DaVinci Resolve 21: Beta version featuring AI that reads clapperboards, facial aging, synthetic voice, and much more

Blackmagic Design has officially announced DaVinci Resolve 21 in beta, introducing a set of updates that reposition the software beyond its traditional color grading and editing functions. This represents a structural expansion of the product: new pages, artificial intelligence tools applied to specific production challenges, and deeper integration between the editing, VFX, and audio modules.

For professionals who use Resolve as their primary workstation in a Linux environment—whether for film color grading, editing, VFX compositing, or sound mixing—this version deserves careful consideration. The changes directly impact established workflows and introduce capabilities that previously required auxiliary software or external pipelines.

In this analysis, we detail all the relevant resources for film and audiovisual producers, organized by area of impact.

The new "Photo" page

Anyone who works on hybrid productions—where a single project involves both video and still photography—knows the struggle of maintaining two separate workflows. Resolve 21 eliminates this problem once and for all with the introduction of Photo Page.

The idea is simple: to bring the same node-based color grading pipeline you already use in film production to still images. That means qualifiers, Power Windows, curves, and full support for DaVinci’s physical color panels—all available for photos, with completely non-destructive editing.

In practice, the workflow goes like this: you organize your footage into albums by shooting day or camera, use the LightBox view to apply and compare grids to dozens of photos at once, and can even crop and reframe while preserving the file’s original resolution at any time. Those who work with stills on set will especially appreciate the native tethering for Sony and Canon cameras—live capture with ISO, exposure, and white balance control directly from the software, saving images already organized into albums.

Integration with Blackmagic Cloud is also available here: colorists, editors, and cinematographers can access the same albums, metadata, and color looks simultaneously, from anywhere.

Artificial intelligence that solves real-world problems

The Davinci Neural Engine now includes a suite of AI tools designed to address specific day-to-day production challenges—without any marketing hype.

AI Slate ID is likely the most impactful feature for editing assistants: the AI automatically detects and reads the clapperboard in the frame, extracting the scene, take, and reel even under difficult conditions—such as a dark image, a blurry clapperboard, or a poor angle. Hours of manual metadata entry are simply a thing of the past.

AI IntelliSearch revolutionizes clip searching. Instead of watching clip by clip to find that specific shot, you describe what you want—a face, an object, a line of dialogue—and the AI locates it in your media library. It works for both video and photos.

AI Speech Generator solves a classic post-production problem: that 3-second gap in narration that needs filling. With just 10 seconds of reference audio from a voiceover artist or actor, the Neural Engine generates a voice model. You enter the text and adjust the speed, pitch, and intonation. It doesn’t replace actual ADR, but it handles emergencies competently.

To ensure continuity between scenes featuring characters of different ages, the AI Face Age Transformer adds or removes wrinkles and signs of aging directly to the recorded footage. The AI Face Reshaper goes a step further, allowing you to adjust the position and shape of the eyes, nose, mouth, and eyebrows on a moving face. The AI Blemish Removal tool addresses skin imperfections while maintaining natural texture—without the plastic-looking appearance that similar tools often produce.

In the optical domain, AI CineFocus lets you set the focal point and depth of field in post-production—including animating rack focus with keyframes. AI UltraSharpen restores footage with focus issues or upscaled footage that would previously have been discarded. And AI Motion Deblur removes motion blur, which is especially useful for slow-motion and freeze frames.

Edit and Cut: Less friction, more control

The assembly pages have been refined in ways that anyone working on complex projects will notice right away.

The curve editor now automatically scales the zoom in normalized mode, and four-point Bezier easing makes retiming much more precise and organic. The new loop, ping-pong, and relative modes for keyframe animations open up possibilities that previously required workarounds in Fusion. Speaking of Fusion: you can now adjust Fusion effects—text, transitions, motion graphics—directly in the keyframe and curve editors on the editing pages, without having to switch pages all the time.

