Jeff Petkau

Not sure why people assume ChatGPT provided the made-up Megalopolis quotes… here’s Perplexity’s answer when asked for negative quotes about Coppola classics. Yep - they’re made up. And Perplexity is advertised as a search engine. So I suspect people trust it more.

@dannysullivan @MLE_Online Searching for programming docs used to work well; now it's all SEO spam. E.g. try "python list remove". Result: spam, "people also search for", spam, spam, "people also ask", spam, spam, spam, spam, images, related searches. docs.python.org wasn't on first page at all.

@MLE_Online It’s really helpful to get actual examples of a search that disappointed, if people are comfortable sharing. Helps us review & understand how we might make general improvements to search overall.

@_AlecJacobson @DerekRenderling Ok, looks like the main contributions are practical? (Robust SDF conversion from possibly nonmanifold mesh, shader implementation for rendering, measurements on a standard 3D dataset.) Now I'm more curious how this would perform with a periodic activation fn.

@_AlecJacobson @DerekRenderling arxiv.org/abs/2006.09661 compares ReLU networks with sin activation for this this exact technique. Have you tried sin instead of ReLU? What's new in this paper? (Seems like mostly you go deeper on reconstruction quality vs. cost).

Purposefully overfit neural networks are an efficient surface representation for solid 3D shapes In arxiv.org/abs/2009.09808 with Thomas Davies, @DerekRenderling, we make a few observations:

Oh codeproof.com, you so crazy, pointing people here.

Star wars Xmas apecial #lizmas

Spirit of Xmas #lizmas

South park #lizmas

Muppet Xmas #lizmas

Glitter is the new black.

It's about bloody time someone put cheerleaders in a FPS.

Argh

I do not like TFS.

I like cheese