mauricio delbracio

I'm a Senior Staff Research Scientist at Google, working at the intersection of computational photography and generative AI. Since 2019, I have helped pioneer the development and on-device deployment of generative models, driving flagship features such as Pixel 10’s 100x ProResZoom, ZoomEnhance, and Photo Unblur.

My research focuses on novel scaling and distillation techniques that bring state-of-the-art imaging capabilities to mobile devices globally. My work has been recognized with the SIAM Imaging Early Career Prize and a TMLR Outstanding Paper Award. Beyond research, I serve as an Associate Editor for IEEE TCI and TMLR.

Previously, I was an Assistant Professor in Electrical Engineering at Universidad de la República, Montevideo, Uruguay. Before that, I was a postdoctoral researcher with Guillermo Sapiro at ECE, Duke University. And before that, I was a Ph.D. student at CMLA, ENS-Cachan, France under the supervision of Jean-Michel Morel, Pablo Musé and Andrés Almansa.

Full list of publications in Google Scholar.

mdelbra [at] gmail.com

Papers

The Geometry of Noise: Why Diffusion Models Don't Need Noise Conditioning

M. Sahraee-Ardakan, M. Delbracio, P. Milanfar

Preprint, 2026

[arXiv]

Learning from a Generative Oracle: Domain Adaptation for Restoration

Y. Hu, M. Sahraee-Ardakan, A. Bansal, K. Mei, C. Qi, P. Milanfar, M. Delbracio

Preprint, 2025

[arXiv]

Learn to guide your diffusion model

A. Galashov, A. Pokle, A. Doucet, A. Gretton, M. Delbracio, V. De Bortoli

ICLR, 2026

[arXiv]

Kernel density steering: Inference-time scaling via mode seeking for image restoration

Y. Hu, K. Mei, M. Sahraee-Ardakan, U.S. Kamilov, P. Milanfar, M. Delbracio

NeurIPS, 2025

[arXiv]

Denoising: a powerful building block for imaging, inverse problems and machine learning

P. Milanfar, M. Delbracio

Philosophical Transactions of the Royal Society A, 2025

[web] [arXiv]

TextSR: diffusion super-resolution with multilingual OCR guidance

K. Ye, I.G. Dorado, M. Raptis, M. Delbracio, I. Zhu, P. Milanfar, H. Talebi

Preprint, 2025

[arXiv]

Reference-guided identity preserving face restoration

M. Zhou, K. Ye, V. Shah, K. Mei, M. Delbracio, P. Milanfar, V.M. Patel, H. Talebi

Preprint, 2025

[arXiv]

UniRes: Universal image restoration for complex degradations

M. Zhou, K. Ye, M. Delbracio, P. Milanfar, V.M. Patel, H. Talebi

ICCV, 2025

[arXiv]

The power of context: How multimodality improves image super-resolution

K. Mei, H. Talebi, M. Ardakani, V.M. Patel, P. Milanfar, M. Delbracio

CVPR, 2025

[arXiv]

The Diffusion Process as a Correlation Machine: Linear Denoising Insights

D. Weitzner, M. Delbracio, P. Milanfar, R. Giryes

TMLR, 2025

[openreview] [arXiv]

Stochastic deep restoration priors for imaging inverse problems

Y. Hu, A. Peng, W. Gan, P. Milanfar, M. Delbracio, U.S. Kamilov

ICML, 2025

[arXiv]

A survey on diffusion models for inverse problems

G. Daras, H. Chung, C.H. Lai, Y. Mitsufuji, J.C. Ye, P. Milanfar, A.G. Dimakis

Preprint, 2024

[arXiv]

Spire: Semantic prompt-driven image restoration

C. Qi, Z. Tu, K. Ye, M. Delbracio, P. Milanfar, Q. Chen, H. Talebi

ECCV, 2024

[arXiv]

Bigger is not always better: Scaling properties of latent diffusion models

K. Mei, Z. Tu, M. Delbracio, H. Talebi, V.M. Patel, P. Milanfar

TMLR, 2024

[arXiv]

