New Arm tech enables the use of Unreal Engine MegaLights on mobile

Arm is showing off a mobile path to Unreal Engine MegaLights that wasn’t practical before: keep more of the lighting and reconstruction work on-chip, and use neural tech to offset the cost of ray tracing. The demo game, Neural Dawn, is built with Sumo Digital and runs on Arm Mali GPUs, which Arm says will ship later this year.

The two headline pieces are Neural Super Sampling and Denoising (NSSD) and Neural Frame Rate Upscaling (NFRU). NSSD reconstructs image detail from fewer rays, which is how Arm claims MegaLights and ray tracing become workable on mobile. NFRU then generates an intermediate frame so 30 fps content can be presented at 60 fps. Arm is positioning this as low-latency GPU/ML cooperation rather than a cloud or post-process trick.

For developers, the practical implication is less about a single demo and more about a potential shift in mobile content budgets. Sumo says the tech let the team treat the project more like a console/PC production: more assets, larger textures, more materials, and fully dynamic lighting instead of baked lightmaps. The team was only 17 people over 18 months, which is a useful data point for scope planning.

The broader context is that mobile hardware is still constrained by battery and thermals, so any feature that trades compute for quality has to justify itself in real production. Arm is also emphasizing openness: the models are published and can be retrained, which may matter to studios that want to inspect or adapt the pipeline rather than treat it as a black box. Neural Dawn is slated for later this year, with an Android...