Hello Epic Games and Unreal Engine team,
I would like to ask a serious question about the long-term future of Unreal Engine and, at the same time, suggest three AI technologies that I believe could fundamentally change the accessibility and creative potential of the engine.
Are there any plans to integrate technologies like the following directly into Unreal Engine? If so, is there any indication of when we might begin to see them?
I am especially interested in three areas, listed below from the third most important to the one I believe could be truly transformative.
3. Voice/Text-to-3D: Creating 3D Content Through Natural Language
The first technology I would love to see is a deeply integrated AI creation system capable of generating and editing 3D content through normal human language.
Imagine being able to speak into a microphone and say:
“Create an ancient abandoned temple on top of a mountain. The temple should be approximately 200 meters long, partially destroyed, covered in vegetation, with enormous stone statues near the entrance, realistic weathering, physically accurate materials, and an interior consisting of several halls, corridors and underground chambers.”
Instead of manually performing every step, Unreal Engine could interpret the request and begin constructing the object or environment.
The same system could work through text prompts.
However, I believe it would become far more useful if it were not designed only around short prompts.
Ideally, users should be able to provide extensive descriptions, scripts, design documents, lore documents, architectural specifications, or even many pages of detailed instructions, and allow the AI to understand the complete context before generating the content.
For example, someone could provide a 10-, 20-, or 50-page description of a location, vehicle, spaceship, building, creature, city, or fictional world.
The AI could analyze that document, identify all of the objects, relationships, dimensions, materials, environments and design requirements described in it, and then construct the corresponding assets or scenes inside Unreal Engine.
Voice could become another interface to the same system.
The creator could simply talk to Unreal Engine:
“Move that building farther away.”
“Make the mountain twice as large.”
“Add damage to the left side of the spacecraft.”
“Make this city look as if it has been abandoned for 500 years.”
“Create the interior as well.”
“Now optimize everything for real-time rendering.”
In other words, natural language could become another genuine authoring interface for Unreal Engine.
2. Image-to-3D: Turning a Reference Image Into an Editable 3D Asset
The second technology is high-quality image-to-3D generation.
A user could provide a photograph, concept art, AI-generated image, sketch, or other visual reference, and Unreal Engine could reconstruct it as an actual editable 3D asset.
For example, I could first generate a concept image with an image-generation model: a futuristic spacecraft, a fantasy castle, an alien creature, an interior, a vehicle, a piece of furniture or an entire architectural structure.
I could then import that image into Unreal Engine and ask:
“Turn this into a complete 3D asset.”
Ideally, the result would not simply be a visual illusion.
It would become usable Unreal Engine content with real geometry, materials, textures, depth, appropriate scale, separate logical components and, where possible, useful collision and physically meaningful properties.
AI could then help reconstruct areas that are not visible in the original photograph by reasoning about the object’s structure while allowing the user to correct the result.
For artists, filmmakers, independent developers and people working with previsualization, this could dramatically shorten the distance between an idea and a usable 3D scene.
1. The Most Important Technology: Genie 3–Like AI World Generation Inside Unreal Engine
The third technology — and by far the most exciting one to me — is a world-model system inspired by technologies such as Google DeepMind’s Genie 3.
This feels almost naturally suited to something like Unreal Engine.
Imagine providing Unreal Engine with a photograph, concept image, text description, or a combination of all of them.
Instead of generating only one isolated object, the system would understand the entire scene and turn it into an explorable environment.
If I provide an image showing a distant futuristic city, mountains, a lake, forests, roads and enormous structures on the horizon, I would love to be able to tell Unreal Engine:
“Build this world and let me enter it.”
The AI would then reconstruct or generate what exists beyond the original camera view.
I could turn around.
I could walk toward the distant building.
I could enter it.
I could explore streets that were only implied in the original image.
I could travel beyond the mountains.
The world could continue to exist beyond the boundaries of the original 2D reference while attempting to preserve its visual identity, architecture, atmosphere and internal logic.
The truly exciting possibility would be combining a world model with Unreal Engine’s existing real-time technologies.
Instead of the generated environment existing only as temporary generated frames, could a future system progressively translate or “bake” the result into persistent Unreal Engine content — geometry, terrain, materials, lights, vegetation, objects, physics, navigation, animation and other editable elements?
That distinction is extremely important.
A world model could provide imagination and rapid generation, while Unreal Engine could provide persistence, editability, simulation, professional rendering and production tools.
The combination of those two concepts could be extraordinarily powerful.
Why I Believe This Matters
Today, Unreal Engine is an incredibly powerful tool, but fully mastering professional 3D creation still requires a very significant investment of time.
This is not necessarily a problem with Unreal Engine itself. Professional 3D production is simply complicated.
A person may have an extraordinary world in their imagination, an idea for a film, a cinematic, a game, a historical reconstruction, a science-fiction universe or an interactive story — yet not have years available to master modeling, texturing, rigging, animation, lighting, programming, level design and all the other disciplines traditionally required to create it.
That means there is still an enormous gap between human imagination and the ability to transform that imagination into an interactive world.
I do not believe AI should eliminate professional artists, developers or technical specialists.
Instead, I believe it could create an entirely new level of accessibility.
Professional creators could use these systems as extremely powerful acceleration tools, while people with little or no traditional 3D experience could finally have a way to express ideas that would otherwise remain only in their imagination.
Someone who wants to create a short cinematic should be able to describe it.
Someone who has designed a fictional universe for years should be able to give Unreal Engine hundreds of pages of information and begin visualizing it.
Someone who has a photograph or concept image of a place that does not exist should be able to step into it.
And after AI generates the initial result, advanced users should still be able to open everything in the normal Unreal Engine tools and manually modify every part of it.
That, to me, would be the ideal relationship between generative AI and a professional engine.
My Questions to Epic Games
So I would genuinely like to ask the Unreal Engine team:
-
Is Epic Games actively researching native natural-language or voice-driven 3D creation for Unreal Engine?
-
Are there plans for native image-to-3D asset generation or reconstruction directly inside the Unreal Editor?
-
Is Epic researching world-model technology capable of generating explorable, persistent environments from text, images, concept art, or other references — something conceptually similar to Google DeepMind’s Genie 3, but designed around Unreal Engine’s professional real-time ecosystem?
-
Could generative world models eventually work together with technologies such as Nanite, Lumen, PCG, World Partition, MetaHuman, Chaos and Unreal Engine’s existing procedural systems?
-
Most importantly: does Epic see a future in which someone without years of traditional Unreal Engine experience can describe, show or speak an idea and have the engine help transform it into a real, editable interactive world?
If technologies like these are already part of Epic’s internal research or long-term roadmap, I would be extremely interested to know whether there is anything publicly available about them and what kind of timeframe the community might realistically expect.
I genuinely believe this could represent one of the biggest steps in Unreal Engine’s history.
For decades, game engines have become better at rendering worlds.
The next revolution may be allowing far more people to actually create those worlds.
Not by replacing creativity, but by removing the technical wall between imagination and creation.
Thank you to the Unreal Engine team for reading this, and I would be very interested to hear the thoughts of Epic developers and the wider Unreal Engine community.