Alexandre Mercier-Aubin

@ScottTRichmond XPBD is often used for games, and this paper cheaply improves convergence. For a proper implementation I expect it to run smoothly, especially with static layers. However, the prototype is in Matlab, so it would be wise to port it to C++/Cuda.

@AlexMercierA This looks great mate! I haven't yet had a chance to really dive into it, but do you think this is generally fast enough to use in real-time apps like games?

Missed #SCA2024? Maybe you want to see my crazy multigrid-like solver for deformables? Paper, presentation, and code are on my website: alexandremercieraubin.com/Work/papers/Mu…

Was fun to help out as a volunteer. I also had my first experience as a session chair for the physics 1 session @SympCompAnim . It certainly was a nice experience and I hope to do it again soon.

@amirvaxman_dgp @diwlevin Uh oh, here come the questions😂! Let's just take a second to contemplate that early video game NPCs were considered AI. Now you can also bring up the AI effect and completely obliterate any discussion.

@diwlevin @AlexMercierA it's the other way around "it's not intelligence it's optimization".

heraldscotland.com/news/24521881.… "The unnamed figure told the Financial Times that the project was too focused on “traditional computing projects such as scientific simulations” rather than artificial intelligence." I facepalmed so hard I'm now non-orientable. 🤦🤦‍♀️🤦‍♂️

@diwlevin @amirvaxman_dgp You can sell physical simulations as Artificial Intelligence. At the end of the day a simulation is some form of intelligence exhibited by a machine.

@amirvaxman_dgp How dare they try and make a computer that wasn’t going to be used for copyright theft and dank memes !

@GoddenThomas Have you tried sympy?

In my defence I really do not feel like doing derivatives by hand right now.

Opening MATLAB and questioning my life choices

@sellan_s @Columbia Being late to the party here as I never fire up twitter. Congratulations for the position, you really deserve this one. Tbf, you are one of the most prolific graphics researchers and you getting to keep doing what you do is amazing news.

I probably should've posted this here last year since it seems that the graphics community likes to gather here. The presentation, code, and paper are all available for free. Have a look at our cool work on Adaptive Rigidification of Discrete Shells alexandremercieraubin.com/Work/papers/Ad…

@diwlevin As someone who just started as a reviewer, it feels a bit daunting to directly let it go public. I think it could be nice to have a new reviewer track so we can get more feedback and learn to be better at it rather than expecting perfection from the get go.

5. Allow reviewers to opt-in to have their reviews deanonymized 6. Deanonymized reviewers eligible for best reviewer prize 7. If we must do meta reviews release them before the rebuttal is due

Some things SIGGRAPH could do to monitor/improve review quality 1. Openreview 2. Anonymize the post rebuttal discussion and make it public 3. Publish anonymous reviews for all accepted submissions 4. Let authors of accepted submissions review their reviewers

@diwlevin Just pick a github pages template and stick to it. This way you can just drop in a md file in the repo and it updates your page automatically. Here is mine as a reference: alexandremercieraubin.com/Work/

I've ignored my webpage for a whole year but I'd finally like to update it. What's the best framework for creating a non-ugly, easy to update page ?

@AllaSheffer @adamhill While writing the paper is harder, presenting a one-take live which will live on the internet forever is a much more stressful experience. You just can't get the same level of polish. Why not allow both options? Either send a video or record it live.

@adamhill As a non-native English speaker (English is my 3rd language), doing a 10min presentation is 100 times easier than writing a research paper in English (a pre-req to presenting). All non-native English speakers who spoke at the town hall argued for removing the pre-recorded video.

A few SIGGRAPH'23 Technical Papers updates that I am sure many will be happy to hear: 1) We will stream/record talks live instead of asking authors for per-recordings.

@sellan_s @ana_dodik Definitely easier to explain to your relatives than a research project.

@sellan_s @amirvaxman_dgp In my limited experience, the 15 mins videos were of a much higher quality than the live presentations. It probably has to do with being in a stress-free environment. I do agree that it should be allowed to offer both or just one.

@sellan_s It was the same for SIGGRAPH 2022. It is a lot of work, but also a good occasion to make a well polished video about your work. Perhaps they could cut the 3 min video, and allow the possibility to record the 10mins live to replace the 15mins one.

Would it be controversial to say that publicly funded research should be open source and reproducible? @diwlevin, @paulkry, and I got our replicability stamp. I really appreciate the efforts of the Replicability committee. #https-github-com-alexandremercieraubin-adaptiverigidification2022" target="_blank" rel="nofollow noopener">replicabilitystamp.org/#https-github-… #replicability

In-person SIGGRAPH was fun. Glad I could meet so many talented people. Both on the artsy, and techy sides. For those who missed my presentation on Adaptive Rigidification, the long version will release today on YouTube; youtu.be/mgxGp_U1H_I Can't wait to see you next year!

@sellan_s @OdedStein Just sayin', I bookmarked it.

I will add the full 15 mins presentation to the website once SIGGRAPH is over.

The preprint and the code are both available online. If you need help setting this up, you can join me at the siggraph labs where we will present the codebase we've been working on. Make sure to download matlab and an appropriate c++ compiler for your OS.

Apparently all the kool kids make twitter posts for their #siggraph2022 papers. I may not be cool, but here is something that definitely is fresh and awesome: alexandremercieraubin.com/Work/papers/Ad… Feel free to no longer waste your time computing deformation on non-deforming elements.