腾讯混元开源5款3D生成模型：亚秒级生成、多模态支持，重构3D内容创作生态

腾讯混元在3D开源日活动中宣布重磅升级，一次性开源5款基于Hunyuan3D-2.0架构的全新3D生成模型，并同步推出升级后的3D AI创作引擎，面向开发者与C端用户全面开放。此次开源的模型涵盖Turbo加速系列、多视图版本（3Dmv）及轻量版（3Dmini），通过自研加速框架与功能创新，实现了生成速度、细节精度与设备兼容性的跨越式突破。

核心技术突破：亚秒级生成与全场景适配

此次开源的Turbo系列模型（Hunyuan3D-2 Turbo、3Dmini Turbo、3Dmv Turbo）搭载腾讯自研FlashVDM加速框架，通过DiT采样优化与层次化体素解码技术，将3D模型生成时间从传统的30秒级压缩至1秒内，其中轻量版3Dmini Turbo更实现0.5秒亚秒级生成，计算量较前代减少95%以上。该框架支持5GB显存起步的消费级显卡（如NVIDIA 4050/3050/2060、AMD RX 6600等），甚至可在苹果M1芯片及CPU设备上流畅运行，打破了3D生成对高端硬件的依赖。

多模态交互升级：从文本到多视图的精准控制

• 多视图输入（3Dmv系列）：用户上传2-4张标准视角图片（如产品多角度照片、手绘三视图），模型即可生成高精度3D模型。实测中，通过手机拍摄企鹅公仔正反面照片，1分钟内即可完成带场景的3D建模，显著降低游戏原画转3D、手办设计等场景的成本。
• 智能减面与PBR材质：引擎新增3D智能减面功能，可根据需求自动优化三角面数量（数百至数千面），在保留细节的同时降低渲染压力；物理渲染（PBR）技术升级后，模型材质光影效果更贴近真实物理规律，支持电影级质感呈现。

全格式兼容：打通创作到落地的最后一公里

升级后的3D AI创作引擎支持OBJ/GLB/FBX/STL/USDZ等10余种主流格式输出，无缝对接Blender、3D打印设备及移动端交互工具。用户可直接将生成结果导出为GIF/MP4动画，或通过插件在Blender中实时编辑，覆盖从原型设计到产品落地的全流程需求。

开源生态布局：推动3D生成普惠化

此次开源的5款模型（包括完整版Hunyuan3D-2、轻量版3Dmini、多视图3Dmv及三者的Turbo版本）已全部上线Hugging Face与GitHub，开发者可自由调用或二次开发。腾讯数据显示，其3D模型已在游戏资产生成、电商商品建模、UGC内容创作等场景落地，部分游戏3D资产已满足几何布线、贴图精度与骨骼蒙皮的工业化标准。

“我们希望通过开源降低3D创作门槛，让每个人都能高效创造数字资产。”腾讯混元团队表示，未来将持续优化FlashVDM框架对纹理生成的加速，目标将“模型+纹理”全管线耗时压缩至10秒内，并探索AI编辑功能的集成。

行业影响与展望

随着3D生成技术从“可用”走向“易用”，腾讯的开源策略正加速推动行业变革：游戏开发者可通过秒级迭代优化角色设计，电商商家能低成本生成商品3D展示，普通用户也可轻松创作个性化手办。业内认为，此次技术突破不仅为元宇宙、AIGC等领域提供基建支持，更通过开源生态构建了国产3D生成技术的竞争力壁垒。

体验链接：https://3d.hunyuan.tencent.com/
开源地址：GitHub | Hugging Face

—— 腾讯混元以技术创新重塑3D创作边界，开启全民3D内容生产新时代——

快速上手

获取代码

!git clone https://github.com/Tencent/Hunyuan3D-2

安装环境

cd Hunyuan3D-2
pip install -r requirements.txt
# for texture
cd hy3dgen/texgen/custom_rasterizer
python3 setup.py install
cd ../../..
cd hy3dgen/texgen/differentiable_renderer
python3 setup.py install

推理：
Hunyuan3D-2mini

%%bash
cd Hunyuan3D-2
python
import time

import torch
from PIL import Image

from hy3dgen.rembg import BackgroundRemover
from hy3dgen.shapegen import Hunyuan3DDiTFlowMatchingPipeline

image_path = 'assets/demo.png'
image = Image.open(image_path).convert("RGBA")
if image.mode == 'RGB':
    rembg = BackgroundRemover()
    image = rembg(image)

pipeline = Hunyuan3DDiTFlowMatchingPipeline.from_pretrained(
    'tencent/Hunyuan3D-2mini',
    subfolder='hunyuan3d-dit-v2-mini',
    variant='fp16'
)

start_time = time.time()
mesh = pipeline(
    image=image,
    num_inference_steps=50,
    octree_resolution=380,
    num_chunks=20000,
    generator=torch.manual_seed(12345),
    output_type='trimesh'
)[0]
print("--- %s seconds ---" % (time.time() - start_time))
mesh.export(f'demo_mini.glb')

