sakalond

让Agent操控blender建模之前一定要装这个插件！ StableGen把生成模型，模型贴图，甚至自动绑骨串成了一整套流程， 最后生成的模型还能继续编辑，导入动作，最后还可以导出到游戏引擎中使用。 重点是它完全开源免费。 GitHub:sakalond/StableGen t.co

ComfyUIでの AIポリゴンリダクション（リメッシング）だそうです。 AIがポリゴン削減する時代になりました。(^_^) sakalond/ComfyUI-LATO2

@Rotinda_Dag If you're using Fit camera resolution for texturing lowering Target megapixels is one of StableGen's low-VRAM suggestions github.com/sakalond/StableGen/wiki/Troubleshooting

textured an entire street in Blender with Qwen Image 2.1, then made cyberpunk, anime and papercraft versions detailed post: reddit.com GitHub: sakalond/StableGen Guide: github.com/sakalond/StableGen/wiki t.co

⭐ StableGen ⭐ sakalond/StableGen 1. 블렌더 안에서 AI로 3D 모델을 만들고, 텍스처를 입히고, 움직일 뼈대(리그)까지 넣어주는 오픈소스 애드온 StableGen이 나옴 2. 모델 둘레에 카메라를 놓으면 각 방향 그림을 AI가 그려서 표면에 입혀줌. 앞서 그린 그림을 참고해서 다음 방향을 그리니까 한 바퀴 돌아도 텍스처가 이어짐 3. 캐릭터를 글로 설명하고 미리보기 이미지 하나만 고르면 TRELLIS.2가 모델을 만들고 SkinTokens가 뼈대를 넣고 텍스처까지 한 번에 끝냄. 이미 있는 모델에 텍스처만 입히기도 가능하고, 칠한 부분만 골라 다시 생성도 가능 4. PBR 재질 맵 생성, Mixamo 모션 입히기, 게임 엔진용 내보내기(Godot·웹은 glTF, 언리얼·유니티는 FBX), 3D 프린트용 3MF까지 가능 5. GPL-3.0 무료, 스타 1,239개. Blender 4.2~4.5 또는 5.1 이상(5.0은 미지원)에 ComfyUI 서버 필요. 그래픽 메모리는 SDXL 약 8GB, Qwen Image 16GB 이상. Windows·Linux·Apple Silicon 맥 가능 게임용 캐릭터를 블렌더에서 빠르게 뽑아 보고 싶은 NVIDIA GPU 쓰는 사람한테 좋을 듯 ━━━━━━━━━━ 🇺🇸 1. StableGen is an open-source Blender add-on that generates 3D models, textures them, and even adds a rig (skeleton) for animation, all inside Blender 2. Place cameras around a model and the AI draws a view for each angle and projects it onto the surface. Each view uses the earlier ones as reference, so the texture stays consistent all the way round 3. Describe a character in text and pick one preview image: TRELLIS.2 builds the mesh, SkinTokens adds the skeleton, and the texture is done in one run. It can also texture models you already have, and regenerate only the part you paint over 4. Also makes PBR material maps, retargets Mixamo animations, and exports for game engines (glTF for Godot and the web, FBX for Unreal and Unity), plus 3MF for 3D printing 5. Free under GPL-3.0, 1,239 stars. Needs Blender 4.2-4.5 or 5.1+ (5.0 not supported) and a ComfyUI server. About 8 GB of VRAM for SDXL, 16 GB or more for Qwen Image. Works on Windows, Linux, and Apple Silicon Macs Looks great for anyone with an NVIDIA GPU who wants to quickly spin up game characters inside Blender #AI #Blender #3D #OpenSource #GameDev

