In the evolving landscape of digital video production, the gap between generative tools and established post production standards has remained a persistent practical concern for many creators and studios.
Outputs from AI systems typically arrive in standard dynamic range formats with limited bit depth, which can restrict options during color grading, compositing, and final delivery for high end displays or theatrical work.
Recent developments from companies building generative media platforms have begun to address this specific integration challenge.
And Runway is playing its part.
The company introduced a model named 'Ruby' designed for converting standard dynamic range video into high dynamic range material.\
It takes SDR input, whether from user uploads or from generations produced by other models on the platform, and extends the brightness range into HDR headroom while aiming to preserve the original pixel data and accompanying audio.
The process applies a bounded color preserving grade rather than regenerating the content.
Supported outputs include 10 bit HEVC in HDR10 or HLG, ProRes in various profiles with BT.2020 and PQ encoding, and sequences of half float OpenEXR frames suitable for compositing pipelines.
Input constraints limit clips to roughly 30 to 40 seconds in length and under 4096 pixels on a side, with the source required to be tagged as SDR.
The feature became available through the Runway web application for users on Max plans and Enterprise accounts, as well as via the developer API under the video to hdr endpoint.
Company leadership described the capability as one that had been frequently requested over the course of nearly eight years of platform development. Concurrent updates also expanded native HDR delivery options for the Gen 4.5 generation model.
Early responses from users testing the conversion on calibrated displays have noted improvements in highlight recovery and shadow detail for certain AI generated footage when viewed on HDR capable monitors.
Other observations have pointed to residual banding in EXR exports and questioned the extent of new information recovered in extreme tonal ranges.
These assessments remain preliminary as the tool sees broader use across different source materials and production contexts.
Overall the addition positions generative video more closely within conventional professional finishing workflows by providing intermediate formats that align with existing color management and editing software expectations.
