Adonis 2.0 houdini question about gravity and double solver with blendshape

I’ve been using adonis 2.0 since launch and it is doing great, also saw the hyena scene, it was good to see some of you guys’ tricks to the setup, with that I have 2 questions:

1- Why is it simulated without gravity? I can see that it jiggles more (that is good) just want the devs input on that haha (feels weird not to have gravity)

I was testing some scripts to do the network in a more automated way, they worked fine but I came up with a setup that I want to know If there will be any issue from your perspective:

2- lets sim the forearm (bunch of small muscles) I simmed them relaxed in one adn solver and tensioned on another solver, them I’m blending them using a blendshape node (with the tension att to blend), there is any problem from the solver perspective? visually I think it’s fine, but just want to ask if there is any catch on that.

Hey, thanks a lot for the kind words, really glad to hear Adonis 2.0 has been working well for you, and also great to know you enjoyed the hyena scene.

About your first question: simulating without gravity is mostly a simple and artistic choice. AdonisFX is intended to allow for fast setup, and while physically plausible results are desirable in some cases, in others what matters more is getting appealing motion quickly. In some setups, gravity is not really necessary to achieve that. Also, depending on how the muscles are modeled, it could be that gravity is already kind of “baked into” their shape, so it really depends on the use case.

For the second topic, I just want to make sure I understand your setup correctly. Are those small forearm muscles all combined and solved in a single AdnMuscle solver? Is the reason for this setup that you wanted to avoid configuring each muscle individually? Or are you driving the blend using a custom tension attribute (per point maybe)? With a bit more context there, I can give you a better answer about whether there’s any catch from the solver side.

Also worth mentioning that we now have AdnTurbo, which allows you to configure a first pass of the asset in a very short amount of time, which can help speed up this kind of workflow even more.

Also, if you feel like sharing your results, we would be really happy to take a look :slightly_smiling_face:

ooo Ok now the gravity thing makes sense haha.

Yes I’m putting all muscles in one solver, them I have 2 of them:
first relaxed (it doesn’t contract) and the second contracted them I blend between both results using a blendshape att on points.

as soon as I can share some results I will definetly post them in the forum :slight_smile:

Hi! Thanks for sharing your setup, interesting approach :slightly_smiling_face:

AdonisFX AdnMuscle nodes are generally intended to simulate one muscle at a time, rather than multiple muscles within a single solver. Unlike systems like Vellum or Otis, the solver isn’t designed to ingest and handle multiple interacting muscles simultaneously.

To achieve muscle-to-muscle interaction, the typical workflow is to simulate muscles individually and then use AdnGlue as a secondary pass.

What you’re proposing can work, but it’s not the intended workflow for AdonisFX. A few things to keep in mind:

  • The core is designed to simulate a single closed geometry (not necessarily watertight), and handling multiple disconnected pieces can lead to incorrect fiber initialization.

  • Volume preservation may become inaccurate, since the solver isn’t built for that type of interaction.

  • Your setup requires simulating the full muscle system twice, which can significantly impact performance.

  • Blending between relaxed and contracted states may not produce the same results as driving activation through sensors.

  • This approach may also limit cross-DCC compatibility.

That said, we’re definitely curious to see your results. It’s always great to explore alternative workflows. If it’s working well for your case, that’s valuable insight.

Out of curiosity, how are you computing the tension attribute?

Looking forward to seeing what you come up with!

Awesome thanks for the infor, as soon as I can show some results I will definetly post.
I’m computing the tension using the new houdini 21 auto tension lines. I’m only doing the double solver with blendshape workflow on smaller tight muscles like fore arm and the under leg, also with the blendshape workflow it’s possible to do a “franken muscle like approach” by merging all of them and blending the contraction using the tension as blendshape att.

1 Like