Tweets
- Tweets, current page.
- Tweets & replies
- Media
You blocked @morgan3d
Are you sure you want to view these Tweets? Viewing Tweets won't unblock @morgan3d
-
Morgan McGuire Retweeted
Turner Whitted's new IEEE ray tracing retrospective: https://ieeexplore.ieee.org/document/8951772 …
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Morgan McGuire Retweeted
The Khronos Vulkan Ray Tracing extension is here! We now have a cross-vendor solution for GPU accelerated ray tracing on Vulkan as well as on DirectX. It is very similar to the DirectX Raytracing and OptiX APIs to make it easy to learn and port to.https://www.khronos.org/blog/ray-tracing-in-vulkan …
Show this threadThanks. Twitter will use this to make your timeline better. UndoUndo -
Morgan McGuire Retweeted
I got through some of my game backlog last weekend. A Short Hike (https://store.steampowered.com/app/1055540/A_Short_Hike/ …, free on Epic this week) Minit (https://store.steampowered.com/app/609490/Minit/ …) Are both friendly, cute Zelda-likes that play very quickly and have great graphics. Perfect for stressed-out adults and for kids.pic.twitter.com/TqcEdCLEUu
Show this threadThanks. Twitter will use this to make your timeline better. UndoUndo -
Lample and Charton's new neural network symbolic mathematics solver includes integration and differentiation, and can beat Mathematica: https://arxiv.org/pdf/1912.01412.pdf …pic.twitter.com/hxl8UQk3FO
Thanks. Twitter will use this to make your timeline better. UndoUndo -
The cheap and cheerful alternative to EMWA and Kalman filters for noisy input at varying frequencies by Casiez et al. 2012: http://cristal.univ-lille.fr/~casiez/acm.php?id=N05397 … http://cristal.univ-lille.fr/~casiez/1euro/ pic.twitter.com/DhfBAWj3gf
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Mitsuba 2: A Retargetable Forward and Inverse Renderer by Nimier-David, Vicini, et al. ToG 2019 describes a technique for building and compiling efficient, modular, and parallel renderers in C++ for both CPUs and GPUs. The full source will soon be public. https://rgl.epfl.ch/publications/NimierDavidVicini2019Mitsuba2 …pic.twitter.com/jLsQBeZAOG
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Low-Discrepancy Blue Noise Sampling, Ahmed et al. 2016 https://projet.liris.cnrs.fr/ldbn/ gives low-discrepancy sample points which are good for quasi-Monte Carlo integration that also have blue-noise properties and are thus perceptually desirable for residual noise and direct visualizationpic.twitter.com/HT6kvQa1Dp
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Non-linear sphere tracing for rendering deformed signed distance fields, Seyb et al. '19 Ray trace skinned SDFs by intersecting a bounding hull and then treating the gradient of the SDF as a flow field to traverse using an ODE solver--like a physics sim. https://cs.dartmouth.edu/~wjarosz/publications/seyb19nonlinear.pdf …pic.twitter.com/HdoscNJVd8
Thanks. Twitter will use this to make your timeline better. UndoUndo -
González-Acuña & Chaparro-Romo, General formula for bi-aspheric singlet lens design free of spherical aberration, Applied Optics 2018 https://www.osapublishing.org/ao/abstract.cfm?uri=ao-57-31-9341 …pic.twitter.com/1RRCPZc6Qo
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Wang et al. Detecting Photoshopped Faces by Scripting Photoshop, 2019 (with code!) detects limited image manipulations using deep learning and attempts to undo them. https://peterwang512.github.io/FALdetector/ pic.twitter.com/70qksiZDqD
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Morgan McGuire Retweeted
In 2016,
@iliyang introduced the concept of distributing MC errors as a blue noise in screen space with Blue-noise Dithered Sampling.@_Laurent and I were inspired by the concept and we have been thinking about it for a while.pic.twitter.com/ZtEGNajxkb
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Kim et al., NVGaze: An Anatomically-Informed Dataset for Low-Latency, Near-Eye Gaze Estimation, CH'2019 uses ray tracing and machine learning to build a fast gaze tracking system and shares their full high-res synthetic training dataset. https://sites.google.com/nvidia.com/nvgaze …pic.twitter.com/smEyXxIc3b
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Morgan McGuire Retweeted
I cobbled together an unofficial updated PDF of the book "Ray Tracing Gems," correcting all known errata (thanks,
@self_shadow and others for your fixes!). Go get it, and see times & places for GDC/GTC signings and talks, here: http://raytracinggems.comThanks. Twitter will use this to make your timeline better. UndoUndo -
Morgan McGuire Retweeted
Just uploaded a preprint of "Real-Time Continuous Level of Detail Rendering of Point Clouds": https://www.cg.tuwien.ac.at/research/publications/2019/schuetz-2019-CLOD/ … Essential source code samples here: https://github.com/m-schuetz/ieeevr_2019_clod … Will follow up with videos/slides after the conference: http://ieeevr.org/2019/ pic.twitter.com/d0o52pWeDf
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Morgan McGuire Retweeted
Baseline Pupil Diameter Is Not a Reliable Biomarker of Subjective Sleepiness, because it has a circadian rhythm and it is sensitive to sleep pressure https://www.frontiersin.org/articles/10.3389/fneur.2019.00108/full …pic.twitter.com/JPgbl4Gd3V
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Morgan McGuire Retweeted
Enriching mesh vertices with filtered quadrics helps smoothing and clustering geometry at high speed, while preserving large salient structures. Find out more in our new
#CGF2019 paper, now online: https://perso.telecom-paristech.fr/boubek/papers/QGF/ …pic.twitter.com/79rXkkJZai
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Geirhos et al.'19 (https://openreview.net/forum?id=Bygh9j09KX …) show that image classification CNNs are remarkably similar to non-ML Bag of Feature models and largely focus on texture over relationships. Explained more clearly by https://medium.com/bethgelab/neural-networks-seem-to-follow-a-puzzlingly-simple-strategy-to-classify-images-f4229317261f …pic.twitter.com/n00rJQZ0WI
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Geirhos et al.'19 (https://openreview.net/forum?id=Bygh9j09KX …) show that image classification CNNs are remarkably similar to non-ML Bag of Feature models and largely focus on texture over relationships. Explained more clearly by https://medium.com/bethgelab/neural-networks-seem-to-follow-a-puzzlingly-simple-strategy-to-classify-images-f4229317261f …pic.twitter.com/1ci5BCTCTC
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Čadík, Perceptual Evaluation of Color-to-Grayscale Image Conversions, PG'08 has terrific evaluations of perceptual metrics for color -> luminance while preserving image content edges. http://cadik.posvete.cz/color_to_gray_evaluation/ …pic.twitter.com/z4qBGF4C34
Thanks. Twitter will use this to make your timeline better. UndoUndo -
Geirho et al. 2019, ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness Counters the overfitting to texture instead of shape of many classifiers by improving the training set. https://medium.com/@robgeirhos/why-deep-learning-works-differently-than-we-thought-ec28823bdbc …pic.twitter.com/8J5vfpbjzp
Thanks. Twitter will use this to make your timeline better. UndoUndo
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.