Video Watermarker

{{ watermarkKindLabel }} {{ positionLabel }} Local only

Source video:

One video up to 40 MB, 2 minutes, and 2.1 megapixels. Output is a re-encoded MP4.

{{ selectedFile ? selectedFile.name : 'Drop a video here or browse' }}

{{ selectedFile ? sourceMetaLine : 'MP4, MOV, WebM, or MKV that this browser can decode' }}

{{ sourceChecking ? 'Reading video metadata…' : sourceStatusLine }}

The demo clip is generated in this browser and uses the same encoder path.

Watermark type:

Choose short text or one PNG, JPEG, or WebP logo.

Watermark text:

Enter 1–60 visible characters. A dark outline keeps light text readable.

Logo image:

One PNG, JPEG, or WebP up to 5 MB and 4096 × 4096 pixels.

{{ logoFile && logoWidth ? `${logoWidth} × ${logoHeight} · ${formatBytes(logoFile.size)}` : 'Transparent images are preserved' }}

{{ logoChecking ? 'Reading logo…' : (logoError || 'Held in memory for this page only') }}

Position:

Choose one of nine stable frame anchors and confirm it in the preview.

Watermark width:

{{ size_percent }}%

Set the mark from 5% to 40% of the frame width.

Opacity:

{{ opacity_percent }}%

Set visible strength from 10% to 100%.

{{ processingProgress }}%

Rotation:

The neutral default is 0°. Use −45° to 45°.

{{ row.label }}{{ row.value }}

The chart renderer is unavailable. The same placement values remain in the render plan.

Signal	Value	Meaning	Copy
{{ row.label }}	{{ row.value }}	{{ row.detail }}

A visible watermark gives viewers context that travels with a video. A creator may add a logo for attribution, a short copyright notice before publishing, or a review label to distinguish an approval copy from a final cut. Because the mark is rendered into every frame, it remains visible when the file is renamed or separated from its original post.

Visibility is a compromise. A small, faint corner mark interferes less with the picture but is easier to crop or miss against detailed footage. A larger or central mark is harder to overlook, yet it can cover faces, captions, product details, or other important content. Position should therefore be judged across the whole clip, not from a convenient opening frame.

Text marks suit proof labels, short ownership notices, and version names. A contrasting outline helps light text remain readable over changing scenes.
Logo marks preserve a brand shape. Transparent PNG or WebP artwork usually blends more cleanly than a JPEG with a fixed rectangular background.
Corner placement protects the main subject in many compositions, while centered placement provides a stronger deterrent for review copies.

Watermarking changes the video stream. Each frame must be decoded, composited with the mark, and encoded again, so the output can differ in file size, fine detail, and color sampling even when the frame dimensions stay the same. Audio can also depend on the available encoding path.

A visible mark is attribution or deterrence, not copy protection. It does not encrypt the file, prevent screen recording, prove authorship by itself, or stop determined removal. Keep original project files and publication records when provenance matters.

How to Use This Tool:

Choose the video and mark first, inspect placement over representative scenes, then leave the page open while every frame is re-encoded.

Choose one browser-playable MP4, MOV, WebM, or MKV file in Source video. It must be no larger than 40 MB, two minutes, or 2.1 megapixels per frame, with even pixel dimensions.
Select Text and enter 1 to 60 visible characters, or select Logo image and choose one PNG, JPEG, or WebP file up to 5 MB and 4096 by 4096 pixels.
Choose one of the nine Position anchors. Set Watermark width from 5% to 40% of the frame width and Opacity from 10% to 100%.
Scrub the source preview to scenes with bright areas, dark areas, captions, faces, or edge action. If needed, open Advanced and rotate the mark from −45° to 45°, then confirm that it still fits inside the frame.
Apply the watermark and keep the page open until processing reaches 100%. Play the complete Watermarked video, confirm the mark in Frame coverage, and check the output encoding and audio note in Render plan before downloading.

Interpreting Results:

Successful encoding confirms that the output MP4 can be read and that its frame dimensions and duration remain close to the source. It does not confirm that the watermark stays legible or avoids important content. Watch the rendered file from beginning to end and listen for audio before publishing it.

Frame coverage shows the rotated mark's rectangular footprint and the pixels before and after it on each axis.
Mark bitmap is the unrotated text or logo size. Rendered footprint includes the extra width and height created by rotation.
Safe margin is three percent of the shorter frame dimension. It keeps edge anchors inset, but it cannot predict platform overlays, player controls, or later cropping.
Output encoding identifies MP4 with H.264/AVC and states whether audio was preserved or the compatibility path produced video only.

