How to add readable captions for makeup

The approach

A useful makeup video gives the viewer something they can inspect. For a cream blush, that means: blend one swipe on a cheek. The focus here is how to make words legible without covering proof; it does not assume a particular tool can reproduce the real product perfectly.

Define the output

Choose one viewer question: shade and blending time. The clip should make the unfiltered blend in one take visible. For this format, make one production decision early: review the export on a phone-size screen. Decide what a pass looks like before generating: the item must match the source, the action must be readable, and the claim must be supported.

Work through the brief

1. Capture an accurate reference of a cream blush, with enough room around it for a vertical crop.
2. Frame the scene around a window-lit vanity and ask for caption placement. Describe the action in plain terms: blend one swipe on a cheek.
3. Write a short line that explains only what is visible. Read it aloud at the intended duration.
4. Generate one rough version. Watch it once at full speed, then pause where the product is closest to camera.
5. Record each defect, fix the largest one, and rerun only if the change can improve the final ad.

Review the result

The common failure here is text over hands or product. Check the product shape, text, hand placement, voice, and transition into the final frame. If the clip implies all-day wear without a real wear test, remove the line or replace the whole shot with real footage. Keep a note of the rejected versions; they matter when calculating the cost of a usable video. A convincing image alone does not prove an accurate product demonstration.

The VibesUGC blog has a reference-frame workflow and a cost worksheet for the next pass.