%%bash
cd Hunyuan3D-2
python
import time

import torch
from PIL import Image

from hy3dgen.rembg import BackgroundRemover
from hy3dgen.shapegen import Hunyuan3DDiTFlowMatchingPipeline
from hy3dgen.texgen import Hunyuan3DPaintPipeline

image_path = 'assets/demo.png'
image = Image.open(image_path).convert("RGBA")
if image.mode == 'RGB':
    rembg = BackgroundRemover()
    image = rembg(image)

pipeline = Hunyuan3DDiTFlowMatchingPipeline.from_pretrained(
    'tencent/Hunyuan3D-2mini',
    subfolder='hunyuan3d-dit-v2-mini',
    variant='fp16'
)
pipeline_texgen = Hunyuan3DPaintPipeline.from_pretrained('tencent/Hunyuan3D-2')

start_time = time.time()
mesh = pipeline(
    image=image,
    num_inference_steps=50,
    octree_resolution=380,
    num_chunks=20000,
    generator=torch.manual_seed(12345),
    output_type='trimesh'
)[0]
print("--- %s seconds ---" % (time.time() - start_time))
mesh.export(f'demo_mini2.glb')

mesh = pipeline_texgen(mesh, image=image)
mesh.export('demo_textured_mini.glb')

Hunyuan3D-2mv

%%bash
cd Hunyuan3D-2
python
import time

import torch
from PIL import Image

from hy3dgen.rembg import BackgroundRemover
from hy3dgen.shapegen import Hunyuan3DDiTFlowMatchingPipeline

images = {
    "front": "assets/example_mv_images/1/front.png",
    "left": "assets/example_mv_images/1/left.png",
    "back": "assets/example_mv_images/1/back.png"
}

for key in images:
    images[key] = Image.open(images[key]).convert("RGBA")
    image = images[key]
    if image.mode == 'RGB':
        rembg = BackgroundRemover()
        image = rembg(image)
    images[key] = image

pipeline = Hunyuan3DDiTFlowMatchingPipeline.from_pretrained(
    'tencent/Hunyuan3D-2mv',
    subfolder='hunyuan3d-dit-v2-mv',
    variant='fp16'
)

start_time = time.time()
mesh = pipeline(
    image=images,
    num_inference_steps=50,
    octree_resolution=380,
    num_chunks=20000,
    generator=torch.manual_seed(12345),
    output_type='trimesh'
)[0]
print("--- %s seconds ---" % (time.time() - start_time))
mesh.export(f'demo_mv.glb')

%%bash
cd Hunyuan3D-2
python
import time

import torch
from PIL import Image

from hy3dgen.rembg import BackgroundRemover
from hy3dgen.shapegen import Hunyuan3DDiTFlowMatchingPipeline
from hy3dgen.texgen import Hunyuan3DPaintPipeline

images = {
    "front": "assets/example_mv_images/1/front.png",
    "left": "assets/example_mv_images/1/left.png",
    "back": "assets/example_mv_images/1/back.png"
}

for key in images:
    images[key] = Image.open(images[key]).convert("RGBA")
    image = images[key]
    if image.mode == 'RGB':
        rembg = BackgroundRemover()
        image = rembg(image)
    images[key] = image

pipeline = Hunyuan3DDiTFlowMatchingPipeline.from_pretrained(
    'tencent/Hunyuan3D-2mv',
    subfolder='hunyuan3d-dit-v2-mv',
    variant='fp16'
)
pipeline_texgen = Hunyuan3DPaintPipeline.from_pretrained('tencent/Hunyuan3D-2')

start_time = time.time()
mesh = pipeline(
    image=images,
    num_inference_steps=50,
    octree_resolution=380,
    num_chunks=20000,
    generator=torch.manual_seed(12345),
    output_type='trimesh'
)[0]
print("--- %s seconds ---" % (time.time() - start_time))
mesh.export(f'demo_white_mesh_mv.glb')

mesh = pipeline_texgen(mesh, image=images["front"])
mesh.export('demo_textured_mv.glb')