Resolve FX · Blur · 02

AI CineFocus in DaVinci Resolve

AI CineFocus builds a virtual lens around an AI-generated depth model. You can choose a focal point, change aperture and depth of field, shape bokeh, add optical character, improve foreground edges and keep the final defocus integrated with grain and a global blend.

The screenshots below come from the real DaVinci Resolve 21 AI CineFocus 2 panel, so every section maps directly to the interface.

Depth MapFocusBokehLens PropertiesGrainGlobal Blend

Controls

Build the virtual lens from focus to final blend

CineFocus is easier to understand when you separate the task into depth, focus, lens behaviour, edge cleanup and final integration.

Focus: focal point and depth of field

The Focus section defines the core behaviour of the virtual lens. Aperture Size shapes the defocus, Focus Distance sets the focal plane, and Depth of Field controls how much of the scene remains sharp. Track Focus to Point helps keep the selected focus point over time, while Focus Peaking makes the focus region easier to judge.

Aperture Size — strength and character of defocus
Focus Distance — location of the focal plane
Depth of Field — width of the sharp region
Track Focus to Point — follows the focus point through motion

Bokeh Quality: Faster, Better and Enhanced

Bokeh Quality lets you choose the trade-off between processing speed and the quality of the virtual defocus. Faster is useful while adjusting the effect interactively, Better gives a cleaner result, and Enhanced is appropriate when bokeh quality matters more than preview speed.

Faster — for responsive setup
Better — cleaner rendering
Enhanced — highest quality priority

Depth Map: how CineFocus understands scene depth

CineFocus builds a depth model from a regular 2D image. In Depth Map you can choose the source and analysis quality, then invert or refine the map when needed. Separate Layers helps create clearer depth separation, while Post-Processing can clean the map before the final defocus is generated.

Source — Internal or From Alpha Input
Quality — Faster / Better
Adjust Map Levels — manual range refinement
Separate Layers — stronger separation between depth planes

Lens Properties and Aperture

After the basic focus is set, CineFocus can shape the character of the virtual lens itself. Chromatic Aberration and Spherical Aberration add optical traits, while Shaped Bokeh changes highlight geometry through blade count, Rotation, Curvature and Pinwheel. Anamorphic Lens provides a separate control for an anamorphic-style response.

Chromatic / Spherical Aberration — optical character
Shaped Bokeh — shape of out-of-focus highlights
Number of Blades / Rotation / Curvature / Pinwheel
Anamorphic Lens — stretched bokeh character

Lens Distortion, Foreground Edge Overlap and Grain

The most obvious artificial-depth artefacts often appear where foreground and background meet. Foreground Edge Overlap helps handle those boundaries more gracefully. Lens Distortion adds controlled curvature and aberration, while Remove and Replace Grain can remove grain before defocus and restore it afterward.

Automatic Edge Handling — helps with difficult boundaries
Overlap Boost — strengthens foreground overlap handling
Remove Grain — cleans grain before defocus
Replace Grain After Defocus — restores texture after processing

Additional Grain and Global Blend

Additional Grain is useful when the defocused image becomes too clean or digital. Strength, Size, Softness and Saturation can restore organic texture. Global Blend then controls the amount of the entire CineFocus result and quickly brings back part of the original image.

Add Grain — extra texture after defocus
Strength / Size / Softness / Saturation
Global Blend — overall AI CineFocus strength

Depth model

The depth map is the foundation of convincing defocus

If the depth model is wrong, no aperture shape will save the result. Check the map first, then refine the focus and lens character.

Practical use

Where AI CineFocus is actually useful

Change focus after the shoot

Choose the subject that should remain sharp and rebuild the perceived focus around it.

Create shallow depth of field

Useful for footage where everything was captured sharply but a more selective cinematic focus is needed.

Shape bokeh and lens character

Change aperture geometry, aberrations and anamorphic character instead of applying a generic blur.

Keep the result integrated

Grain controls and edge handling help the defocus feel like part of the shot rather than a layer pasted on top.

Working order

A clean workflow prevents artificial-looking depth

01

Inspect depth first

Confirm that foreground, subject and background are separated correctly before tuning the lens.

02

Set the focal point

Use Focus Distance and Depth of Field to place the sharp zone where the story needs it.

03

Shape the lens

Only after focus is stable should you tune aperture shape, aberrations and anamorphic traits.

04

Clean edges and integrate

Use edge overlap, grain and Global Blend to remove the obvious synthetic feel.

Typical mistakes

The effect usually breaks at depth boundaries, not in the centre of the subject

Trusting the depth map without checking it

Hair, thin objects and overlapping planes can expose depth errors immediately.

Using too much aperture blur

A dramatic blur is easy to notice but often makes the shot feel artificial. Pull back before the effect becomes the subject.

Ignoring foreground edges

Foreground Edge Overlap exists because boundaries are where synthetic defocus is easiest to detect.

Removing texture and never restoring it

Heavy defocus can make grain behaviour inconsistent. Remove/replace or additional grain can help the processed region match the source.

AI CineFocus vs ordinary blur

Use depth-aware defocus only when the shot needs depth logic

Box Blur

Simple blur without depth logic

Choose Box Blur when the whole task is simply to soften or blur pixels. It is lighter conceptually and does not attempt to reconstruct scene depth.

