The argument in one line.
DaVinci Resolve's AI subtitle tool is most powerful when used as a free word-timing ruler — generate single-word subtitles for their cut marks, then delete the layer and animate your own styled text to those marks.
Read if. Skip if.
- You edit talking-head, real estate, or social media videos in DaVinci Resolve and want animated word-synced captions without a separate caption tool.
- You know your way around the DaVinci Edit page but have never used Fusion's camera tracker or the Color page magic mask.
- You want a step-by-step method for placing text behind a subject or locking text into a 3D scene without a motion-graphics background.
- You edit in Premiere Pro, CapCut, or any other NLE — almost nothing here transfers directly.
- You only need static or auto-generated captions and have no interest in per-word animation.
The full version, fast.
The core insight is that DaVinci's AI subtitle generator is a better timing tool than a caption tool — set it to one character per word, let it cut the timeline, then hide the layer and use those cuts as rails for your own animated text. From there, per-word animation direction is set clip by clip in the animation tab, with a universal 5-frame leftward offset to ensure animations resolve before the spoken word lands. Compositing text behind a subject runs through the Color page magic mask with an alpha output, requiring no Fusion work. For 3D scene tracking, the workflow is: compound clip → Fusion clip → camera tracker solve → image plane transform → mask-inverted drop shadow → magic mask subject back in front.
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.

01 · Intro
Host introduces the three techniques: timing, text-behind-subject, and 3D tracking.

02 · Timing Subtitles
AI subtitle tool set to 1 char/word max → generates cut marks → layer disabled, marks used as timing rails.

03 · Building the Visuals
One clip per word, mixed fonts/sizes, Lux Slide and Shine effects, positional layout.

04 · Fixing Animation Directions
Per-clip animation tab: uncheck mirror-out, set directional in/out, Fix Animation button to auto-recalculate duration.

05 · Fine Tuning Timing
Select all text, shift 5 frames left using comma key, extend clip ends to compensate.

06 · The Best Text Pack for Resolve
Launch pitch for Lux pack — fix animation, customization depth, launch sale.

07 · Placing Text Behind a Subject
Cut footage, unlink audio, drag footage above text, Color page magic mask → alpha output → slight blur, shift all text up by Y to partially hide behind subject.

08 · Bringing Subtitles into a 3D Scene
Compound clip → Fusion clip → magic mask as garbage mat → camera tracker solve → export → image plane in 3D space → apply mode screen → mask-inverted drop shadow → magic mask subject back in front.

09 · Lux Title Pack Overview
Full walkthrough: Slide (position/size/blur/opacity/motion blur/apply mode/glow), Count (prefix/append/easing), Flicker (strength/duration), Follower (order/delay), Shine (offset/length/angle/width), Tracking, Write-On.

