{"id":{"repo_id":"mcmaster","oai_identifier":"oai:macsphere.mcmaster.ca:11375/33607"},"canonical_url":"https://search.ndltd.org/etd/mcmaster/oai:macsphere.mcmaster.ca:11375/33607","repository":{"repo_id":"mcmaster","name":"McMaster University","base_url":"https://macsphere.mcmaster.ca/server/oai/request"},"display":{"title":"Explicit Motion Representations for Generative Visual Models","abstract":"Generative visual models have achieved strong performance in image and video synthesis, but precise control over motion remains difficult. This thesis studies explicit point-based motion representations for generative visual models. The central argument is that motion can serve as a structured bridge between visual content and generative synthesis when it is represented explicitly, anchored to meaningful references, and stabilized during generation. The thesis develops this argument in two complementary settings. The first setting is sparse motion control for image editing. GoodDrag uses externally specified handle-target point pairs to guide diffusion-based image editing. It addresses two failure modes of diffusion drag editing: accumulated latent perturbation and handle-point feature drift. Alternating Drag and Denoising interleaves motion supervision with diffusion denoising to preserve image fidelity, while Information-Preserving Motion Supervision anchors motion control to the original handle-point feature to preserve semantic correspondence. The second setting is dense future motion generation from video history. TrajLoom generates dense future trajectories and visibility from observed video and past motion. It represents trajectories using Grid-Anchor Offset Encoding, compresses dense motion into coherent latent tokens with TrajLoom-VAE, and samples future motion with TrajLoom-Flow, a history-conditioned flow-matching generator. Boundary hints and on-policy fine-tuning improve history-future continuity and reduce long-horizon drift. Together, GoodDrag and TrajLoom study opposite directions of the relation between motion and generation: using given motion to control visual editing, and generating motion as an intermediate representation for future video synthesis. The thesis shows that explicit motion representations can improve controllability, temporal coherence, and interpretability in generative visual models.","abstract_html":"Generative visual models have achieved strong performance in image and video synthesis, but precise control over motion remains difficult. This thesis studies explicit point-based motion representations for generative visual models. The central argument is that motion can serve as a structured bridge between visual content and generative synthesis when it is represented explicitly, anchored to meaningful references, and stabilized during generation. The thesis develops this argument in two complementary settings. The first setting is sparse motion control for image editing. GoodDrag uses externally specified handle-target point pairs to guide diffusion-based image editing. It addresses two failure modes of diffusion drag editing: accumulated latent perturbation and handle-point feature drift. Alternating Drag and Denoising interleaves motion supervision with diffusion denoising to preserve image fidelity, while Information-Preserving Motion Supervision anchors motion control to the original handle-point feature to preserve semantic correspondence. The second setting is dense future motion generation from video history. TrajLoom generates dense future trajectories and visibility from observed video and past motion. It represents trajectories using Grid-Anchor Offset Encoding, compresses dense motion into coherent latent tokens with TrajLoom-VAE, and samples future motion with TrajLoom-Flow, a history-conditioned flow-matching generator. Boundary hints and on-policy fine-tuning improve history-future continuity and reduce long-horizon drift. Together, GoodDrag and TrajLoom study opposite directions of the relation between motion and generation: using given motion to control visual editing, and generating motion as an intermediate representation for future video synthesis. The thesis shows that explicit motion representations can improve controllability, temporal coherence, and interpretability in generative visual models.","abstract_has_math":false,"creators":["Zhang, Zewei"],"institution":null,"degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":"Electrical and Computer Engineering","school":null,"contributors":[],"advisors":["Chen, Jun"],"committee_chairs":[],"committee_members":[],"year":2026,"date_issued":"2026","date_published":"2026","updated_at":"2026-11-18T19:38:33Z","subjects":["Generative visual models","Motion representations","Diffusion-based image editing","Drag-based image editing","Dense point trajectories","Future motion generation","Flow matching","Video generation","GoodDrag","TrajLoom"],"languages":["en"],"rights":["Attribution-NonCommercial-NoDerivatives 4.0 International"],"rights_urls":["http://creativecommons.org/licenses/by-nc-nd/4.0/"],"identifier_entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://doi.org/10.71548/25378"],"render_values":[{"text":"https://doi.org/10.71548/25378","href":"https://doi.org/10.71548/25378","code":true}]}],"person_identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/11375/33607","outbound_label":"Handle","outbound_source":"dc:identifier.uri","record_url":"https://hdl.handle.net/11375/33607","record_label":"Repository record","record_kind":"handle","record_source":"dc:identifier.uri","download_url":null,"download_label":null,"download_source":null,"access_status":null,"embargo_until":null,"access_note":null},"source_record":{"url":"https://macsphere.mcmaster.ca/server/oai/request?verb=GetRecord&metadataPrefix=dim&identifier=oai%3Amacsphere.mcmaster.ca%3A11375%2F33607","prefix":"dim"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Chen, Jun"]},{"key":"dc:contributor.department","label":"Department","values":["Electrical and Computer