PhD studentship – Neural rendering for real-time graphics Full-time

Next Generation Neural Rendering Techniques Studentship

Imagine a future where most of the pixels on your screen are generated, not rendered. This studentship offers the chance to shape that future – exploring how next generation neural rendering techniques can push the boundaries of visual fidelity, performance, and perception.

Building on recent advances such as Intel's XeSS 2 and NVIDIA's DLSS 4, this project will investigate intelligent neural upsampling and frame generation methods that do more than boost resolution. The goal is to design perceptually guided systems that adapt rendering effort in real time – deciding which pixels and frames need to be rendered, and which can be convincingly imagined by the network.

You will explore techniques that increase perceived resolution, synthesize new frames, and dynamically control rendering parameters to deliver exceptional visual quality within strict computational budgets – all while minimizing artifacts that disrupt human perception.

This is an excellent opportunity for a student passionate about computer graphics, vision science, and machine learning to contribute to the next evolution of real time rendering.

The topics of this studentship include: Neural frame generation/extrapolation for real-time rendering; Neural super-sampling; Neural shading; Perceptual optimization of rendering parameters; GPU power optimization.

We are looking for candidates with a strong background and interest in computer graphics and machine learning. Ideally, the candidates should have taken advanced courses, have professional or research experience in those areas. We also recommend consulting the entrance requirements for the PhD programme under the Expected Academic Standard.

The project is a collaboration with LightSpeed Studios. The project is based at the University of Cambridge.

We recommend contacting Prof. Rafal Mantiuk (rafal.mantiuk@cl.cam.ac.uk) in advance to assess topic and background fit. Please include a CV and a 2-paragraph research statement that shows evidence of engagement with this advert. Further information on the PhD in Computer Science programme can be found at: https://www.postgraduate.study.cam.ac.uk/courses/directory/cscspdpcs/apply

All applications should be made online via the University's Applicant Portal: https://www.postgraduate.study.cam.ac.uk/courses/directory/cscspdpcs/apply. Please quote the reference NR48987 in the Research Topic so that applications can be routed directly to Prof. Mantiuk.

Applications should include academic transcripts, a CV, a research proposal, and 2 references. An application is only complete when all supporting documents, including the 2 academic references, are submitted. It is your responsibility to ensure that both referees submit their references before the closing date. The research proposal should expand on at least two topics listed in the bullet points above.

This studentship provides full approved tuition fees and maintenance at recommended UKRI rates for 3 years (the expected duration). Both home and overseas students are welcome to apply.

We encourage groups currently underrepresented in Engineering and Physical Science subjects. Amongst UK-domiciled students, this includes women, Black British, British Bangladeshi and British Pakistani applicants. Amongst UK-domiciled and international applicants, we also particularly welcome applications from people from low-income backgrounds, mature students, care-experienced students, and students from families where no parent or care-giver went to university. Further information can be found on our widening participation webpages https://www.postgraduate.study.cam.ac.uk/apply/before/widening-access

Please quote reference NR48987 on your application and in any correspondence about this vacancy.

Relevant papers: [1] Mantiuk, R.K., Hanji, P., Ashraf, M., Asano, Y., Chapiro, A., 2024. ColorVideoVDP: A visual difference predictor for image, video and display distortions. ACM Transactions on Graphics 43, 129. https://doi.org/10.1145/3658144 [2] Jindal, A., Wolski, K., Myszkowski, K., Mantiuk, R.K., 2021. Perceptual model for adaptive local shading and refresh rate. ACM Transactions on Graphics 40. https://doi.org/10.1145/3478513.3480514 [3] Denes, G., Jindal, A., Mikhailiuk, A., Mantiuk, R.K., 2020. A perceptual model of motion quality for rendering with adaptive refresh-rate and resolution. ACM Trans. Graph. 39. https://doi.org/10.1145/3386569.3392411 [4] Denes, G., Maruszczyk, K., Ash, G., Mantiuk, R.K., 2019. Temporal Resolution Multiplexing: Exploiting the limitations of spatio-temporal vision for more efficient VR rendering. IEEE Trans. Visual. Comput. Graphics 25, 2072-2082. https://doi.org/10.1109/TVCG.2019.2898741

The University actively supports equality, diversity and inclusion and encourages applications from all sections of society.