Every filmmaker understands the struggle. The performance is perfect. The timing is right. But the lighting is wrong — the sky looks flat; colors shift between angles, and a scene that felt cinematic on set appears lifeless on screen.

Fixing these issues has always demanded hours of painstaking work. And even then, changing lighting or color without introducing visual artefacts remains one of post-production's most persistent challenges.

Two of Dolby’s research efforts address this problem from different directions.

Relighting Scenes With AI Precision

The first uses diffusion models to realistically transform lighting conditions after capturing. This is not a brightness adjustment. The system understands how light interacts with surfaces, shadows, and the environment — producing coherent results where conventional methods break down most visibly: complex skies, outdoor landscapes, and mixed light sources. The outcome is lighting that looks like it was always there. Read More: Outdoor Scene Relighting with Diffusion Models

Color That Understands What It Is Changing

The second tackles semantic video color transfer, enabling creators to draw color characteristics from reference images or videos while preserving realism and temporal consistency. Rather than applying a global filter, the system targets skies, water, foliage, and other scene elements independently. An online training strategy reduces flickering between frames, delivering the stability ofprofessional workflows' demand. Read More: Temporal Consistent Semantic Video Color Transfer from Multiple References

What This Means for Creators

The impact of these technologies is less about automation and more about creative flexibility. A creator could shoot a scene under overcast skies and later explore how it looks at sunset. A colorist could match footage from different cameras, locations, or lighting conditions without spending hours making manual corrections frame by frame.

As these systems continue to mature, they could become powerful tools that help creators spend less time correcting footage and more time shaping the visual story they want to tell.

From Flat Footage to Cinematic Storytelling: AI-Powered Relighting and Color Transfer

{{ pageEducationalSlot.headingContent[index] }}