Artificial intelligence is no longer just generating images. It is beginning to construct spaces.

That shift became visible in recent tests with ChatGPT Images 2.0, OpenAI’s latest image generation model. The technology is not yet ready for professional VR production, and its most obvious limitation remains resolution. But for immersive creators, the direction is becoming difficult to ignore.

Daniel Vega, founder of ZeusXR and a VR180 content producer, says the most surprising result was not simply the quality of the image, but the way the model appeared to understand the scene as a 360-degree environment.

“What really surprised me was its ability to generate equirectangular 360 images with spatial coherence, without the need for stitching,” Vega said. “I generated a landscape, tested it immediately inside a headset like Meta Quest 3, and the sense of presence was much more convincing than I expected at this stage.”

This does not mean AI-generated VR is here. Professional immersive production operates at a very different technical level. VR180 and 360 workflows often depend on high-resolution stereoscopic files, with delivery targets ranging from 8K to 16K depending on the platform, headset and production standard. Current generative image tools are still far from that threshold.

They also lack real stereoscopy, precise control over scale and consistent depth. For final production, these are not minor limitations. They are structural barriers.

“We are not talking about professional VR180 quality yet,” Vega said. “But there is a technical foundation that did not exist before: an AI system that can begin to treat a scene as a 360-degree space. Conceptually, that changes the starting point for immersive creation.”

For now, the most immediate impact is not in final delivery, but in preproduction.

Concept design, visual scouting and scene exploration are often time-consuming parts of immersive production. A director or studio may need to evaluate locations, moods, spatial layouts and visual references before cameras, crews or physical scouting enter the process. In that context, AI-generated 360 environments could become a practical creative tool.

Instead of waiting days to visualize a concept, teams can explore multiple spatial directions in minutes. The result is not production-ready, but it can help accelerate decision-making before the real production begins.

That distinction matters. In immersive media, every creative decision affects budget, logistics, camera placement, blocking, lighting and post-production. A faster previsualization process can reduce uncertainty before a project moves into execution.

Why this feels different

AI tools such as Midjourney and Stable Diffusion have previously been used to experiment with panoramas and immersive-looking images. Some results were visually appealing on a flat screen, but often broke down when viewed inside a headset. The image might look wide, but the spatial logic did not always hold.

The difference in these new tests is not only the output format. It is the impression that the model is beginning to build a more coherent environment from the start. Scale, continuity and scene logic feel more convincing, even if the results are still imperfect.

That is the key signal for XR creators. A 360 image does not become immersive just because it wraps around the viewer. It needs spatial consistency. It needs a believable relationship between foreground, background, horizon, scale and visual continuity. Without that, the illusion collapses quickly inside a headset.

The latest generation of image models suggests that AI is getting closer to that requirement.

The next threshold

For AI-generated immersive media to move from concept material to publishable XR content, two major advances are still needed: real stereoscopy and temporal consistency.

Stereoscopy would allow the image to carry correct binocular depth, which is essential for convincing VR180 and high-end immersive formats. Temporal consistency would allow AI to generate video where objects, depth, lighting and camera continuity remain stable over time.

Without those two elements, AI-generated 360 images remain useful as references, prototypes or creative exploration. With them, the conversation changes entirely.

“I would not say this is immediate,” Vega said. “But the direction is clear. First, this opens the door to production support tools. Eventually, it could lead to immersive content generated entirely with AI.”

That future is not here yet. Today, the technology still lacks the resolution, stability and stereoscopic depth required for professional XR distribution. OpenAI’s own documentation describes GPT Image 2 as a state-of-the-art image generation and editing model, not as a dedicated VR or 360 production engine.

But the early signal is important.

The question is no longer whether AI will enter the XR pipeline. It already has. The more relevant question is where it will become useful first.

For now, the answer is preproduction.

Final immersive production remains an open challenge. But the starting point has changed. AI is beginning to move from generating pictures to imagining spaces. And for XR, that is a much more important shift.