2026-09-21T12:18:38Z https://www.ideals.illinois.edu/oai-pmh

oai:www.ideals.illinois.edu:2142/129197 2025-10-20 col_2142_5131 col_2142_10761 com_2142_5130 com_2142_10755 com_2142_234

application/pdf en text Submission original under an indefinite embargo labeled 'Open Access'. The submission was exported from vireo on 2025-10-19 without embargo terms The student, Hao-Yu Hsu, accepted the attached license on 2025-04-13 at 13:44. The student, Hao-Yu Hsu, submitted this Thesis for approval on 2025-04-13 at 13:54. This Thesis was approved for publication on 2025-04-15 at 05:21. DSpace SAF Submission Ingestion Package generated from Vireo submission #21755 on 2025-10-19 at 18:09:20 Physically realistic video editing from natural language instructions Hsu, Hao-Yu 2025-04-15 Wang, Shenlong Visual effects text-guided video editing 3D simulation physical-based editing eng Modern visual effects (VFX) software has made it possible for skilled artists to create imagery of virtually anything. However, the creation process remains laborious, complex, and largely inaccessible to everyday users. In this work, we present AutoVFX, a framework that automatically creates realistic and dynamic VFX videos from a single video and natural language instructions. By carefully integrating neural scene modeling, LLM-based code generation, and physical simulation, AutoVFX is able to provide physically-grounded, photorealistic editing effects that can be controlled directly using natural language instructions. We conduct extensive experiments to validate AutoVFX’s efficacy across a diverse spectrum of videos and instructions. Quantitative and qualitative results suggest that AutoVFX outperforms all competing methods by a large margin in generative quality, instruction alignment, editing versatility, and physical plausibility. 2025-05 Thesis https://hdl.handle.net/2142/129197 © 2025 Hao-Yu Hsu. Siebel School Comp & Data Sci Computer Science University of Illinois Urbana-Champaign M.S. Thesis