Generated the props and the character with TRELLIS.2, then projection-textured the whole street with Qwen Image 2.1 (Blender, local models) Every texture on this street, and the 3D assets in it, were generated with local models in Blender, using StableGen, my free and open-source add-on. The process: Meshes (TRELLIS.2) - The props (car, bench, lamp, hydrant, bin, crates, bush) each started as a text prompt → a Qwen Image 2.1 reference image → TRELLIS.2 at 1536 (cascade). The character came from a concept image (the best of a few seeds). The buildings, road and sidewalk are simple blocks. - Decimated on the server to 200k–500k faces each (car 500k, character 400k, bench 300k, lamp and bush 250k, the rest 200k), with QDC remesh on. With remesh off, the decimated meshes had flipped and broken triangles. - Tip: keep damage words (rust, moss, cracks, broken) out of the mesh prompt. TRELLIS.2 turns them into dents and noise in the geometry. Ask for clean, intact shapes and put the wear in the texture prompt. - These are dense triangle meshes, fine for renders like this; they are not retopologized or game-ready. Texturing (Qwen Image 2.1, int8) - 10 cameras around the scene. Each view is generated from that camera's geometry render, then projected and blended onto all the meshes at once. - Each camera's prompt lists only the objects it actually sees. With one prompt for the whole scene, views painted objects onto surfaces where they aren't. - The character was textured on her own with her own cameras; everything was baked to 8K textures (the photoreal look with Marigold V2 PBR maps) and rendered in Cycles. - The logo is a Masked Edit: paint a region, describe the change, and only that region is regenerated through the cameras that see it. It follows a reference image of the logo, which is on the development branch and coming in the next release. - Each look is a separate texturing run with its own prompt. Generated on rented RTX 3090/4090 cloud GPUs and my RTX 4080 (16 GB). Qwen Image 2.1 is under a non-commercial license. GitHub: sakalond/StableGen Guide: github.com/sakalond/StableGen/wiki

I textured an entire street in Blender with Qwen Image 2.1, then made cyberpunk, anime and papercraft versions of it Every texture on this street corner was generated with local models and projected onto the geometry in Blender, using StableGen, my free and open-source add-on. How it was made: - The 3D assets (the car, the woman, the bench, lamp, hydrant, bin, crates and bush) are TRELLIS.2 meshes, each from a text prompt via a Qwen Image 2.1 reference image. The buildings, road and sidewalk are simple blocks. - Texturing: Qwen Image 2.1 (int8), 10 cameras. Each view is generated from a depth/geometry render of that camera, then projected and blended onto all the meshes at once. Each camera also got a prompt listing only the objects it actually sees, which stopped views from painting things that aren't there. The woman was textured separately with her own cameras. - The logo is a Masked Edit: paint a region, describe the change, and only that region is regenerated through the cameras that see it. Here it follows a reference image of the logo, which is on the development branch and coming in the next release. A few more Masked Edits fixed problem spots. - The looks are separate texturing runs with different prompts (the photoreal one with Marigold V2 PBR maps), baked and rendered in Cycles. Lighting is set up per look in Blender. Generated on rented RTX 3090/4090 cloud GPUs and my RTX 4080 (16 GB). Qwen Image 2.1 is under a non-commercial license; check each model's license before commercial use. GitHub: sakalond/StableGen Guide: github.com/sakalond/StableGen/wiki

