Skip to content
Technical University of Munich

Master-Seminar – Deep Learning in Computer Graphics (IN2107, IN0014)

Chengyun Wang, Parsa Vatani and Prof. Dr. Nils Thuerey

Time, Place

Tuesdays 14:00-16:00 in room: MI 02.13.010

Begin: Tuesday, Oct. 13, 2026

First official seminar with presentation: Tuesday, Oct. 27, 2026

Further information under: ge.in.tum.de/teaching/seminars-general-info/

PrerequisitesIntroduction to Deep Learning

Content

In this course, students will autonomously investigate recent research about machine learning techniques in computer graphics. Independent investigation for further reading, critical analysis, and evaluation of the topic are required.

Requirements

Participants are required to first read the assigned paper and start writing a report. This will help you prepare for your presentation.

Attendance
  • It is only allowed to miss two sessions. If you have to miss any, please let us know in advance, and write a one-page summary about the paper in your own words. Missing the third one means failing the seminar. 
Report
  • A short report (4 pages max. excluding references in the ACM SIGGRAPH TOG format (acmtog) - you can download the precompiled latex template) should be prepared and sent until January 11th 2027.
  • Guideline: You can begin with writing a summary of the work you present as a start point; but, it would be better if you focus more on your own research rather than just finishing with the summary of the paper. We, including you, are not interested in revisiting the work done before; it is more meaningful if you make an effort to put your own reasoning about the work, such as pros and cons, limitation, possible future work, your own ideas for the issues, etc.
  • For questions regarding your paper or feedback for a semi-final version of your report you can contact your advisor.
Presentation (slides)
  • You will present your topic in English, and the talk should last 15 to 20 minutes. After that, a discussion session of about 5 minutes will follow.
  • The slides should be structured according to your presentation. You can use any layout or template you like, but make sure to choose suitable colors and font sizes for readability.
  • Plagiarism should be avoided; please do not simply copy the original authors' slides. You can certainly refer to them.
  • The semi-final slides (PDF) must be sent within one week before the talk, otherwise the talk will be canceled.
  • We strongly encourage you to finalize the semi-final version as far as possible. We will take a look at the version and give feedback. You can revise your slides until your presentation.
  • The final slides and report should be sent within one week after the talk.

Topics

The paper list for this seminar can be found below. Please send us an email with your 3 ranked preferences by September 29th.  We will assign papers based on these preference. The paper-matching results will be made available before the Kick-off meeting.

Paper NumberPaper
12020, Xiao et al., Neural Supersampling for Real-Time Rendering
2 2020, Mildenhall et al., Representing Scenes as Neural Radiance Fields for View Synthesis
32021, Dosovitskiy et al., An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
42022, Hertz et al., Prompt-to-Prompt Image Editing with Cross Attention Control
52022, Saharia et al., Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding
62023, Kerbl et al., 3D Gaussian Splatting for Real-Time Radiance Field Rendering
72022, Poole et al., DreamFusion: Text-to-3D using 2D Diffusion
82023, Deng et al. , Fluid Simulation on Neural Flow Maps
92024, Hao et al., Meshtron: High-Fidelity, Artist-Like 3D Mesh Generation at Scale
102024, Fan et al., InstantSplat: Sparse-view Gaussian Splatting in Seconds
112024, Genie: Generative Interactive Environments 
122025, Stochastic Preconditioning for Neural Field Optimization
132025, RenderFormer: Transformer-based Neural Rendering of Trangle Meshes with Global Illumination
142025, Cosmos-Transfer1: World Generation with Adaptive Multimodal Control
152025, GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Control
162025, GENMO: A GENeralist Model for Human MOtion
172025, VGGT: Visual Geometry Grounded Transformer
182026, LagrangianSplats: Divergence-Free Transport of Gaussian Primitives for Fluid Reconstruction

References

To top