Tip: Text-driven image processing with semantic and restoration instructions

C. Qi, Z. Tu, K. Ye, M. Delbracio, P. Milanfar, Q. Chen, H. Talebi

Preprint, 2023

[arXiv]

A Restoration Network as an Implicit Prior

Y, Hu, M. Delbracio, P. Milanfar, U.S. Kamilov

ICLR, 2024

[openreview] [arXiv]

CoDi: Conditional Diffusion Distillation for Higher-Fidelity and Faster Image Generation

K. Mei, M. Delbracio, H. Talebi, Z. Tu, V.M. Patel, P. Milanfar

CVPR, 2024

[arXiv]

Prompt-tuning latent diffusion models for inverse problems

H. Chung, J.C. Ye, P. Milanfar, M. Delbracio

ICML, 2024

[web] [arXiv]

Inversion by Direct Iteration: An Alternative to Denoising Diffusion for Image Restoration

M. Delbracio, P. Milanfar

TMLR, 2023 (Outstanding Certification)

[openreview] [web] [arXiv]

Multiscale Structure Guided Diffusion for Image Deblurring

M. Ren, M. Delbracio, H. Talebi, G. Gerig, P. Milanfar

ICCV, 2023

[pdf] [arXiv]

Soft Diffusion: Score matching for general corruptions

G. Daras, M. Delbracio, H. Talebi, A. G. Dimakis, P. Milanfar

TMLR, 2023

[openreview] [arXiv]

Deblurring via stochastic refinement

J, Whang, M. Delbracio, H. Talebi, C. Saharia, A. G. Dimakis, P. Milanfar

CPVR, 2022

[pdf] [arXiv]

Interpretable Unsupervised Diversity Denoising and Artefact Removal

M. Prakash, M. Delbracio, P. Milanfar, F. Jug

ICLR, 2022

[openreview] [arXiv] [github]

Mobile Computational Photography: A Tour

M. Delbracio, D. Kelly, M. S. Brown, P. Milanfar

Annual Review of Vision Science, 2021

[doi] [arXiv]

Projected Distribution Loss for Image Enhancement

M. Delbracio, H. Talebi, P. Milanfar

ICCP, 2021

[doi] [arXiv]

Learning to Reduce Defocus Blur by Realistically Modeling Dual-Pixel Data

A. Abuolaim, M. Delbracio, D. Kelly, M.S. Brown, P. Milanfar

ICCV, 2021

[pdf] [arXiv]

Polyblur: Removing mild blur by polynomial reblurring

M. Delbracio, I. Garcia-Dorado, S. Choi, D. Kelly, P. Milanfar

IEEE TCI, 2021

[pdf] [arXiv]

Single Image Non-uniform Blur Kernel Estimation via Adaptive Basis Decomposition

G. Carbajal, P. Vitoria, M. Delbracio, P. Musé, J. Lezama

Preprint, 2021

[arXiv]

Differential 3D Facial Recognition: Adding 3D to Your State-of-the-Art 2D Method

J.M. Di Martino, F. Suzacq, M. Delbracio, Q. Qiu, G. Sapiro

IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2020

[doi] [arXiv]

Solving Inverse Problems by Joint Posterior Maximization with a VAE Prior

M. Gonzalez, A. Almansa, M. Delbracio, P. Musé, P. Tan

Preprint, 2019

[arXiv]

Efficient blind deblurring under high noise levels

J. Anger, M. Delbracio, G. Facciolo

International Symposium on Image and Signal Processing and Analysis (ISPA), 2019

[arXiv]

Image Anomalies: a Review and Synthesis of Detection Methods

T. Ehret, A. Davy, J.-M. Morel, M. Delbracio

Journal of Mathematical Imaging and Vision (JMIV), 2019 (to appear)

[arXiv]

Modeling Realistic Degradations in Non-blind Deconvolution

J. Anger, M. Delbracio, G. Facciolo

IEEE International Conference on Image Processing (ICIP), 2018

[arXiv]

