Brand new video up on the Future Familiar YT channel.
When I tested AI 3D generators two years ago my takeaway was we were nowhere close to anything usable.
Since then every few months something would pop up on my feed claiming to be the next breakthrough but I just kept scrolling assuming the updates were minor.
But with the start of a new year it felt like the right time to finally check back in and see how far things have actually come.
So I went through and tested all of the top mesh generation platforms side by side including Meshy, Tripo, Rodin, and the new entries from studios like Tencent and NCSoft. The results were surprising.
Admittedly, it was a tough one to put together (and expensive) but I think it's a useful snapshot for anyone in the 3D or gamedev space who's been tuning this stuff out the same way I was. More on this down below.
But first here's the weekly roundup of tech and tool updates that are shaping the future of worldbuilding, gamedev, and creative work.
NEW TECH UPDATES
1. NCSoft Launches Varco 3D AI
NCSoft's AI division (the Guild Wars and Lineage studio) launched Varco 3D over the holidays. Like many similar tools it offers the ability to input text or an image and get a 3D asset out with the full pipeline handled in the browser from concept ideation to model generation, remeshing, and even basic animation all in one place.
This puts them in direct competition with Meshy and Tripo for rapid prototyping but with a AAA game studio's production pipeline behind (similar to Hunyuan 3D with Tencent) rather than a pure AI startup.
Takeaway: In my testing for this week’s video it seems many of these platforms try to be all-in-one’s catering to 3D printing or just simply AI enthusiasts looking to play around in 3D but this one's interesting because it's a major game publisher putting their internal production AI into public hands.
Check out the Varco 3D platform.
2. Ramen VR Launches UE5 AI assistant “Aura”
Ramen VR launched Aura on January 2nd as an AI coding assistant built specifically for Unreal Engine that is designed to understand your project's codebase and work with Blueprints and C++ in context. Their goal is to get you out of Claude Code or Cursor and into an environment actually trained on Unreal workflows rather than general software development.
They published a case study alongside the launch where Sinn Studio (the VR team behind Swordsman) used Aura during development and shipped Zombonks in early access in five months. According to the studio their asset sourcing pipeline went from 2-3 days per level down to 4-5 hours.
Access is invite-only right now with a two-week free trial. Your assets stay local and they claim no training on your work without permission.
"Aura and I rapidly iterated to meticulously craft a vision for the game. We make VR games, so coming up with an idea and then seeing it come to life in 3D just a few minutes later was game-changing for me."
Check out the Aura official release announcement.
3. Hy-Motion 1.0 Releases
Tencent open-sourced HY-Motion 1.0, a text-to-motion model that generates skeleton-based 3D character animations from text prompts. The output is SMPL-H format with FBX export, so the results drop straight into Blender, Unity, or Unreal.
The model was trained on over 3,000 hours of motion data across 200+ categories. There's a full 1B parameter version and a lighter 0.46B version for lower-spec machines, and both run locally on Mac, Windows, and Linux (if you happen to have 26+GB of Vram).
Takeaway: The likely application is blocking out character movement during previs or prototyping NPC behavior before committing to mocap. Current limitations are the usual text-to-motion gaps: no seamless loops, and anything involving precise object interaction (sitting, grabbing, opening doors) tends to clip or float. But for roughing in combat sequences or movement passes, this is free and runs locally (again, assuming the Vram requirements).
Check out the project on Github. Or check out the ComfyUI Plugin.
4. Polygonflow 1.9.3 adds 1-click assets from Sierra Division and Dekogon Studios inside Unreal 5
Polygonflow released Dash 1.9.3 and the headline feature is single-asset purchasing from third-party creators directly inside the plugin. Dekogon is the first seller with their full library available and Sierra Division has added a free asset pack ahead of joining as a seller in a future update.
The workflow is straightforward, you just search for what you need right in the viewport, buy/download the individual asset and it's in your scene within seconds. Never thought I’d find micro-transactions in worldbuilding but frankly the workflow feels pretty sweet.
Especially given you can opt to buy individual assets when you just need a single rock or prop for example, as opposed to entire packs that then need sifting through and importing the one or two you actually needed.
Takeaway: Dash has been my favorite worldbuilding plugin for Unreal for quite some time (honestly might be my favorite worldbuilding tool period), and this update pushes it further that direction.
Having quality asset libraries like Dekogon and Sierra Division available at the individual asset level inside the same tool I’m already using for scattering, material blending, and procedural placement removes a lot of the friction that normally fragments a worldbuilding workflow.
Check out Polygonflow Dash.
NEW VIDEO RELEASED
The Best 3D AI Generators in 2026
Two years ago I tested every AI 3D mesh generator I could find.
And they were all garbage.
Even their demo landing page cherry picked examples were… pretty rough when you took a close look at them. So I largely wrote them off at the time and assumed we were still incredibly far off still.
But I blinked, 2025 came and went, and it turns out there are now dozens of new tools with huge new updates shipping almost weekly.
And so over the holidays I spent hundreds of dollars testing all of the best ones currently out there, and honestly, I found the results pretty surprising and legitimately useable (depending on your goals and what it is you’re trying to achieve.)
So, if you haven’t had time to research and personally test every single one of these tools yourself, then this video is designed to catch you up to speed on where we are at going into 2026.
Up next week
I’ve been seeing more and more world model updates on X and Linkedin. Might FINALLY dive into those unless there are other major updates that pop up before next weekend.
If there's something I missed this week, just hit reply to let me know!
Until next week,
Chet - Future Familiar