Open Box Blur
AI CineFocus

Depth-aware virtual lens

Use CineFocus when foreground, subject and background must react differently and the final defocus needs to behave like a lens rather than a uniform blur.

Depth-aware virtual lens

AI CineFocus: how the effect behaves in a real node tree

AI CineFocus in Resolve 21 first builds or receives a Depth Map, then uses it as the basis for virtual focus and defocus. That makes depth-map quality more important than attractive bokeh shape: wrong depth inevitably produces wrong blur.

The official Resolve 21 New Features Guide describes an Internal AI Depth Map or From Alpha Input, Focus Distance as a percentage of the depth map, Depth of Field, Track Focus to Point, Bokeh Quality, Lens Properties and edge/grain tools. It is a Studio AI tool rather than a basic free blur filter.
The Depth Map is an estimate of 3D structure from 2D, not measured scene geometry. Hair, glass, smoke, thin branches, reflections and foreground overlap deserve special scrutiny.
AI CineFocus is a DaVinci Resolve Studio 21 feature; Blackmagic explicitly lists CineFocus among the new AI tools in Studio 21.

Use it when

  • refocusing phone/deep-focus footage
  • virtual focus pulls between depth planes
  • optical bokeh with automatic depth separation

Choose another tool when

  • you simply need to conceal a masked object — a normal Blur is simpler and faster
  • the depth map consistently fails on a complex scene — manual mattes may be more reliable

What each control changes — and what to inspect

Control / layerMeaningInspection point
Depth Source / QualityInternal AI or external Alpha; Faster/Better for the map.Inspect Depth mode before judging final bokeh.
Focus Distance / DoFFocal plane position and in-focus depth width based on the depth map.Focus Distance is not metres/feet — it is relative to the Depth Map.
Bokeh / apertureVirtual shape and quality of defocus after depth is correct.Test bright highlights and foreground edges together.
Edge / grainForeground overlap, remove/replace grain and final integration.Inspect hair/backlight and grain continuity between sharp/blurred regions.

Production workflow

  1. 01Switch to Depth view and refine Far/Near/Gamma/Separate Layers before lens styling.
  2. 02Set Focus Distance and Depth of Field until the in-focus region is correct.
  3. 03Only then tune Aperture/Bokeh/Lens Properties.
  4. 04Check foreground overlap, hair, transparency and motion; use external Alpha when needed.
  5. 05Handle grain around the defocus and finish with Global Blend.

Diagnostics: symptom → cause → fix

Hair blurs like background

Likely cause: Depth Map does not separate thin foreground detail.

Fix: Refine map levels/overlap, use Better quality or external Alpha.

Focus breathes/drifts through motion

Likely cause: Depth estimate varies frame-to-frame or focus point is not tracked.

Fix: Use Track Focus to Point and inspect the depth map across the shot.

Bokeh is good but result feels plastic

Likely cause: Grain was removed/blurred and not restored.

Fix: Use Remove/Replace Grain or restore matched texture after defocus.

Performance

CineFocus combines AI depth estimation with lens simulation, making it one of the heaviest FX in this set. Faster is useful while adjusting; Better/Enhanced for final edge and bokeh checks; long 4K/8K shots benefit from caching.

Animation & keyframes

Focus Distance can be keyframed for virtual focus pulls, while Track Focus to Point can follow a moving subject. Curve timing should resemble real focus behaviour or the post-process becomes obvious.

Node placement

Bring the base normalisation/grade to a stable state before CineFocus and place final grain after it. When Depth Source comes from Alpha Input, matte routing becomes part of the pipeline and should be explicit in the node tree.

Scene / Display

Depth Map is not colour science, but CineFocus placement still matters: technical transforms and display rendering should not accidentally complicate matte/depth workflow.

Primary references used for this technical layer

Control names and panel layout are also cross-checked against the real DaVinci Resolve 21 screenshots shown earlier on this page. Where Blackmagic does not publish implementation math, this guide describes observable behaviour and does not invent a proprietary kernel.

FAQ

AI CineFocus in DaVinci Resolve: common questions

What does AI CineFocus do in DaVinci Resolve?

It builds a depth-aware virtual lens from a 2D image, letting you choose a focus point and control aperture, depth of field, bokeh and other optical characteristics in post.

What is the difference between Faster, Better and Enhanced Bokeh Quality?

They trade processing speed for bokeh quality. Faster is useful for responsive adjustment, Better improves the result, and Enhanced is intended when visual quality has priority over speed.

Can AI CineFocus use an external alpha instead of its internal depth map?

Yes. The Depth Map Source menu shown in the Resolve 21 panel offers Internal and From Alpha Input, allowing an external alpha-driven workflow when needed.

Why does CineFocus include grain controls?

Defocus changes fine texture. Removing grain before the blur and restoring or adding it afterward can make processed regions match the original shot more naturally.

When is Box Blur better than AI CineFocus?

Use Box Blur when you simply need a controllable blur without scene-depth logic. CineFocus makes sense when the blur must depend on the depth of the scene and behave more like a virtual lens.