Inbibo has released Adonis 2.1, the latest version of its digital anatomy framework for Autodesk Maya and SideFX Houdini. The update introduces a new product split, adds machine-learning tools for faster deformation feedback, extends the anatomy transfer workflow and introduces a new deformer for cleaner simulation output. Because apparently muscle, fat, skin and fascia simulation still left too much spare time in the day.
From version 2.1, Adonis is split into two products. AdonisFX remains the core digital anatomy toolset for muscle, fat and skin simulation in Maya and Houdini. AdonisML is the new higher-end version and includes everything in AdonisFX, plus two machine-learning systems: ML Deformer and SmartTissue.
The overall idea is simple enough, at least by the deranged standards of character effects: keep the high-quality anatomical setup, but give animators a faster, more interactive representation of the final character while they work.
ML Deformer for interactive character feedback
The new ML Deformer is trained on AdonisFX characters and creates a pose-based machine-learning approximation of the deformation produced by the heavier tissue setup. Inbibo describes the system as a custom neural network training pipeline.
SmartTissue brings back the secondary motion
The limitation of pose-based ML deformation is that it usually loses dynamic information. The trained result may reproduce the shape well, but it does not automatically preserve the jiggle, settle and overshoot that come from simulating fascia and fat. That missing motion is often exactly what makes a creature or character feel less like a rig and more like a living, slightly expensive problem.
SmartTissue is Inbibo’s answer to that limitation. The system uses a neural network architecture to predict how the underlying muscles are activating for the current pose, then uses that prediction to drive a faster dynamics layer. The goal is to restore the kind of secondary motion that would otherwise require a full anatomy simulation.
According to Inbibo, SmartTissue can also be cached, so animators can generate playblasts with the dynamics already present. In many cases, the company says the cached result is good enough for shots with multiple simulated creatures to move on to lighting or client review within minutes. That does not remove the need for final validation, because pipelines enjoy punishing optimism, but it should make iteration less glacial.
Anatomy transfer without matching topology
AdonisFX 2.1 also gets a topology-independent anatomy transfer workflow. The new AdnTransfer system is designed to move anatomy geometry from one character or creature to another by mapping anatomical correspondence between the source and target. The important part is that the source and target do not need to share topology or identical proportions.
That is useful because many vertebrates share broadly comparable anatomical structures, even when the mesh, scale and proportions differ heavily. A transfer workflow like this can reduce the amount of manual rebuilding needed when moving muscle and tissue setups between related characters or creatures.
Two new deformers sit at the centre of the workflow. AdnRigidWrap attaches a mesh to one or more target meshes, locking each point to the closest location on the targets while preserving its original offset. AdnSoftWrap handles the softer regions between those attachments by blending nearby influences within a defined radius, creating continuous soft deformation across the mesh. A new Landmark Tool provides a visual interface for placing and configuring correspondence pairs directly on the meshes.
AdnMush smooths without killing detail
The other major addition is AdnMush, a new deformer for reducing skinning and post-simulation artefacts while preserving anatomical surface detail. Inbibo specifically mentions details such as ribs and pectoral muscle striations, which is the sort of fine structure that a generic smoothing pass can quickly turn into athletic porridge.
In Adonis 2.1, AdnMush runs inside every solver as part of the solve itself, so the output should require less post-processing. It also remains available as a standalone post-process node for shot-level correction when artists need to clean up a result without rerunning the full simulation. For users of the Turbo auto-setup workflow, generated rigs now use AdnMush for artefact correction, reducing the amount of cleanup needed before a setup is ready for initial supervisor feedback.
Pricing and compatibility
Adonis 2.1 is available now. Inbibo lists AdonisFX Pro licences as starting from £1,500 per year, with Indie licensing also available. AdonisML Pro licences start from £3,000 per year and include AdonisFX plus ML Deformer, SmartTissue and the related machine-learning tools. AdonisFX has a free 30-day trial, while evaluation access for AdonisML requires contacting Inbibo directly.
Adonis 2.1 is compatible with Maya 2023 and later, and Houdini 20 and later. Supported operating systems are listed as Windows 10 and later, CentOS 7 and later, RHEL 8 and later, and Rocky Linux 8 and later. AdonisFX Indie subscriptions cost at £25 per month or £250 per year per node-locked licence for artists earning under £100,000 per year. Pro subscriptions are at £110 per month or £1,100 per year for node-locked licences, and £150 per month or £1,500 per year for floating licences.
Download / trial // Inbibo Downloads