Technical Details:

Video watermarking is frame compositing followed by a complete video encode. Text or logo artwork is first rendered into one transparent bitmap at the chosen opacity and rotation. That bitmap is placed at a fixed pixel coordinate over every source frame, and the combined frames are written to a new MP4.

Formula Core

The requested width is a percentage of the source frame width. Height preserves the text or logo aspect ratio. Each dimension is rounded to a whole pixel and never falls below one pixel.

w_{m} = \max (1, round (w_{f} \times \frac{p}{100}))

h_{m} = \max (1, round (w_{m} \times \frac{h_{i}}{w_{i}}))

Here w_f is frame width, p is Watermark width in percent, w_i and h_i are the mark's intrinsic dimensions, and w_m and h_m are its unrotated pixel dimensions.

Rotation expands the axis-aligned rectangle needed to hold the mark. The absolute sine and cosine terms make positive and negative angles produce the same footprint dimensions.

w_{r} = ceil (| w_{m} \cos (θ) | + | h_{m} \sin (θ) |)

h_{r} = ceil (| w_{m} \sin (θ) | + | h_{m} \cos (θ) |)

margin = round (0.03 \times \min (w_{f}, h_{f}))

θ is the absolute rotation angle in radians. Edge anchors place the rotated footprint one margin from the chosen sides; center anchors round half of the remaining horizontal or vertical space. A plan is rejected when the rotated footprint plus two margins is larger than the frame. Opacity is the chosen whole percent divided by 100.

Transformation Core

Video watermark rendering and encoding stages
Stage	Behavior	Output consequence
Prepare mark	Logo pixels retain transparency. Text is drawn in bold white with a black outline, then opacity and rotation are applied once to the mark bitmap.	The same rendered mark is reused for the complete clip.
Overlay	The mark bitmap is composited at the calculated `x` and `y` coordinates on every frame.	Position, size, opacity, and rotation remain fixed over time.
Primary encode	Video is encoded as H.264/AVC in YUV 4:2:0 with constant-rate factor 23. The first audio stream, when present and supported, is encoded as AAC at 128 kbit/s.	The output is a web-oriented MP4 with fast-start layout and stripped source metadata and chapters.
Compatibility encode	When the primary browser path is unavailable, frames are sampled at 12 frames per second and encoded as H.264 video without audio.	Motion may be less smooth and the result is explicitly labeled video only.
Verify	The finished MP4 is decoded again. Its dimensions must equal the source, and duration may differ by no more than 0.2 seconds or 8%, whichever allowance is larger.	A file that fails either check is not offered as a completed result.

Input and Placement Rules

Video and watermark validation limits
Input	Accepted range	Boundary behavior
Video	At least 0.2 seconds and at most 120 seconds; no more than 40 MB; each side at most 1,920 pixels; at most 2,073,600 pixels per frame.	Width and height must both be even for H.264 output.
Logo	PNG, JPEG, or WebP from 32 bytes to 5 MB; each side at most 4,096 pixels.	JPEG has no transparent background; PNG and WebP may retain alpha.
Size and opacity	Width 5% to 40% inclusive; opacity 10% to 100% inclusive.	A valid percentage can still be rejected if rotation makes the mark exceed the safe frame area.
Rotation	−45° to 45° inclusive, in whole degrees.	Positive angles rotate clockwise around the mark center.

Changing any watermark setting invalidates the prior output because placement and encoding must agree with the displayed plan. The output dimensions match the source, but re-encoding is lossy and does not preserve source metadata, chapters, subtitles, or non-primary audio streams.

Privacy and Limitations:

The source video and logo stay on the device and are processed in browser memory; they are not uploaded for watermarking. An internet connection may still be needed to load the browser encoder code before processing begins.

Keep the tab open during encoding. Closing it or cancelling processing discards the in-memory job.
Expect a lossy H.264 re-encode rather than a bit-for-bit copy of the source.
Check the output audio note. The compatibility path intentionally produces video without audio.
Do not rely on a visible watermark as digital rights management, forensic identification, or legal proof of ownership.

Worked Examples:

Client review copy

A centered text mark such as “DRAFT — NOT APPROVED” at moderate opacity remains obvious across changing scenes. Scrub to shots with faces and subtitles before encoding, then confirm the message stays readable without covering the exact detail the client needs to review.

Corner logo for publication

Use a transparent logo at a corner anchor and start near 20% frame width. Test both bright and dark scenes, because transparency and low opacity can make the mark disappear. After encoding, verify playback and audio rather than accepting the placement preview as proof of the final file.

References:

FFmpeg Filters Documentation: overlay, FFmpeg Project.
ffmpeg.wasm usage guide, ffmpeg.wasm project.