AI People Wiki

Builders & open source

Miika Aittala

Member of NVIDIA Research (joined November 2019), NVIDIA

Miika Aittala joined NVIDIA Research in November 2019 after a postdoctoral position at MIT CSAIL, where Aittala worked with Frédo Durand on computer graphics, computer vision, and machine learning. Aittala's MIT page lists research interests in material appearance capture, rendering and light transport, image processing, and deep learning. At NVIDIA, Aittala is the second author of the StyleGAN3, EDM, and EDM2 papers, whose official implementations NVIDIA publishes in its NVlabs GitHub organization. NVIDIA Research also lists Aittala as a co-author of a NeurIPS 2025 paper on rolling diffusion models for probabilistic forecasting.1234567

At a glance

Current role
Member of NVIDIA Research (joined November 2019)
Company
NVIDIA
Sector
Company
Education
MIT
Work listed
2
Sources cited
10
Profile reviewed

Career

Aittala received M.Sc. and Ph.D. degrees from Aalto University, advised by Jaakko Lehtinen, and has worked at or visited Inria's GraphDeco group, the VTT Technical Research Centre of Finland, and NVIDIA Research as an intern. Aittala was then a postdoctoral researcher at MIT CSAIL with Frédo Durand, and in November 2019 joined NVIDIA Research, where NVIDIA lists Aittala's research areas as generative AI and machine learning with a main field of interest in computer graphics.12

Research

Aittala's early papers, many published in the proceedings of SIGGRAPH, dealt with capturing the appearance of real materials and with rendering. They include two-shot and frequency-domain methods for capturing spatially varying reflectance (SVBRDF), reflectance modeling by neural texture synthesis (2016), and gradient-domain path tracing and Metropolis light transport. At MIT, Aittala worked on burst image deblurring with permutation-invariant networks and on blind inverse light transport.1

With Jaakko Lehtinen, Tero Karras, Timo Aila and others, Aittala co-authored Noise2Noise (2018), which trains image restoration networks using only noisy examples. Since joining NVIDIA, Aittala has been second author, after Tero Karras, of StyleGAN3 (2021), EDM (2022), EDM2 (2023) and the 2024 autoguidance paper, and in 2025 co-authored elucidated rolling diffusion models for probabilistic forecasting of complex dynamics.834597

Work and contributions

  • EDM and EDM2 diffusion models Published research

    Second author, with Tero Karras, Timo Aila, Samuli Laine, and others, of "Elucidating the Design Space of Diffusion-Based Generative Models" (EDM, 2022), "Analyzing and Improving the Training Dynamics of Diffusion Models" (EDM2, 2023), and "Guiding a Diffusion Model with a Bad Version of Itself" (2024). The EDM2 repository releases code and pretrained models under CC BY-NC-SA 4.0.4596

  • StyleGAN3 (Alias-Free Generative Adversarial Networks) Published research

    Second author of the NeurIPS 2021 StyleGAN3 paper; the official implementation in NVlabs lists the paper's authors and is released under the NVIDIA Source Code License.3102

Timeline

  1. Published "Practical SVBRDF Capture in the Frequency Domain" in the proceedings of SIGGRAPH 2013.1
  2. Co-authored "Two-shot SVBRDF Capture for Stationary Materials" and "Gradient-Domain Path Tracing" in the proceedings of SIGGRAPH 2015.1
  3. Co-authored "Burst Image Deblurring Using Permutation Invariant Convolutional Neural Networks" with Frédo Durand, an ECCV 2018 oral paper.1
  4. Co-authored "Noise2Noise: Learning Image Restoration without Clean Data".8
  5. Joined NVIDIA Research after a postdoc at MIT CSAIL.1
  6. Posted "Alias-Free Generative Adversarial Networks" (StyleGAN3) as second author.3
  7. Posted "Elucidating the Design Space of Diffusion-Based Generative Models" (EDM) as second author.4
  8. Posted the EDM2 paper on the training dynamics of diffusion models as second author.5
  9. Co-authored "Guiding a Diffusion Model with a Bad Version of Itself".9
  10. Co-authored "Elucidated Rolling Diffusion Models for Probabilistic Forecasting of Complex Dynamics".7

Other documented roles

Roles the sources document that are not confirmed as current.

  • Postdoctoral researcher, working with Frédo Durand, MIT CSAIL1

Where they work

Suggest a correction or removal

Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.

References

10 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026

  1. Miika Aittala — MIT CSAIL pageMiika Aittala (MIT CSAIL) · Official page · accessed Oct 21, 2026
  2. Miika Aittala — NVIDIA ResearchNVIDIA Research · Official page · accessed Oct 21, 2026
  3. Alias-Free Generative Adversarial Networks (arXiv 2106.12423)arXiv · Paper · published Jun 23, 2021 · accessed Oct 21, 2026
  4. Elucidating the Design Space of Diffusion-Based Generative Models (arXiv 2206.00364)arXiv · Paper · published Jun 1, 2022 · accessed Oct 21, 2026
  5. Analyzing and Improving the Training Dynamics of Diffusion Models (arXiv 2312.02696)arXiv · Paper · published Dec 5, 2023 · accessed Oct 21, 2026
  6. NVlabs/edm2 README (EDM2 and Autoguidance; CC BY-NC-SA 4.0)NVlabs (GitHub repository) · Repository · accessed Oct 21, 2026
  7. Elucidated Rolling Diffusion Models for Probabilistic Forecasting of Complex Dynamics (arXiv 2506.20024)arXiv · Paper · published Jun 24, 2025 · accessed Oct 21, 2026
  8. Noise2Noise: Learning Image Restoration without Clean Data (arXiv 1803.04189)arXiv · Paper · published Mar 12, 2018 · accessed Oct 21, 2026
  9. Guiding a Diffusion Model with a Bad Version of Itself (arXiv 2406.02507)arXiv · Paper · published Jun 4, 2024 · accessed Oct 21, 2026
  10. NVlabs/stylegan3 README (official PyTorch implementation; NVIDIA Source Code License)NVlabs (GitHub repository) · Repository · accessed Oct 21, 2026