For projects that use motion graphics, native support for Lottie animations (.lottie and .json) and OGraf HTML graphics is a handy addition: simply drag them onto the timeline, and the alpha channel is automatically recognized. The Text tool now features a multilingual spell checker, a dedicated font browser, and per-character styling within the same block—essential for productions with complex titles or those in multiple languages.

The Smart Bins on the Cut page feature dropdown menus and sophisticated metadata filters. Here’s a simple example: filtering to display only .braw clips with a green flag—exactly what you need when organizing footage from a large shoot.

Color: The centerfold has become more striking

The color page—the heart of Resolve—has received updates that directly impact highly complex projects.

The MultiMaster Trim Manager is the latest innovation for colorists who deliver in multiple formats. When you enable this mode, the node editor displays additional layers for trim operations, allowing you to manage HDR and SDR from the same timeline and render both at once. No more duplicate pipelines.

The Layer List View in the node editor is an interface change that will divide opinions—in a good way. If the node graph view used to be confusing in long projects, you can now switch to a layered list, with bypass, lock, label, and delete controls that are more directly accessible.

Render in Place for Magic Mask is one of those improvements that should have come sooner: now you can cache the traced mask as a traveling matte directly in the node, freeing up processing power for the next steps without disrupting the pipeline.

Group Versions are the highlight of this section: you can create and manage multiple grid versions for entire groups of pre-clips and post-clips, applying uniform styles in bulk and removing individual clips as needed.

Fusion and Fairlight: Deeper Integration

Fusion natively includes the Krokodove library—more than 70 compositing tools ranging from productivity utilities to customizable 2D and 3D templates. For those already familiar with Krokodove as a standalone plugin, having it built-in greatly simplifies project setup.

The most interesting new feature in Fusion, however, is the Fairlight Animator modifier: a modifier that connects Fusion directly to Fairlight’s audio engine. Audio levels from the timeline or Media Pool now automatically control visual parameters—lip sync, reaction to music beats, any animation that needs to respond to sound in real time. It’s the kind of integration that eliminates an entire export and reimport process.

The USD toolset has been updated to SDK 25.11 with the Hydra 2.0 API, adding support for 3D matte objects, textures via the USD Texture Projector and Catcher, and global in/out controls for the USD loader—which is relevant for VFX pipelines that already use Universal Scene Description.

In Fairlight, organizing tracks into collapsible folders solves a real problem in feature-length projects with dozens of audio tracks. Clip EQ now features 6 bands across all tabs, matching the mixer’s track EQ. The EQ and Level Matcher automates the adjustment of a clip’s tone and volume to match another—essential when interleaving material from different sources. And Chain FX lets you save and reuse chains of up to 6 plugins with fixed settings, speeding up the setup of recurring projects.

Technical improvements

Under the hood, the Davinci Neural Engine runs with full support for the Apple M Series and Snapdragon X Elite. For immersive content delivery, Resolve 21 offers comprehensive support for VR180 and VR360, direct export to Meta Quest and YouTube VR, spherical rotation via Panomap, and support for Apple Immersive Foveated Rendering —which prioritizes high-resolution rendering only in the central field of view, significantly reducing the GPU load.

In terms of codecs, MainConcept now offers encoding options for H.265 and MV-HEVC in 4:2:0 and 4:2:2, ensuring higher quality in 2D and 3D renders for compatible hardware.

For screenwriters and directors who use Final Draft, IntelliScript now natively imports the format and compares the script with the transcribed audio to automatically assemble a rough cut of the scene. For live productions, ATEM Mini ISO projects can be opened as a timeline and output in Blackmagic RAW.

Verdict

DaVinci Resolve 21 is not a minor update. It’s a major expansion in scope—with the Photo Page bringing the film pipeline to photography, the Neural Engine solving chronic on-set issues with applied AI, and deeper integrations between Fusion, Fairlight, and the editing pages.

For those who use Linux as their primary post-production platform, Resolve remains the most robust option on the market—and this version further reinforces that claim.

Download the beta from the Blackmagic website, take your time testing it out, and share your feedback here in the comments. Which of these tools solves a real pain point in your workflow?