The author of this article is Li Dan
Source: Hard AI
Generative artificial intelligence (AI) star startup Stability AI released its latest breakthrough 3D model, Stable Fast 3D, in August, which dramatically accelerates the speed at which a single image can be generated into a 3D image.
According to Stability AI, Stable Fast 3D converts a single input image into a detailed 3D asset in just 0.5 seconds, arguing that this innovative model sets a new benchmark for speed and quality in the field of 3D reconstruction. It quickly generates complete 3D assets, including UV split networks, material parameters, and albedo colors that reduce light baking, and users can remesh with optional quad or triangle remeshing in just 100 to 200 milliseconds.
Stable Fast 3D问世代表Stability AI在缩短AI模型处理时间方面实现重大飞跃。
In March of this year, Stability AI released the 3D model Stable Video 3D (SV3D), which is an improvement on Stable Video Diffusion, which can generate the corresponding 3D model with just one image, which is suitable for new perspective compositing tasks and 3D generation tasks. SV3D takes up to 10 minutes to generate 3D assets, while Stable Fast 3D accomplishes the same task in 0.5 seconds, 1,200 times faster.
When Stability AI released Stale Fast 3D, it specifically mentioned that the model outperformed the competition in several key areas, one of which was that it took 0.5 seconds to generate 3D assets per time on a GPU with 7GB of VRAM memory and nearly a second on the Stability AI API. In addition, the model features a high-quality UV splitting network and material parameters that reduce entangled lighting in textures and generate additional material parameters and normal maps.
According to Stability AI, Stable Fast 3D is based on TripoSR, a model developed in collaboration with Tripo AI, which significantly improves the architecture and enhances functionality. The model is suitable for game and virtual reality (VR) developers, as well as professionals in retail, architecture, design, and other graphics-intensive industries. There are a variety of use cases in games and film production:
- During pre-production, where experimentation is critical, use the model to achieve fast inference.
- Static assets for the game (background objects, clutter, furniture)
- 3D model of e-commerce
- Rapid model creation for enhanced display (AR)/VR.
目前，Stable Fast 3D可通过Stability AI 的聊天机器人Stable Assistant 和 Stability AI API 使用。 用户可以在AI开源社区Hugging Face上使用该模型，Stability AI基于免费用于研究、非商业和商业用途的Stability AI 社区许可而发布。
After Stability AI officially announced the launch of Stable Fast 3D, netizens on social media quickly responded to the post, and some netizens asked how clear the UV network generated by the model is and how many polygons there are. Some netizens said they liked the GIF that Stability AI attached when it released the model, which showed the machine spitting out generic AI image models and throwing them directly into the trash. Some netizens asked if it was possible to print out the generated 3D results.
Some commentators say that content creation in the 3D world is a bottleneck at the moment. We don't want to spend money to handcraft the metaverse. Auto-generated assets should be the default, just like in the real world. Stability AI's new model is a big step in the right direction.
This article is from Wall Street News, welcome to download the APP to see more