AI video generation has exploded over the past year, and ByteDance's two offerings—JiMeng (即梦) for the Chinese market and Dreamina for international users—have become serious contenders against tools like Runway, Pika, and Kling. Both platforms run on the same underlying AI model, produce stunning videos from text prompts or reference images, and both slap watermarks on their output. But here's the question every content creator eventually asks: which one is harder to clean up?
If you've spent any time testing these tools, you've probably noticed that the watermarks aren't identical. They differ in placement, opacity, behavior during camera movements, and how they're embedded into the video stream. This comparison breaks down the technical differences, the real-world hassle of dealing with each platform's watermark, and—most importantly—how to get clean footage without re-encoding or quality loss.
---
Let's start with the basics. JiMeng (即梦) is ByteDance's AI video generation app tailored for Chinese users, available on iOS and Android. Dreamina is essentially the same product rebranded for international markets. The AI model, the generation pipeline, the output resolution—all identical. The watermark strategy, however, diverges.
JiMeng offers a feature called Camera Mode (出镜模式) where you upload a photo of yourself (or a video clip), and the AI generates a new video featuring your face or appearance in various scenarios. Think of it as a digital分身 that can dance, speak, or act in AI-generated environments.
The watermark in Camera Mode is particularly aggressive. It appears as a semi-transparent "JiMeng" logo overlaid near the bottom-right corner. But here's the catch: because the AI is animating your face and body, the background behind the watermark is constantly shifting. In videos with rapid camera movement, the watermark's position relative to the background changes frame by frame. This makes simple blurring or cropping ineffective—you'd lose too much of the frame.
For standard text-to-video or image-to-video generations (non-Camera Mode), JiMeng applies a lighter watermark that's easier to handle. But the Camera Mode watermark is where most creators struggle.
Dreamina, being the international version, uses a more subtle watermark. It's typically a small "Dreamina" text logo in the bottom corner, with lower opacity than JiMeng's. The international version also tends to place the watermark slightly further from the edge, giving you a bit more cropping room if needed.
However, Dreamina's watermark is still embedded in the output video file. You can't just toggle it off in settings. And because Dreamina videos often feature complex visual effects and motion, the watermark area can contain important visual information that you'd lose by cropping.
Here's what most comparison articles miss: both JiMeng and Dreamina store the original, unwatermarked video on their servers. The watermark isn't burned permanently into the pixels during generation—it's added as a layer when the video is delivered to your device. This means the clean version exists. The question is whether you can access it.
ByteDance's API returns the watermarked version by default, but the underlying CDN often hosts the original source file alongside the watermarked copy. Tools like OffWatermark work by fetching that original source directly, bypassing the watermark overlay entirely. This is fundamentally different from trying to remove a watermark from the video itself—it's about finding the clean file that already exists.
---
| Factor | JiMeng (即梦) | Dreamina |
|---|---|---|
| Watermark Type | Semi-transparent logo, heavier opacity | Lighter text logo, lower opacity |
| Camera Mode Watermark | Aggressive, positioned over dynamic backgrounds | Not applicable (Camera Mode not available in Dreamina) |
| Watermark Position | Bottom-right, close to edge | Bottom-right, slightly inset |
| Cropping Viability | Poor—too much frame loss, especially in Camera Mode | Fair—some cropping room, but loses content |
| Blur/Clone Stamp Viability | Poor—background movement makes cleanup inconsistent | Moderate—static backgrounds work, complex scenes fail |
| Source File Existence | Yes—original clean file on ByteDance servers | Yes—original clean file on ByteDance servers |
| Extraction Difficulty | Moderate (needs proper API handling) | Moderate (same backend, different endpoint) |
| Quality After Removal | 100% original quality when source-extracted | 100% original quality when source-extracted |
| Re-encoding Required | No (with proper tool) | No (with proper tool) |
| Time Investment | ~30 seconds with automated tool | ~30 seconds with automated tool |
---
Before diving into the solution, it's worth understanding why most watermark removal methods don't work well on JiMeng and Dreamina videos.
Blurring is the most common approach. Open any video editor, slap a blur effect over the watermark area, and call it done. The problem? AI-generated videos have incredibly detailed textures and lighting. A blur patch is immediately obvious—it creates a smudge that breaks the immersion. On Camera Mode videos where the background is constantly shifting, the blur patch also needs to track the watermark position, which adds complexity.
Cropping seems simpler. Just trim the bottom-right corner. But JiMeng's Camera Mode watermark is placed close enough to the edge that cropping removes 10-15% of the frame. On a 1080p video, that's a significant loss of visual real estate. Plus, many AI-generated videos have important action happening in the lower portion of the frame—dancing legs, moving objects, text overlays.
Clone stamping or content-aware fill works on static backgrounds, but AI videos often feature camera movement, particle effects, or lighting changes that make frame-by-frame cleanup impractical. You'd spend hours on a 30-second clip.
Screen recording is the worst option. It introduces compression artifacts, frame drops, and audio sync issues. The quality plummets.
The only reliable method is extracting the original source file from ByteDance's servers. This preserves 100% of the original quality, maintains the exact frame rate and bitrate, and requires zero editing skills.
---
OffWatermark handles both platforms (and their Chinese/international variants) through the same simple workflow. Here's exactly what to do:
For JiMeng (即梦) videos:
For Dreamina videos:
That's it. No app installation, no video uploading, no complex settings. The tool handles the server-side extraction automatically.
Important note: OffWatermark supports multiple platforms beyond JiMeng and Dreamina. If you're also dealing with watermarks from Douyin (抖音), TikTok, Kuaishou (快手), or Xiaohongshu (小红书), the same tool works for all of them. Just paste the share link from whichever platform you're using.
---
If you're using traditional methods (blur, crop, clone stamp), JiMeng's Camera Mode watermark is significantly harder to remove than Dreamina's. The dynamic backgrounds, aggressive logo placement, and constant motion make manual cleanup a nightmare. Dreamina's lighter watermark gives you more options for cropping or blurring without ruining the frame.
But if you're using source extraction (the correct approach), both are equally easy. They share the same backend infrastructure. The clean source file exists for both. The extraction process takes the same amount of time and effort. The result is identical quality.
The real variable is whether the tool you're using properly handles both platforms. Some watermark removers only work with Douyin or TikTok. Some don't recognize JiMeng links at all. OffWatermark was built specifically to handle the full ByteDance ecosystem, including both JiMeng and Dreamina, plus the Camera Mode variant that other tools often miss.
---
| Scenario | Winner |
|---|---|
| Manual removal (blur/crop) | Dreamina (easier) |
| Automated source extraction | Tie (both work identically) |
| Camera Mode videos | OffWatermark (only reliable method) |
| Batch processing multiple videos | OffWatermark (same workflow for both) |
| Maintaining original quality | OffWatermark (zero re-encoding) |
If you're generating AI videos for professional use, social media repurposing, or content creation, don't waste time with editing software. The clean source exists. You just need the right tool to pull it.
> Disclaimer: OffWatermark is an independent tool and is not affiliated with, endorsed by, or connected to ByteDance, JiMeng (即梦), Dreamina, Kuaishou (快手), or Xiaohongshu (小红书) in any way. All trademarks belong to their respective owners. Users are solely responsible for ensuring their use complies with applicable laws and terms of service. Only remove watermarks from videos you personally created.
← Back to OffWatermark