{"id":{"repo_id":"mit","oai_identifier":"oai:dspace.mit.edu:1721.1/164029"},"canonical_url":"https://search.dev.ndltd.org/etd/mit/oai:dspace.mit.edu:1721.1/164029","repository":{"repo_id":"mit","name":"MIT","base_url":"https://dspace.mit.edu/oai/request"},"display":{"title":"Efficient Generative Models for Visual Synthesis","abstract":"While current visual generative models produce high-quality outputs, they suffer from significant computational costs and latency, limiting their applicability in interactive settings. In this dissertation, we introduce a suite of techniques designed to enhance the efficiency of generative models for image and video synthesis. First, we propose distribution matching distillation, a method that enables the training of one- or few-step visual generators by distilling knowledge from computationally expensive yet highly capable diffusion models. Next, we develop improved distillation techniques that enhance robustness and scalability, culminating in a production-grade few-step image generator. This system is now deployed in widely used software, generating hundreds of millions of images annually. Finally, we extend our approach to video generation by adopting an autoregressive paradigm, significantly reducing latency and enabling fast interactive video generation and world simulation.","abstract_html":"While current visual generative models produce high-quality outputs, they suffer from significant computational costs and latency, limiting their applicability in interactive settings. In this dissertation, we introduce a suite of techniques designed to enhance the efficiency of generative models for image and video synthesis. First, we propose distribution matching distillation, a method that enables the training of one- or few-step visual generators by distilling knowledge from computationally expensive yet highly capable diffusion models. Next, we develop improved distillation techniques that enhance robustness and scalability, culminating in a production-grade few-step image generator. This system is now deployed in widely used software, generating hundreds of millions of images annually. Finally, we extend our approach to video generation by adopting an autoregressive paradigm, significantly reducing latency and enabling fast interactive video generation and world simulation.","abstract_has_math":false,"creators":["Yin, Tianwei"],"institution":"Massachusetts Institute of Technology","degree_name":"Doctoral","degree_level":null,"degree_discipline":null,"degree_department":"Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science","school":null,"contributors":[],"advisors":["Freeman, William T.","Durand, Frédo"],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-05","date_published":"2025-05","updated_at":"2026-07-22T22:21:53Z","subjects":[],"languages":[],"rights":["In Copyright - Educational Use Permitted","Copyright retained by author(s)"],"rights_urls":["https://rightsstatements.org/page/InC-EDU/1.0/"],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/1721.1/164029","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Freeman, William T.","Durand, Frédo"]},{"key":"dc:contributor.department","label":"Department","values":["Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science"]},{"key":"dc:creator","label":"Author","values":["Yin, Tianwei"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2025-11-25T19:37:24Z"]},{"key":"dc:date.available","label":"Dc Date Available","values":["2025-11-25T19:37:24Z"]},{"key":"dc:date.issued","label":"Date","values":["2025-05"]},{"key":"dc:publisher","label":"Institution","values":["Massachusetts Institute of Technology"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["Doctoral","Doctor of Philosophy"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:rights","label":"Dc Rights","values":["In Copyright - Educational Use Permitted","Copyright retained by author(s)"]},{"key":"dc:rights.uri","label":"Rights URI","values":["https://rightsstatements.org/page/InC-EDU/1.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/1721.1/164029"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["While current visual generative models produce high-quality outputs, they suffer from significant computational costs and latency, limiting their applicability in interactive settings. In this dissertation, we introduce a suite of techniques designed to enhance the efficiency of generative models for image and video synthesis. First, we propose distribution matching distillation, a method that enables the training of one- or few-step visual generators by distilling knowledge from computationally expensive yet highly capable diffusion models. Next, we develop improved distillation techniques that enhance robustness and scalability, culminating in a production-grade few-step image generator. This system is now deployed in widely used software, generating hundreds of millions of images annually. Finally, we extend our approach to video generation by adopting an autoregressive paradigm, significantly reducing latency and enabling fast interactive video generation and world simulation."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Ph.D."]},{"key":"dc:title","label":"Title","values":["Efficient Generative Models for Visual Synthesis"]}]}],"canonical_facts":{"dc:contributor.advisor":["Freeman, William T.","Durand, Frédo"],"dc:contributor.department":["Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science"],"dc:creator":["Yin, Tianwei"],"dc:date.accessioned":["2025-11-25T19:37:24Z"],"dc:date.available":["2025-11-25T19:37:24Z"],"dc:date.issued":["2025-05"],"dc:description.abstract":["While current visual generative models produce high-quality outputs, they suffer from significant computational costs and latency, limiting their applicability in interactive settings. In this dissertation, we introduce a suite of techniques designed to enhance the efficiency of generative models for image and video synthesis. First, we propose distribution matching distillation, a method that enables the training of one- or few-step visual generators by distilling knowledge from computationally expensive yet highly capable diffusion models. Next, we develop improved distillation techniques that enhance robustness and scalability, culminating in a production-grade few-step image generator. This system is now deployed in widely used software, generating hundreds of millions of images annually. Finally, we extend our approach to video generation by adopting an autoregressive paradigm, significantly reducing latency and enabling fast interactive video generation and world simulation."],"dc:description.degree":["Ph.D."],"dc:identifier.uri":["https://hdl.handle.net/1721.1/164029"],"dc:publisher":["Massachusetts Institute of Technology"],"dc:rights":["In Copyright - Educational Use Permitted","Copyright retained by author(s)"],"dc:rights.uri":["https://rightsstatements.org/page/InC-EDU/1.0/"],"dc:title":["Efficient Generative Models for Visual Synthesis"],"dc:type":["Thesis"],"thesis:degree_name":["Doctoral","Doctor of Philosophy"]},"updated_at":"2026-07-22T22:21:53Z"}