Engineering"]},{"key":"dc:creator","label":"Author","values":["Zhang, Zewei"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2026-09-25T19:31:05Z"]},{"key":"dc:date.issued","label":"Date","values":["2026"]},{"key":"dc:type","label":"Dc Type","values":["Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Generative visual models","Motion representations","Diffusion-based image editing","Drag-based image editing","Dense point trajectories","Future motion generation","Flow matching","Video generation","GoodDrag","TrajLoom"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution-NonCommercial-NoDerivatives 4.0 International"]},{"key":"dc:rights.uri","label":"Rights URI","values":["http://creativecommons.org/licenses/by-nc-nd/4.0/"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/11375/33607","https://doi.org/10.71548/25378"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Generative visual models have achieved strong performance in image and video synthesis, but precise control over motion remains difficult. This thesis studies explicit point-based motion representations for generative visual models. The central argument is that motion can serve as a structured bridge between visual content and generative synthesis when it is represented explicitly, anchored to meaningful references, and stabilized during generation. The thesis develops this argument in two complementary settings. The first setting is sparse motion control for image editing. GoodDrag uses externally specified handle-target point pairs to guide diffusion-based image editing. It addresses two failure modes of diffusion drag editing: accumulated latent perturbation and handle-point feature drift. Alternating Drag and Denoising interleaves motion supervision with diffusion denoising to preserve image fidelity, while Information-Preserving Motion Supervision anchors motion control to the original handle-point feature to preserve semantic correspondence. The second setting is dense future motion generation from video history. TrajLoom generates dense future trajectories and visibility from observed video and past motion. It represents trajectories using Grid-Anchor Offset Encoding, compresses dense motion into coherent latent tokens with TrajLoom-VAE, and samples future motion with TrajLoom-Flow, a history-conditioned flow-matching generator. Boundary hints and on-policy fine-tuning improve history-future continuity and reduce long-horizon drift. Together, GoodDrag and TrajLoom study opposite directions of the relation between motion and generation: using given motion to control visual editing, and generating motion as an intermediate representation for future video synthesis. The thesis shows that explicit motion representations can improve controllability, temporal coherence, and interpretability in generative visual models."]},{"key":"dc:description.degree","label":"Dc Description Degree","values":["Doctor of Philosophy (PhD)"]},{"key":"dc:title","label":"Title","values":["Explicit Motion Representations for Generative Visual Models"]},{"key":"others:access_status","label":"Others Access Status","values":["open.access"]}]}],"canonical_facts":{"dc:contributor.advisor":["Chen, Jun"],"dc:contributor.department":["Electrical and Computer Engineering"],"dc:creator":["Zhang, Zewei"],"dc:date.accessioned":["2026-09-25T19:31:05Z"],"dc:date.issued":["2026"],"dc:description.abstract":["Generative visual models have achieved strong performance in image and video synthesis, but precise control over motion remains difficult. This thesis studies explicit point-based motion representations for generative visual models. The central argument is that motion can serve as a structured bridge between visual content and generative synthesis when it is represented explicitly, anchored to meaningful references, and stabilized during generation. The thesis develops this argument in two complementary settings. The first setting is sparse motion control for image editing. GoodDrag uses externally specified handle-target point pairs to guide diffusion-based image editing. It addresses two failure modes of diffusion drag editing: accumulated latent perturbation and handle-point feature drift. Alternating Drag and Denoising interleaves motion supervision with diffusion denoising to preserve image fidelity, while Information-Preserving Motion Supervision anchors motion control to the original handle-point feature to preserve semantic correspondence. The second setting is dense future motion generation from video history. TrajLoom generates dense future trajectories and visibility from observed video and past motion. It represents trajectories using Grid-Anchor Offset Encoding, compresses dense motion into coherent latent tokens with TrajLoom-VAE, and samples future motion with TrajLoom-Flow, a history-conditioned flow-matching generator. Boundary hints and on-policy fine-tuning improve history-future continuity and reduce long-horizon drift. Together, GoodDrag and TrajLoom study opposite directions of the relation between motion and generation: using given motion to control visual editing, and generating motion as an intermediate representation for future video synthesis. The thesis shows that explicit motion representations can improve controllability, temporal coherence, and interpretability in generative visual models."],"dc:description.degree":["Doctor of Philosophy (PhD)"],"dc:identifier.uri":["https://hdl.handle.net/11375/33607","https://doi.org/10.71548/25378"],"dc:language.iso":["en"],"dc:rights":["Attribution-NonCommercial-NoDerivatives 4.0 International"],"dc:rights.uri":["http://creativecommons.org/licenses/by-nc-nd/4.0/"],"dc:subject":["Generative visual models","Motion representations","Diffusion-based image editing","Drag-based image editing","Dense point trajectories","Future motion generation","Flow matching","Video generation","GoodDrag","TrajLoom"],"dc:title":["Explicit Motion Representations for Generative Visual Models"],"dc:type":["Thesis"],"others:access_status":["open.access"]},"updated_at":"2026-11-18T19:38:33Z"}