I made a ComfyUI node for LATO.2: turn dense TRELLIS.2 / Hunyuan3D meshes into usable low-poly meshes Image-to-3D models like TRELLIS.2 and Hunyuan3D give you great shapes, but as 500k–1M triangles of soup that you can't use in a game or animate. LATO.2 (HUST & Meshy, MIT license) is a new model that generates a new low-poly mesh for a shape: you ask for a vertex count (200–5000), it places the vertices and predicts how they connect. It's not a decimator, so the edges mostly follow the form. It's not hand-modelled quality either: think fairly clean, not artist-clean. ComfyUI-LATO2 wraps it as a ComfyUI node and makes it usable out of the box: - Works with ComfyUI's native 3D nodes: plug in a MESH straight from native TRELLIS.2 or Hunyuan3D, or a 3D file from Load 3D; get a MESH for Save 3D Model and a 3D file for Preview 3D. Or just pick/upload a .glb/.obj. - Closed surface: LATO.2's raw output stacks overlapping faces, leaves holes and flips faces at random (last image). The node rebuilds the faces from LATO.2's edge predictions into one closed surface with the faces pointing outward. - Inner walls removed first: TRELLIS.2 meshes have a hidden second copy of their surface inside; a voxel flood fill strips it so LATO.2 doesn't mesh both walls. - Same position and scale as the input, so it drops in as a replacement. - Windows and Linux, nothing to compile: LATO.2 normally needs CUDA extensions (o_voxel, flash-attn, ...) that have no Windows builds. The node ships pure PyTorch/NumPy stand-ins (the voxelizer gives identical voxels to the original, checked on millions of voxels), and a one-command installer sets up its own environment so your ComfyUI Python stays untouched. Install: git clone https://github.com/sakalond/ComfyUI-LATO2 into ComfyUI/custom_nodes, then run python install_lato2.py once in the node's folder (about 12 GB: environment + weights) and restart. It's also submitted to the Comfy Registry as comfyui-lato2, but that listing is still pending review, so ComfyUI Manager may not find it yet. The LATO.2 Setup Check node tells you if anything is missing. Four example workflows are included (Browse Templates → Extensions → ComfyUI-LATO2), including one that plugs into ComfyUI's own TRELLIS.2 template. Needs: NVIDIA GPU with 8 GB VRAM (peaks around 7 GB). A run takes 30–60 s on a 3090/4080. Caveats: - Thin parts like windmill blades or slats can come out badly, so try 2–3 seeds. - Geometry only: no textures or UVs, so texture it afterwards. GitHub: sakalond/ComfyUI-LATO2 All credit for the model to the LATO.2 authors: LoHhhha/LATO.2 (paper: arxiv.org). The meshes in the images were generated with TRELLIS.2 in StableGen, my Blender add-on for AI texturing and 3D generation (sakalond/StableGen). Happy to answer questions or take feature requests.

I made a free Blender add-on that turns a prompt into a textured, rigged character, with local models (and showcased it in Unreal) StableGen is my free, open-source Blender add-on. It runs local models through ComfyUI, so nothing goes to a paid API: - Prompt -> a few candidate images -> a TRELLIS.2 mesh. - Texturing: every camera view is generated (Qwen Image 2.1 here) and projected onto the mesh, with PBR maps. - Masked Edit: paint a part of the model and describe the change, for example swapping the jacket for armor, without redoing the rest. - Auto-rig with SkinTokens, then retarget walks or runs (the walk in the video is from Mixamo). - Export as GLB/FBX: the village in the video is 12 props made the same way and dropped into Unreal Engine 5.7. Current limits: the meshes are dense triangles, so for production you'd retopologize or decimate and bake (this is possible within the plugin). Texture quality depends a lot on the prompt and the model you pick. Qwen Image 2.1 is non-commercial; the add-on also supports models you can use commercially, such as SDXL and MOSS Image. GitHub: sakalond/StableGen Guide: github.com/sakalond/StableGen/wiki

Qwen Image 2.1 + TRELLIS.2 in Blender: from a prompt to a textured, rigged character, then a whole village in Unreal (StableGen 0.4.0) StableGen is my free and open-source Blender add-on that generates and textures 3D models with local models through ComfyUI. 0.4.0 is out, and everything in the video is a real run in the add-on: - Prompt -> mesh: a Preview gallery of candidate images to pick from, then TRELLIS.2 builds the mesh. - Texturing with Qwen Image 2.1: every camera view is generated and projected onto the mesh, then Marigold V2 makes the PBR maps (albedo, roughness, metallic, normal). - Masked Edit (new): paint a region on the model, describe the change, and only that part is regenerated, through the cameras that see it. - Auto-rigging (new) with SkinTokens, plus retargeting FBX/BVH animations onto the rig. - Experimental Qwen Image 2.1 Turbo, MOSS Image and FIBO 2.0 Edit, a reworked projection with cleaner seams, a new panel and a full user guide. In the video: Qwen Image 2.1, 10 cameras, on an RTX 4080 (16 GB). Time-lapse parts are marked. The village props went through the same pipeline and were exported to Unreal Engine 5.7 as GLB/FBX. Walk animation from Mixamo. Licenses: each model has its own; Qwen Image 2.1 and FIBO 2.0 Edit are non-commercial, MOSS Image is Apache-2.0. Check before commercial use. GitHub: sakalond/StableGen User guide: github.com/sakalond/StableGen/wiki Discord: discord.gg