#Trellis2
I made a brain.

My first non-human 3D model through Trellis2. A few holes, but nothing to worry about.

Next I'm going to make it explode 🤯

solvx.uk/go/3d-bs

#buildinpublic #3D #blender #gamedev #trellis2 #3dart #b3d
September 15, 2026 at 8:49 PM
Back to faces, and this one is hard.

Last time I just needed a face. This time it has to match a specific image, and Trellis2 is not good at faces.

Splitting the head into its own mesh failed. So: sweep the seeds, pick the best, run again.

#buildinpublic #3D #gamedev #blender
September 20, 2026 at 3:00 PM
Trellis2, we love you.

Sub 2k tris. Complete models — one piece, no loose parts. Running locally on an 8GB card, for free.

Animation pipeline coming soon.

#buildinpublic #3D #blender #gamedev #lowpoly
September 9, 2026 at 6:32 PM
Best result yet.

Model generated in Meshy.ai, rigged locally with AccuRig, re-textured with Claude Code — and now he walks.

Yesterday nine Trellis2 faces fell apart under a close-up. Different pipeline, completely different outcome.

#buildinpublic #3D #AI #gamedev #blender
September 2, 2026 at 8:42 PM
Nine models. One source image. All generated locally with Trellis2, no cloud, no credits.

Some of them are genuinely good. Together, T-posed in a row, they are the world's worst digital boy band.

Auditions are closed.

#buildinpublic #3D #AI #gamedev
September 1, 2026 at 4:49 PM
Yesterday I lined up nine Trellis2 models and called them a boy band. Then I zoomed in on the faces.

Only 2 of the 9 survive a close-up. The rest are less boy band, more horror show.

That run only varied the seed. Today I sweep the actual settings.

#buildinpublic #3D #AI #gamedev
September 2, 2026 at 6:02 PM
- a background manifold with input videos projected onto it via a conformal mapping (joukowski here, from memory)
- a 3d mesh model of the pleopenguin generated with trellis2 (tripo3d etc cheapest paid variants are fine for simple geometries like this), pleorama projected onto this geometry too
May 7, 2026 at 3:52 AM
Ran "Measuring Pixal3D on ComfyUI 0.35" via Colab A100. Eight runs took around 28 minutes total. Dropped the resulting GLB outputs in a gallery.

https://yosuke4061.com/new_toppage/briefings/brief150_en/
Measuring Pixal3D on ComfyUI 0.35
Measured Pixal3D (Switch to Trellis2 = false) on ComfyUI 0.35 via Colab A100 and a GCP reverse tunnel. Eight white-background inputs, GLB gallery, Per
yosuke4061.com
September 19, 2026 at 4:01 PM
Microsoftが1枚の画像から複雑かつシャープなトポロジー・PBRマテリアルを備えた高品質な3Dアセットを生成する技術をオープンソース公開!

TRELLIS 2
3dnchu.com/archives/tre...

#TRELLIS2
December 19, 2025 at 3:09 AM
ComfyUI-TRELLIS2
ComfyUI custom nodes for TRELLIS.2 - Microsoft's state-of-the-art image-to-3D generation model.
https://github.com/PozzettiAndrea/ComfyUI-TRELLIS2?utm_source=wikitd.com&ref=wikitd.com
February 26, 2026 at 8:37 AM
Trying trellis2-mlx to run Trellis.2 on Apple Silicon — local 3D model generation on a MacBook is dreamy! I made an SF-style companion robot; quality's surprisingly high. Some mesh holes and rough unseen parts remain, but it’s tweakable. So fun! #3D
July 2, 2026 at 1:06 AM
Running TRELLIS.2 on my Mac showed a night-and-day difference in meshing! 🤯

Inference with `trellis2-mlx` was smooth, but meshing left holes, ran slow & was unstable.

So I compared 5 meshing methods using the same voxel data—the difference in results is absolutely insane!!
July 9, 2026 at 7:06 AM
August 12, 2026 at 11:42 PM
July 2, 2026 at 1:13 AM
April 17, 2026 at 11:45 PM
February 25, 2026 at 1:30 AM
だから「誰でも3秒で革命」より、「かなり強い研究成果が、まだ重い机の上で動いてる」が近いかも。速くなったのは本当。でも、作業台まで消えたわけじゃない。 #AI検証 #3D生成 #TRELLIS2
June 27, 2026 at 5:14 AM
「Blender内で完結」と書かれてるけど、ComfyUIは別プロセスで動いてて手動起動が必要💎

画面が一つに見えてるだけで、裏では二つのアプリが走ってる🫧

「Open」という名前だけど、アドオン自体のGitHubは見つからなかった。中のFLUX.2-Klein、LTX 2.3、Trellis2、HY-Motion——全部、別のチームが作ったオープンソースモデル💎
June 2, 2026 at 3:29 AM
→ Trellis 2 - Image To 3D
Image to 3D Model Generator

#ar/vr #product

https://trellis2.app/
April 9, 2026 at 1:28 AM
🎨 Being an Artist 16 - "Wonderland"

Nothing like a one hour moment with the best arty friends to bang out a piece! Feels so good.

Technique had ComfyUI text to img to Trellis2 models, then lit and laid out in 3D, then the effects using compositor techniques in Bforartists.
February 21, 2026 at 5:41 AM
✏️ Drawing practice always goes so left field...

Technique: ComfyUI Trellis2 models to Blender/Bforartists with Grease Pencil (3d brush drawing) with the inkguashy look I did last Friday.
February 4, 2026 at 3:32 AM
Trellis.2: Turn one image or text into PBR-ready 3D in seconds.

Image-to-3D.
Text-to-3D.
PBR materials.
Open source.
Free commercial use.
Engine-ready (GLB/FBX).

Perfect for games, e-commerce AR, XR, product design.

#Trellis2 #ImageTo3D #TextTo3D

Read more:

aiadoptionagency.com/trellis-2-im...
Trellis.2: Image to 3D Guide - Ai Adoption Agency
Trellis.2 is an advanced open-source image-to-3D and text-to-3D model from Microsoft that turns simple images or prompts into high-fidelity, PBR-ready 3D assets within seconds, dramatically compressin...
https://aiadoptionagency.com/trellis-2-image-to-3d-guide/"
December 22, 2025 at 8:19 PM
Trellis2.com is a practical image-to-3D workbench for TRELLIS.2, helping you generate, preview, and validate GLB models before importing them into Blender, Unity, or Unreal.
January 1, 2026 at 12:05 PM
「Blender内で画像・動画・3D生成が完結する『革命』の神ツール」と話題のOpenBlenderを検証した。

紹介: https://x.com/ai_hakase_/status/2025542347195285565
公式: https://pgcrt.github.io/

ツール自体は実在する。LTX 2.3、FLUX.2-Klein、Z-Image Turbo、Trellis2、OpenRouter経由チャット——10モジュールをBlender内から呼べる統合UIだ。💎
May 15, 2026 at 1:05 PM
What I'm watching this week: OpenClaw, Trellis.2, Gemma 4, and two more

Five open-source AI releases worth bookmarking from the week of May 26–30 2026, annotated for builders running lean AI stacks.

https://dev.to/morinaga/what-im-watching-this-week-openclaw-trellis2-gemma-4-and-two-more-p70
June 27, 2026 at 10:11 PM