The New Shape of 3D VFX Pipelines
The latest 3D VFX software updates show a clear trend: machine learning simulation and modern GPU workflows are moving from research experiments into daily production tools, promising faster iterations, more interactive character rigs, and smarter 3D compositing that cuts down on manual fix‑ups and technical overhead for professional VFX artists. Inbibo’s Adonis 2.1 and Foundry’s Nuke 17.1 beta are not minor point releases; they are strategic bets on automation and real‑time visualization. By putting ML‑driven soft tissue rigs inside Maya and Houdini, and by expanding Gaussian splats and Hydra 2.0 in compositing, these releases aim to shorten the distance between simulation, layout, and final pixels. The message is blunt: if your pipeline still relies on heavy, slow solvers and flattened renders, you are about to feel outdated.
Adonis 2.1: ML-Powered Soft Tissue, Less Pain for Rigging
Inbibo has released Adonis 2.1, upgrading its modular anatomy framework for Maya and Houdini with a clear goal: replace slow, hand‑tuned soft tissue setups with machine learning‑accelerated rigs. The new high‑end AdonisML edition introduces an ML Deformer and Smart Tissue system for training lightweight approximations of complex, slow‑to‑process tissue rigs. The machine‑learned deformer represents the same mesh deformations but processes faster, so animators can pose and iterate in near‑interactive time instead of waiting for heavy simulations to finish. Smart Tissue then restores the lost secondary motion—the jiggle, settle, and overshoot—so high‑fidelity results survive the speed‑up. According to Inbibo, the combination provides an “ML-powered path for animators that need high-fidelity results at interactive speed.” This is exactly what the market has lacked since the popular Ziva VFX left commercial availability: a production‑ready, layered muscle–fat–skin solver that does not grind artists to a halt.
Adonis 2.1 is not just about ML. The AdnTransfer anatomy transfer system lets teams move tissue setups between characters or creatures with very different proportions, using AdnRigidWrap to lock landmark points and AdnSoftWrap to handle the soft tissue in between. That makes reusing high‑quality anatomy rigs across assets more realistic, rather than rebuilding them from scratch every time. A Landmark Tool lets riggers refine the mapping when automation is not quite enough. On top of that, the AdnMush deformer smooths simulation artifacts while preserving fine anatomical detail—especially useful for defined or “shredded” characters—and it runs automatically inside every solver to reduce post‑processing in the first place. Indie subscriptions for AdonisFX start at £25/month (around $33/month, approx. RM153) or £250/year per node‑locked license, while Pro subscriptions start at £110/month or £1,100/year, and £150/month or £1,500/year for floating licenses. AdonisML is priced on enquiry, with Pro subscriptions starting at £3,000/year.
Nuke 17.1: Gaussian Splats Grow Up
Foundry’s Nuke 17.1 open beta shows how compositing is catching up to the new wave of 3D capture and rendering: 3D Gaussian splats are no longer a novelty, they are becoming a practical environment tool. Nuke 17.1 extends the major 3D Gaussian Splatting toolset introduced in 17.0, now letting artists import, manipulate, render, and animate splat sequences as part of regular 3D workflows. The new GeoSequencer node can process sequences of USD, PLY, or SPLAT files and output time‑sampled USDC that Nuke’s existing nodes can handle. More importantly, the updated SplatRender node supports 2D rendering of Direct, Point, and Spot lights placed in the 3D system, enabling “basic relighting” of splats. That moves splats from static set‑extension plates toward lighting‑responsive assets that can sit more convincingly alongside CG, without round‑tripping everything back to a DCC for changes. Nuke now also exports SPLAT files in addition to PLY, USD, and ABC, which matters for studios experimenting with splat‑based capture.
The practical impact is obvious: Nuke artists can iterate on scanned environments, tweak lighting, and test compositing decisions without constantly bouncing between departments. When combined with the new non‑destructive USD export workflows—explicit sublayers and overs (sparse overrides) instead of flattened scenes—teams can share changes between studios or departments with less data loss and fewer heavy exports. GeoExport now defaults to exporting frames individually and stitching them into single files, cutting memory usage for flattened exports. Annual subscriptions cost $3,839/year (approx. RM17,810) for Nuke, $5,219/year (approx. RM24,220) for NukeX, $6,379/year (approx. RM29,620) for Nuke Studio, and $462/year (approx. RM2,145) for Nuke Render licenses, which have gone up by $22/year (approx. RM102). Nuke Indie, limited for smaller commercial users, is $499/year (approx. RM2,315). These prices make clear that Foundry expects serious studios to treat splats and USD as standard production tools, not side experiments.
Hydra 2.0 and Real-Time Visualization: Compositors Want Speed Too
While Adonis focuses on character simulation, Nuke 17.1 quietly upgrades how compositors look at 3D: the Hydra‑based 3D viewer now supports Hydra 2.0, the current USD rendering framework. That sounds like plumbing, but it matters. Hydra 2.0 opens up new options for displaying 3D scenes when using the HdStorm render delegate, including previewing individual AOVs such as depth and position directly in the viewer. For artists, that means faster diagnosis of compositing issues—matte edges, parallax problems, depth mixes—without sending shots back to lighting. It also “paves the way” for support for third‑party render delegates in future, which is a polite way of saying the viewer is being positioned for whatever GPU‑friendly rendering stacks studios may adopt next. Combined with new cutout controls in ScanlineRender and several shaders, Nuke’s 3D view feels less like a bolt‑on and more like a serious part of the pipeline.
Nuke Studio’s timeline‑oriented updates round out this focus on speed. The Annotations toolset now includes rectangles and arrows, with color, width, and opacity controls tuned for mouse and trackpad users—small, but telling, because it pushes review directly into the timeline. The Contact Sheet shifts to per‑sequence operation, letting artists load all visible tracks with one click and switch between grid view and a single full‑frame cell. A new Compare Versions feature brings A/B comparisons into the timeline, using the same comparison modes as the viewer. It is now possible to burn in custom metadata and expressions as soft effects. None of this is flashy, but combined they cut friction in day‑to‑day compositing, where dozens of tiny interactions often cost more time than any big render. Foundry has released Nuke 17.1 in open beta, with a stable release expected in roughly three to six months based on past update cycles.
Conclusion: Faster Iterations, Fewer Heroics
Taken together, the Adonis 2.1 update and Nuke 17.1 features mark a practical shift in 3D VFX software: pipelines are being built for faster iteration, not heroic last‑minute saves. AdonisML’s machine learning simulation systems aim to automate complex soft tissue workflows in Maya and Houdini, giving animators interactive rigs that still respect anatomical detail. Nuke’s Gaussian splat relighting and Hydra 2.0 viewer support push compositing closer to real‑time 3D visualization, reducing the need to flatten scenes and improving round‑tripping between departments or studios. The impact on ordinary users is clear: fewer hours spent waiting on solvers and exports, more time thinking about performance, framing, and story. The risk is that teams who ignore these shifts will end up locked into slower, more manual pipelines while competitors iterate twice as fast. The opportunity is to treat ML and modern GPU workflows not as buzzwords, but as tools to remove drudgery from simulation and compositing—and that is a trade most VFX artists are ready to make.
