Let me start with the conclusion: I've processed more than 30 videos with fixed station logos and corner marks using this workflow. After automatic removal, the original footage, duration, and audio track all remain fully intact—no cropping, no speed changes, no audio-video desync. That said, this only holds when the watermark position is fixed and the background isn't too complex.
My use case is organizing course screen recordings and interview footage. Many clips have a fixed station logo in the top-right corner and a fixed subtitle bar at the bottom. Over about two months, I processed more than 30 videos totaling over ten hours, ranging from 720p to 1080p, mostly MP4 with some MOV files. Below I'll share three things that worked reliably in practice, followed by two caveats.
Strength one: after removing a fixed watermark, the frame size and composition barely change. For example, a 1920×1080 interview video had a semi-transparent logo in the bottom-right corner. After processing, the export was still 1920×1080, with no cropping of the subject's shoulders or background edges. I zoomed in on the same frame before and after to compare: apart from the filled watermark area, the other pixels were virtually unchanged. Unlike some crop-based approaches, it doesn't enlarge the whole frame or trim the edges.
Strength two: duration and audio stay as they were, with no realignment needed. I tested a 12-minute-37-second course screen recording. The duration was exactly the same before and after, the audio waveform matched, and there was no offset between lip movements and sound. I batch-processed ten clips of different lengths and checked each export: duration error was zero across the board, and the audio wasn't re-encoded to some odd sample rate—it dropped straight into editing software and was ready to use.
Strength three: it's fairly stable for watermarks in fixed positions, with good consistency across repeated runs. Applying the same set of parameters to a series of videos shot from the same camera position produced consistent removal areas and edge transitions—one clip wasn't clean while another left a ghost. For a set of five lecture videos from the same scene with the logo in an identical position, I set the area once and ran the batch: none of the five finished clips showed obvious flickering or smearing in that region, saving the time of tuning each clip individually.
Limitation one: when the watermark sits on a face or high-frequency texture, the fill-in is more noticeable. In one case, a corner mark landed right on the pattern of a person's collar. After removal, that small patch of texture looked a bit blurry—acceptable in a still frame, but a slight smearing was visible once the camera moved. So if the watermark covers key subject detail, I'd suggest grabbing a frame to check before deciding whether to process it.
Limitation two: when the watermark moves or is semi-transparent over a complex background, automated removal becomes less stable. I tried a clip where the watermark drifted slowly with the frame—the fixed-area approach couldn't keep up, and residue appeared at the edges. In another clip, the watermark sat over swaying leaves, and removal left a slight ghosting in places. These cases require manual segment-by-segment work, and the batch workflow loses much of its convenience.
Who this suits: if your footage has watermarks in fixed positions—station logos, corner marks, subtitle bars—and the watermark doesn't cover a large area of a face or fine texture, this workflow is fairly hassle-free. It fits organizing scenarios like course screen recordings, interviews, and meeting notes. If the watermark moves or sits right on key subject detail, budget some time for manual touch-ups rather than expecting full automation in one pass. Different tools involve different trade-offs—just pick according to your footage's characteristics.