An NSFW AI Video Generator That Adds No Watermark

Short answer: no watermark, on any plan, including the cheapest clip you can make. There is no watermarking step in the pipeline at all, which also means there is no upgrade that removes one - a 500-coin clip comes out of the system looking exactly like a 33,000-coin one.

That is the question most people arrive with, so it goes first. The rest of this is the arithmetic and the limits, because those are what actually decide whether this is useful to you.

Why “no watermark” is usually a paid feature elsewhere

The standard model is a free tier that stamps the output and a paid tier that does not. It works as marketing because the watermark is a visible reminder on something you already made and might have shared.

We do not have a free video tier at all, which is the honest reason there is no watermark question here. Video is paid per second from the first clip. That is worse if you wanted to try it for nothing, and better if you were going to pay anyway: there is nothing held back from the cheap end of the product.

What a clip costs, in real numbers

Pricing is per second of finished video, and it depends on the model and the resolution:

• Turbo models (fast, and what most people should use): 125 coins per second at 720p, 150 at 1080p.
• Full models (slower, more detail): 115 a second at 480p, 225 at 720p, 550 at 1080p, 1,100 at 4K.

So the common cases: a 4-second 720p turbo clip is 500 coins, 10 seconds is 1,250, and the longest allowed clip - 30 seconds - is 3,750. At the full model in 4K, that same 30 seconds is 33,000 coins, which is the honest reason to start with turbo.

Coins start at 500 for EUR 2.99, so a first 4-second clip is about two euros. Larger packages bring the per-coin price down by roughly a third, and coins never expire. The full breakdown including images is in the costs piece.

Four models, not one

Both text-to-video and image-to-video, each in a turbo and a full variant. That matters more than the naming suggests:

Image-to-video takes a still you provide and animates it, which is the way to keep a specific face or body consistent. Text-to-video starts from nothing but a description, which is faster to try and much harder to control.

Clips run 4 to 30 seconds, at 480p, 720p, 1080p or 4K depending on the model, with optional generated audio and an optional end frame that makes the render travel between two stills. The difference between the two starting points is covered in the comparison post.

What it will not make

Three things, without exception: anything sexual involving minors, anything involving age-play, and anything involving non-consent. Those are refused regardless of phrasing, and they are the reason everything else can be permissive.

Separately, the vendor model runs its own moderation and sometimes rejects a prompt we would have allowed. When that happens you get the coins back in full - a rejected or failed render is refunded automatically, and we have never charged for one. Of the video generations in the last week, the failures were entirely vendor moderation, and every one was refunded.

Two practical things worth knowing before you start

Storage is 30 days. After that the bytes are deleted automatically, along with any start image you uploaded. Download what you want to keep; there is no archive tier.

Commercial use is fine. Our Terms place no restriction on it and we claim no ownership of what you generate, so client work or a monetised channel is not a problem from our side. One honest caveat: our permission is not the same as you holding copyright, which for a purely machine-made clip is unsettled in several countries.

Where to actually start

A 4-second 720p turbo clip from a still you already have, for 500 coins. It is the cheapest honest test of whether the motion is what you wanted, and if it is not, you have spent two euros rather than committing to a 30-second render.

The studio is at Fellowi Video, on the same account and the same coin wallet as the image generator. If you want the full walk-through of producing one clip end to end, mistakes included, that account is here. And the broader piece on what an uncensored video generator does and does not allow is here.