While “Midjourney” for smartphones has officially arrived on the App Store, AI for image creation has now become a reality with many formidable competitors. Meanwhile, Nvidia similarly, he succeeded in producing a tool that could create polygonal models.
called Magic3D and an AI-based tool, including academic publications by Lin et al.
Its goal is to go and attack a market it currently dominates. Dream Fusionis the result of research and development by Google. The main difference between the two lies in the diffusion method.
DreamFusion uses a NeRF model While traditional, the Magic3D approach involves two basic steps, i.e. interpreting the input with the creation of a low resolution model, then optimizing and upgrading accordingly to make it usable.
According to the authors of the article, Magic3D is twice as fast Mountain View competitor. In addition, it will always be possible to create custom meshes that will be applied to the model via text input to create as reliable a rendering as possible.
An interesting project that adds to what has already been produced with dissemination tools. An example? Our final experience where we want the end of God of War Ragnarok from Midjourney!
Source: Every Eye