[BlackLotus89]

Everyone in the comments until now seems to focus on the simulation properties for true to life simulations for ads and movies. Isn't this way more interesting for culinary advancements? I for one am way more interested if we could tweak ingredient ratios and cooking methods for the optimal result. Fluffy yet cross pizza dough with leoparding, how much sourdough/yeast is really needed for my bread and how can I estimate the rise based on the pre-rise, how can I optimize for flour x, ...

Until now I tried to measure exactly and get every parameter right and document everything correctly, with this I could just simulate everything if I get decent measurements for my parameter once.

[jey]

It's not meant to be a simulation that's actually accurate. It's instead meant to "look accurate". The full chemistry is not being simulated.

[Doxin]

> It's not meant to be a simulation that's actually accurate.

Most computer simulations aren't. That doesn't mean they aren't accurate enough for some real world applications anyways. Take the fluid simulation in blender for example: It was designed for visual effects, not for sea defence simulations. It works really quite nicely for both however.

[hatsunearu]

You sure? It sounds like it could be useful for capturing a variety of thermodynamic phenomena provided that you capture the chemistry properly (the stuff about changing elasticity and material properties is really a simplified model of the chemistry, i think)

[hyperbovine]

I’m surprised they get realistic browning. In real life that’s caused by the Maillard reaction. They did not mention this in the video. I wonder it they specifically modeled this process, or if the behavior emerges from their other chemical models.

[jey]

The paper says that "coloring is based on temperature". I think what they mean is that the coloring of each point is determined by the peak temperature that has been reached by that point. (Which seems perfectly appropriate for the intended application to visual effects.)

Source: https://www.math.ucla.edu/~myding/papers/baking_paper_final....

[RomanBob]

those things are figured out centuries ago via iterated optimisation.