":22,"doc_module_name":25,"category_name":119,"show_sort_weight":120,"slug":121},21,"Paper Templates",5,"papers-templates",{"id":123,"doc_module":22,"doc_module_name":25,"category_name":124,"show_sort_weight":4,"slug":125},158,"General","general-158",{"code":4,"msg":81,"data":127},{"doc_id":78,"user_id":128,"nickname":42,"user_avatar":129,"doc_module":22,"category_id":84,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":130,"file_id":131,"file_url":132,"file_type":133,"file_size":134,"view_count":33,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":135,"language":136,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":137,"faqs":138,"seo_title":139,"seo_description":12,"update_tm":140,"read_time":141},962090880963,"https://ap-avatar.wpscdn.com/davatar_6f874abed73319feea01a86fa6f0fab8","CultureVidBench: Benchmarking Cultural Understanding in  \nText-to-Video Generation  \nXianjing Han1 , Yuhan Su2 , Yang Deng3 , Dong Ma4 , Wee Peng Tay1 , Bin Zhu3 †  \n1 Nanyang Technological University 2 Centrale Supélec  \n3 Singapore Management University 4 University of Cambridge  \nCorrespondence: [binzhu@smu.edu.sg](binzhu@smu.edu.sg)  \n[https://hanxjing.github.io/CultureVidBench/](https://hanxjing.github.io/CultureVidBench/)  \narXiv : 2608 .0 1942v2 [cs .CV] 27 Aug 2026  \nAbstract  \nText-to-video (T2V) generation models have advanced rapidly, yet their ability to represent diverse cultural contexts remains underexplored. Existing benchmarks mainly focus on perceptual quality, physical plausibility, and text-video alignment, but do not directly assess whether generated videos capture culturally specific objects, actions, rituals, visible text, or audio cues. We introduce CultureVidBench, a comprehensive benchmark for evaluating cultural understanding in T2V generation. CultureVidBench contains 1,000 curated prompts covering 12 countries, 6 continents, 8 cultural regions, and 14 cultural aspects organized into three categories: material culture, social practice & performance, and ritual & ceremony. Designed specifically for video generation, CultureVidBench emphasizes dynamic and multimodal cultural representation, including social interactions, ritual procedure, and culturally appropriate visible text and audio. We evaluate seven representative T2V models through human user studies and MLLM-based automatic assessment across cultural faithfulness, multimodal cultural rendering, semantic adherence, and perceptual quality. Results show that although current models achieve strong semantic adherence and visual quality, they often fail to faithfully capture fine-grained cultural details, particularly for underrepresented regions, rituals, and multimodal cultural cues.  \n1 Introduction  \nRecent advances in text-to-video (T2V) generation have enabled models to produce increasingly realistic, temporally coherent, and semantically aligned videos from natural language prompts (Wan et al., 2025 ; Google DeepMind, 2025 ; Alibaba Cloud, 2026) . These models are rapidly adopted to support a growing range of applications, including filmmaking, social media content creation, advertising, and education (Zheng et al., 2025 ; Chen  \n†Corresponding author and project lead.  \net al., 2026b) . As generated videos become widely consumed across different societies, an important question emerges: can T2V models faithfully understand and represent cultural contexts?  \nCulture is deeply embedded in real-world videos. It shapes not only visible objects such as clothing, food, architecture, and decorations, but also human actions, social interactions, ritual procedures, performances, music, and symbolic practices (Geertz, 2017) . Therefore, cultural understanding in T2V generation cannot be limited to general prompt following or visual realism. As shown in Figure 1(a), a generated video may appear visually plausible and semantically related to the prompt while still misrepresenting the intended culture. For example, a model may depict a traditional Chinese music performance using culturally inconsistent instruments or text, or generate a Nigerian wedding scene with generic attire and ceremonial elements. Such videos may score highly for visual quality and coarse semantic alignment, but they remain culturally inaccurate, misleading, or inappropriate for viewers in the target culture. These failures suggest that cultural understanding is a central requirement for trustworthy T2V generation, rather than a secondary aesthetic preference.  \nDespite its importance, cultural understanding remains largely underexplored in existing T2V evaluation. Current benchmarks primarily assess perceptual quality (Huang et al., 2024), motion consistency (Liao et al., 2024), physical plausibility (Gu et al., 2025), and text-video semantic alignment (Sun et al., 2025) . While ","cbCaijClAQGyA0Ct","https://ap.wps.com/l/cbCaijClAQGyA0Ct","pdf",9130318,19,"English","# Abstract\n# Introduction\n# Categories, 14 Cultural Aspects for Evaluation","[{\"question\":\"What gap does CultureVidBench address in current T2V evaluation?\",\"answer\":\"It fills the lack of direct assessment for culturally faithful content, beyond perceptual quality, physical plausibility, and general text-video alignment.\"},{\"question\":\"What does CultureVidBench include in its evaluation scope?\",\"answer\":\"It provides 1,000 curated prompts covering 12 countries, 6 continents, 8 cultural regions, and 14 cultural aspects organized into material culture, social practice \\u0026 performance, and ritual \\u0026 ceremony.\"},{\"question\":\"How are T2V models evaluated on CultureVidBench?\",\"answer\":\"Through human user studies and MLLM-based automatic assessment across cultural faithfulness, multimodal cultural rendering, semantic adherence, and perceptual quality.\"}]","CultureVidBench - Benchmarking Cultural Understanding in Text-to-Video Generation | PDF",1790240728,7]