Why Do Small Colored Letters Smear After Video Export While the Picture Stays Sharp?

Understand why color edges can fail first

Many video formats separate brightness-related information from color-difference information. Chroma subsampling reduces the spatial resolution of the color channels while retaining more luma detail.

FFmpeg's official pixel-format definitions identify YUV 4:2:0 as one chroma sample per two-by-two luma sample block, 4:2:2 as one chroma sample per two horizontal luma samples, and 4:4:4 as one chroma sample per luma sample.

Consequently, tiny red or blue lettering against a contrasting background can appear smeared even while the label's black outlines and surrounding product texture remain relatively sharp. This is especially relevant when the color distinction itself carries the letter shape.

However, chroma subsampling is not the only possible cause. Source focus, motion blur, resizing, compression quantization and the generated content itself can all damage lettering. The first task is to identify the stage where the colored edge deteriorates.

Compare identical label crops

Choose a stable frame where the approved product label is readable in the original. Keep the source file and exact frame identity. Compare the same crop in the original and exported versions at matched display size.

Inspect colored letters separately from black or neutral edges. If the black edges remain distinct but the colored boundaries spread, chroma resolution is a plausible contributor. If both luma and chroma edges become soft, investigate scaling, focus or general compression first.

Record each file's actual pixel format, chroma subsampling, dimensions and encoder settings. Do not infer a format solely from an MP4 filename. Different encoders and delivery platforms may use different internal formats even when the visible container extension is identical.

Where the editor and destination support it, create controlled test exports from the same source with different chroma formats while holding dimensions and other relevant settings as constant as possible. Some delivery systems will transcode the file again, so the local comparison and published result require separate checks.

Choose the appropriate information format

If the high-quality source retains the text and the delivery copy loses it, investigate the export's chroma format and compression path. A higher-chroma-resolution intermediate may help preserve color boundaries during editing, but it does not guarantee that the final platform will retain them.

If the original recording already has insufficient chroma resolution, converting it to 4:4:4 does not recreate the missing original color samples. Similarly, sharpening can create halos without restoring the true letter shapes.

When tiny colored label text is essential for the buyer, consider a verified still photograph or separate readable text in the product page rather than requiring a compressed moving image to carry every character.

Any reconstructed or generated lettering must be checked against the real product. A visually plausible word is not proof of the actual label.

Hypothetical case: red lettering on a white tube

Hypothetical export exercise; no video was encoded. A fictional skincare tube has a small red product name printed on white packaging. The original source is stipulated to preserve readable colored edges.

Candidate A is an 8-bit 4:2:0 delivery in which the small red letters are stipulated to lose boundary definition while neighboring black text remains more legible. Candidate B is a higher-chroma-resolution editing derivative from the same source. It is proposed for comparison, not claimed to restore the final published file automatically.

If B retains the red edges locally but the platform copy loses them again, the publishing derivative becomes the next investigation. If the source itself lacks the correct letters, neither export is acceptable evidence of their spelling.

Record your own observations

The filled rows are illustrative examples. Use the blank row for your own files and observations.

Download the editable CSV worksheet

What the references establish

• FFmpeg — Pixel Format Definitions: Primary technical definitions of YUV 4:2:0, 4:2:2 and 4:4:4 chroma sampling relationships. Supports the central mechanism.
• FFmpeg — Command-Line Tool Documentation: Primary context for encoded video processing and format selection. The red-lettering diagnostic is an original hypothetical example.