📺 S42 CutFlow Lower Third

Lower thirds are the name cards that slide in at the bottom of the screen to tell you who's speaking - the most tedious thing to build in a traditional editor because every one needs positioning, timing, and animation. S42CF_LowerThird is the S42 CutFlow node that renders them procedurally: name, title, style, timing, and slide-in/out animation, all as parameters on one node.

What you set

• name - the primary text (default "John Smith"). The headline.
• title - the secondary line (default "Lead Designer"). The descriptor.
• style - modern (colored accent bar + clean text, the corporate/tech look), minimal (text only, documentary/indie), broadcast (full background bar, news/TV), or cinematic (large text, subtle background, film credits).
• accent_color - hex, with a color-picker swatch in the widget. Defaults to a warm orange.
• start_frame - which frame the card appears on.
• duration_frames - how long it stays visible (default 72 = 3 seconds at 24fps).
• animate_in / animate_out - frames for the eased slide-in and slide-out (default 12 each).

Outputs are clip (your frames with the card burned in), frame_count, and info.

The workflow move

Drop it late in the chain - after color grade, after your subject's in the right place - and it stamps the card over whatever's there. To identify a speaker: put one LowerThird with the name/title for the first 3 seconds, another later for the next speaker, each with its own start_frame. Since it's all frame-index based, it lines up precisely with the rest of the pack's clip ops: trim first, then add the card, and the frames will be exactly where you expect.

Two habits that separate good from garish: keep animate_in/animate_out short (12 frames is a beat, not a production number - big values make it feel sluggish), and don't run the accent color hot over busy footage. The broadcast style's full background bar is the readability safety net when your background is chaotic; minimal is for when the footage is clean enough to let the text float.

Text is rendered with PIL - anti-aliased, with stroke and shadow support under the hood - so it's CPU-only and fast, with no font downloads or model dependencies. This is one of the four text nodes in the pack, and it's the one you'll reach for when someone's face needs a name attached.

Install

ComfyUI Manager → search "S42 CutFlow" → Install → restart. Or:

cd ComfyUI/custom_nodes/
git clone https://github.com/GeekyGhost/S42-CutFlow.git
pip install -r S42-CutFlow/requirements.txt

Restart fully; console prints [S42 CutFlow] Loaded 53 total nodes. The pack's only mandatory dependency is OpenCV (opencv-python-headless in the shipped requirements.txt; README's troubleshooting suggests opencv-python if you see "No module named 'cv2'").

Standing note: Geeky Ghost ships S42 CutFlow as a work-in-progress being refined for LTX Desktop. Text nodes are part of the stable core - the expanded experimental nodes are where the churn lives.