Generative AI has taken a giant leap forward with the release of Midjourney v5, an achievement in photorealism that the computer graphics industry has been chasing for decades. The astonishing progress has been made possible by a group of just 11 talented individuals with a shared vision. In this article, we delve into the potential implications of these groundbreaking advancements in Generative AI for the 3D and VFX industries.
A significant improvement over its predecessor, Midjourney v5 boasts photorealistic results right out of the box. In comparison, the earlier version leans more towards a painterly, stylized approach. The new iteration has successfully crossed the uncanny valley, garnering recognition and admiration from professionals in the field.
With such capabilities, it’s no surprise that Generative AI, particularly Midjourney v5, is expected to disrupt 3D engines like Unreal, Unity, Octane, and Redshift. The generated outputs, even for digital humans, are more realistic than ever before. Midjourney v5 can create virtual sets with ease, surpassing the quality of Unreal Engine or Octane renders. The intricate details and dynamic range achieved are simply stunning.
The potential applications for Generative AI are numerous. In addition to digital art and gaming, product photography will benefit greatly. Designers can now visualize products before creating them, fine-tuning models and staging virtual shoots on demand. Previously, these tasks required scanning assets, modeling from scratch, and spending hours working with 3D tools.
Jensen Huang, CEO of NVIDIA, has acknowledged the significance of this shift, stating, “Every single pixel will be generated soon. Not rendered: generated.” Hybrid approaches that combine classical 3D with generative AI will likely dominate the market in the short term. For example, a lightweight 3D engine could be used for the first pass, with a generative filter applied on top to achieve AAA quality. This approach has been compared to NVIDIA’s “DLSS” on steroids.
As Generative AI continues to develop, video production is an obvious next target. Short films have already been created using nothing more than an iPhone, Midjourney, and a first-generation AI model, all within an apartment setting. This level of virtual production has the potential to democratize the filmmaking process and dramatically lower costs.
The ultimate goal of Midjourney’s creator, David Holz, has always been to revolutionize interactive content. The generative AI technology will initially transform ideation and asset creation, followed by 3D engine embellishment. Eventually, the vision is to enable users to experience their dreams through cloud-based interactions.
The rapid advancements in Generative AI are an exciting step forward for the world of 3D and VFX. As Midjourney v5 and similar technologies continue to evolve, the possibilities for creativity, productivity, and accessibility in these industries will be virtually limitless.