Reducing Anomaly Detection in Images to Detection in Noise

A. Davy, T. Ehret, J.-M., M. Delbracio

IEEE International Conference on Image Processing (ICIP), 2018

A Practical Guide to Multi-image Alignment

C. Aguerrebere, M. Delbracio, A. Bartesaghi, G. Sapiro

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2018

[arXiv]

Deep Video Deblurring

S. Su, M. Delbracio, J. Wang, G. Sapiro, W. Heidrich, O. Wang

IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2017

[web] [code] [arXiv]

Fundamental Limits in Multi-image Alignment

C. Aguerrebere, M. Delbracio, A. Bartesaghi, G. Sapiro

IEEE Transactions on Signal Processing (TSP), 2016

[doi] [arXiv] [bib]

An analysis of the factors affecting keypoint stability in scale-space

I. Rey-Otero, J.-M. Morel, M. Delbracio

Journal of Mathematical Imaging and Vision (JMIV), 2016

[doi] [arXiv] [bib]

Hand-held Video Deblurring via Efficient Fourier Aggregation

M. Delbracio and G. Sapiro

IEEE Transactions on Computational Imaging (TCI), 2015

[web] [doi] [arXiv] [bib]

Removing Camera Shake via Weighted Fourier Burst Accumulation

M. Delbracio and G. Sapiro

IEEE Transactions on Image Processing (TIP), 2015

[web] [doi] [arXiv] [bib]

Is Repeatability an Unbiased Criterion for Ranking Feature Detectors?

I. Rey-Otero, M. Delbracio

SIAM Journal on Imaging Sciences (SIIMS), 2015

[doi] [arXiv] [bib]

Burst Deblurring: Removing Camera Shake Through Fourier Burst Accumulation

M. Delbracio and G. Sapiro

IEEE International Conference on Computer Vision and Pattern Recognition (CVPR), 2015

[web] [doi] [preprint] [bib]

An analysis of scale-space sampling in SIFT

I. Rey-Otero, J.-M. Morel and M. Delbracio

IEEE International Conference on Image Processing (ICIP), 2014

[doi] [preprint] [code] [bib]

Boosting Monte Carlo Rendering by Ray Histogram Fusion

M. Delbracio, P. Musé, A. Buades, J. Chauvier, N. Phelps and J.-M. Morel

Transactions on Graphics (TOG), January 2014.

[web] [doi] [preprint] [demo] [code] [bib]

Subpixel Point Spread Function Estimation from Two Photographs at Different Distances

M. Delbracio, A. Almansa, J.-M. Morel and P. Muse

SIAM Journal on Imaging Sciences (SIIMS), November 2012.

[doi] [preprint] [demo] [code] [bib]

The Non-parametric Sub-pixel Local Point Spread Function Estimation Is a Well Posed Problem

M. Delbracio, P. Muse, A. Almansa and J.-M. Morel

International Journal of Computer Vision (IJCV'2012)

[doi] [preprint] [demo] [code] [bib]

Aguara: An Improved Face Recognition Algorithm through Gabor Filter Adaptation

C. Aguerrebere, G. Capdehourat, M. Delbracio, M. Mateur, A. Fernandez and F. Lecumberry.

IEEE Workshop on Automatic Identification Advanced Technologies (AutoID 2007)

[doi] [preprint]

IPOL - Image Processing Online :: reproducible research

How to Reduce Anomaly Detection in Images to Anomaly Detection in Noise

T. Ehret, A. Davy, M. Delbracio, J.-M. Morel

IPOL, 2019.

[doi] [web] [demo] [code]

Blind Image Deblurring using the l0 Gradient Prior

J. Anger, G. Facciolo, M. Delbracio

IPOL, March 2019.

[doi] [web] [demo] [code]

Estimating an Image's Blur Kernel Using Natural Image Statistics, and Deblurring it: An Analysis of the Goldstein-Fattal Method

J. Anger, G. Facciolo, M. Delbracio

IPOL, September 2018.

[doi] [web] [demo] [code]

Computing an Exact Gaussian Scale-Space

I. Rey Otero, M. Delbracio

IPOL, February 2016.