10 · Outro
Sign-off and Jesus sign-off.
Lines worth screenshotting.
- DaVinci's AI subtitle tool generates the most value not as captions but as a one-click word-cut-mark generator — use it for timing, then disable it.
- Animated text that starts at the exact moment a word is spoken will always feel late — pre-roll by 5 frames minimum to sync perception with audio.
- A 'Fix Animation' button that auto-truncates animation length to clip duration eliminates the most common caption-timing error with one click.
- You can composite text behind a subject entirely in DaVinci's Color page using magic mask plus an alpha output — no Fusion required for this use case.
- The DaVinci camera tracker's garbage-mat input accepts a magic mask directly — use it to exclude the moving subject from the track solve and get cleaner point clouds.
- Drop shadows break in screen-blend mode; the fix is to pipe the text layer into the shadow merge as an inverted mask so the shadow appears only where text edges are.
- Converting multiple text clips into a compound clip before entering Fusion keeps the text editable in the Edit page after the Fusion composition is built — edits propagate automatically.
- Setting subtitle max characters per line to 1 in DaVinci forces single-word cuts — this is the precision timing trick the whole workflow depends on.
- A text pack that goes from 7 presets to 200+ effective variations through combinatorial customization (position, size, blur, opacity, motion blur, apply mode, glow) has higher ROI than one with more preset slots.
- The follower animation's out-delay should be set to zero if you want all letters to exit simultaneously — leaving it non-zero makes individual letters stagger out, which rarely looks intentional.
The three layers of professional caption timing.
Getting word-synced captions to look professional in DaVinci Resolve is not one problem — it's three separate problems that must be solved in order: mark the timing, control the animation direction, and then fix the perceptual lag.
- DaVinci's AI subtitle tool can generate word-precise cut marks in seconds by setting max characters per line to 1 — use those marks as timing rails, then hide the subtitle layer entirely.
- Building per-word text stacks means one clip per word on the timeline — the word's cut mark sets the in-point, the next word's cut mark sets the out-point.
- Animation direction is set per clip in the animation tab; the key control is mirror-out, which defaults on and forces the exit to mirror the entry — turn it off to assign independent exit directions.
- The Fix Animation button auto-truncates animation length to fit any clip duration — it eliminates the single most common timing error without manual adjustment.
- All text clips should be shifted 5 frames earlier than the word's cut mark so that the animation resolves at the moment of speech, not after it — viewers perceive text that arrives late as out of sync.
- Placing text behind a subject can be done entirely in the Color page: magic mask the subject, add an alpha output node, drag the footage clip above the text in the timeline, and the subject composites over the text without touching Fusion.
- For 3D-tracked text, converting all text clips to a compound clip before entering Fusion keeps them editable in the Edit page after the Fusion composition is built — changes propagate automatically.
- A drop shadow breaks when the merge node is set to screen mode; route the text layer as an inverted mask into the shadow merge so the shadow renders only at text edges without making the text opaque.
- The garbage mat input on the camera tracker accepts a magic mask as a bitmap — erode the mask edge slightly to prevent imprecise subject selections from corrupting the 3D solve point cloud.
Terms worth knowing.
- Magic mask
- A DaVinci Resolve Color page tool that uses AI to automatically select and track a subject in a clip. When an alpha output node is added, it outputs a matte usable for compositing.
- Alpha output
- A node connection in DaVinci's Color page node graph that exports a clip's transparency mask as its output, allowing the masked subject to be composited over other layers.
- Camera tracker (Fusion)
- A Fusion node that analyzes a clip's motion to reconstruct a 3D camera path, enabling 2D text or graphics to be placed on image planes that follow the original camera move.
- Image plane
- A Fusion node that maps a 2D media input onto a flat surface in 3D space, allowing it to be positioned, rotated, and scaled relative to a tracked 3D scene.
- Compound clip
- A DaVinci Resolve container that collapses multiple timeline clips into a single clip without flattening them — contents remain editable by opening the compound clip in its own timeline.
- Fusion clip
- A clip type in DaVinci Resolve that opens in the Fusion compositing page, allowing node-based effects to be applied to the contained clips as a single unit.
- Mirror out
- An animation control in some DaVinci title presets that automatically creates the exit animation as a mirror of the entry animation — turning it off allows independent exit direction control.
- Garbage mat / garbage matte
- A mask input fed to a tracker to exclude regions of the frame from analysis — here used to prevent the moving subject from corrupting the 3D camera solve point cloud.
- Apply mode
- A blending mode control in some DaVinci title tools (and Fusion merge nodes) analogous to layer blend modes in Photoshop — screen, multiply, color dodge, etc. Screen makes text semi-transparent over footage.
Things they pointed at.
Lines you could clip.
“I don't actually use these subtitles. I just use the cut marks for my timing.”
“By the time the viewer actually sees the text, it's already past the word, and it's a little bit too late.”
“All of the animations are synced up with the audio really nicely.”
Word for word.
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
The bait, then the rug-pull.
Every editor who has tried to build word-synced animated captions in DaVinci Resolve knows the pain: placing markers by hand, wrestling with animation timing, and then figuring out how to get the text behind the subject without breaking anything. This tutorial dismantles that friction step by step — starting with a timing trick buried inside DaVinci's AI subtitle tool that most editors overlook entirely.
Named ideas worth stealing.
AI Subtitle Timing Trick
- Set max chars per line to 1
- Generate subtitles from audio
- Disable subtitle visibility
- Use cut marks as word-timing rails
- Build your own animated text to those marks
Repurposes DaVinci's free AI subtitle tool as a precision timing ruler rather than a caption generator.
5-Frame Pre-Roll Offset
- Select all text clips
- Shift left 5 frames (comma key)
- Extend clip ends to fill gap
- Animation completes at the word, not after it
A universal timing correction: animated text needs to start before the word is spoken so the motion resolves by the time the viewer hears it.
Text-Behind-Subject (Color Page Method)
- Cut footage clip at subject entry/exit
- Unlink audio
- Drag footage above text on timeline
- Color page: magic mask the subject
- Add alpha output node
- Right-click node graph → add alpha output
- Connect blue mask to alpha output
- Subject now composites over text
A Color-page-only approach to text-behind-subject that avoids Fusion entirely, suitable for clips that don't need 3D tracking.
3D Text Tracking Pipeline (Fusion)
- Convert text to compound clip
- Right-click → New Fusion Clip
- Magic mask subject → disconnect from output
- Add Bitmap node (garbage mat, inverted, eroded edge)
- Connect footage + bitmap to Camera Tracker
- Auto track → Solve → Export
- Image plane for text, connect to Merge3D
- Set Z/Y position in 3D space
- Apply screen mode to price text
- Add shadow node, connect camera tracker output as inverted mask
- Magic mask subject back in front
Full pipeline for locking animated caption text to a tracked 3D scene in DaVinci Fusion, with transparent blending and drop shadow.
How they asked for the click.
“Go check it out. It's just a double click install, super super simple. There's a launch sale for 3 days.”
Embedded mid-video after the core workflow is demonstrated, before the 3D tracking section. The product (Lux pack) is the tool used throughout, so the pitch lands as a natural extension rather than an interruption.






































































