Uncensored AI video generator online, image to video
LU Cloud runs fourteen image-to-video models that render uncensored content, and they only unlock after you change the content policy setting in your own account. Every one of them starts from a picture you supply, none of them turns a text prompt alone into a clip, and every clip is five seconds long. The lowest price on the list is 10,000 credits a clip, so the 19 EUR Hosted plan with its 900,000 monthly credits is worth up to 90 clips.
The awkward parts are in there on purpose. We build Locally Uncensored, the free open source AI studio that runs models on your own machine, and the hosted layer at LU Labs exists for people whose machine cannot hold a video model. Renting GPUs by the clip costs real money, so we would rather you knew what you were buying first. What uncensored AI means on either side of that line is set out on its own page.
Fourteen image-to-video models that render uncensored content
All fourteen animate the image you hand over, so the table is fourteen ways to do one job at different prices. Twelve of them carry Uncensored in the name. OpenVideo and OpenVideo + Style LoRA do not, and they are open all the same, because the catalog marks a model open in a field rather than in its name. Wan 3.0 Uncensored and Wan 3.0 Prime Uncensored take a second picture as the end frame, so you can say where the five seconds should land. The numbers below come from the same file the app bills against, and each plan column is the number of clips you get if you spend that plan's entire monthly wallet on that single model. They are alternatives, not fourteen separate allowances, and the same wallet also pays for chat and for images.
| Model | Credits per clip | Hosted | Pro | Max |
|---|---|---|---|---|
| LTX 2.3 Uncensored | 10,000 | 90 | 235 | 500 |
| OpenVideo | 10,000 | 90 | 235 | 500 |
| Wan 2.2 Uncensored | 15,000 | 60 | 156 | 333 |
| OpenVideo + Style LoRA | 15,000 | 60 | 156 | 333 |
| MiniMax H3 Uncensored | 20,000 | 45 | 117 | 250 |
| Wan 3.0 Uncensored (takes an end image too) | 25,000 | 36 | 94 | 200 |
| Seedance 1.5 Pro Uncensored | 26,000 | 34 | 90 | 192 |
| Vidu Q3 Uncensored | 35,000 | 25 | 67 | 142 |
| Wan 3.0 Prime Uncensored (takes an end image too) | 37,500 | 24 | 62 | 133 |
| Wan 2.6 Uncensored | 50,000 | 18 | 47 | 100 |
| Wan 2.7 Uncensored | 50,000 | 18 | 47 | 100 |
| Seedance 2.0 Fast Uncensored | 50,000 | 18 | 47 | 100 |
| Seedance 2.0 Uncensored | 60,000 | 15 | 39 | 83 |
| Seedance 2.5 Uncensored | 90,000 | 10 | 26 | 55 |
Wan 2.2 Uncensored Extend is not in that table, because it does not start a clip. It lengthens one you already rendered, so it is a second paid run on an existing file.
| Tool | Credits per run | Hosted | Pro | Max |
|---|---|---|---|---|
| Wan 2.2 Uncensored Extend | 15,000 | 60 | 156 | 333 |
Picking down the list is the main lever you have on cost. Seedance 2.5 Uncensored at 90,000 credits costs what nine LTX 2.3 Uncensored clips cost, so it is worth saving for a shot you already framed on a cheaper model.
The clip table on our pricing page lists the same fourteen models, with the extension tool beside them, so the two pages agree line for line. The rest of the hosted video catalog is there as well. It includes text-to-video models, among them Wan 2.2 720p, Wan 2.2 Fast, LTX-2 (with audio), HunyuanVideo 1.5 and LTX 2.3, which take a prompt with no picture. None of those is one of the fourteen.
Image in, five seconds out: how the workflow runs
- Open lu-labs.ai and sign in. It runs in the browser, so a Mac or a phone works as well as a PC.
- Buy a credit pack or a Hosted plan. Packs start at 15 EUR and need no subscription.
- Go to Settings, General, Content policy and move the setting off its default. Signed out, that section only offers you a sign-in link.
- Get your input image. Either upload one you already have, or generate it in LU first on one of the seven image models that render uncensored content. Generating it in LU spends credits from the same wallet, from 1,000 credits an image on Neta Lumina (anime) Uncensored, Z-Image Base and Nucleus up to 2,500 on Wan 2.2 Realism, and that is on top of the clip price.
- Open the Animate lane, hand over the image, pick a model from the table above and write what should happen in the frame.
- Start the job. What comes back is five seconds long. Download it, because hosted renders are held for seven days and then removed.
- Optional: run Wan 2.2 Uncensored Extend on the result for another 15,000 credits to make it longer.
Step four surprises people, so plainly: the clip price does not include the input image. Make that image in LU and you pay twice out of one balance, so a session of ten stills and three clips costs more than the video table alone suggests. Exact image quantities are on the image generation page.
How the content policy setting works
Hosted image and video generation follows a setting that lives in your account, not a switch we flip for everybody. Settings, General, Content policy has three options, and the line above them reads "Applies to images and video rendered in the cloud. Text is unaffected, and nothing on your own machine is." Strict is described in the app as "The tightest filter we have. Choose this if others use your screen." Standard is "The default for every account." Off is "No filter beyond the legal limits below."
Those legal limits are printed under all three options and they do not move: "Material involving minors, and photographs of real people uploaded without their consent, are refused on every request whatever this is set to, and reported." Content involving minors is blocked everywhere, always, on every model, with no exception and no setting that changes it. A photograph of a real, identifiable person needs that person's consent before you upload it. The Terms of Service apply on top of both.
We describe hosted text differently, because it works differently: on the 47 chat models we add no moderation layer of our own. Video is not text and we are not going to call it filterless. The account setting is the honest description, and the full walkthrough of it is in the cloud handbook chapter.
What a month of clips costs
One wallet covers chat, images and video. Plans refill it monthly; packs are bought once and do not expire.
| Option | Credits | Clips at 10,000 credits |
|---|---|---|
| Hosted, 19 EUR a month | 900,000 | up to 90 |
| Hosted Pro, 49 EUR a month | 2,350,000 | up to 235 |
| Hosted Max, 99 EUR a month | 5,000,000 | up to 500 |
| 15 EUR pack | 450,000 | up to 45 |
| 35 EUR pack | 1,100,000 | up to 110 |
| 75 EUR pack | 2,400,000 | up to 240 |
| 125 EUR pack | 4,050,000 | up to 405 |
| 300 EUR pack | 9,750,000 | up to 975 |
Annual billing is 190 EUR on Hosted, 490 EUR on Pro and 990 EUR on Max. Every pack is available from your first purchase, so you do not have to climb a ladder to reach the 300 EUR one. Packs do not carry the daily Flash chat allowance that the plans include. Anything you want to check against the live catalog is on our pricing page and on the LU Labs pricing page.
On a plan, the same EUR 15 pack gives 710,000 credits or more, because every pack uses your plan's rate.
The honest limits
- Clips are five seconds. Extending one is a separate paid run.
- The uncensored models have no text-to-video path. If you have no picture, you have to make one first, and that costs credits too. Wan 3.0 Uncensored and Wan 3.0 Prime Uncensored can take a second picture as the end frame, which is another image to find or pay for.
- Hosted renders are kept for seven days, then removed. Download what you want to keep.
- Catalog prices change. The numbers on this page were checked on September 22, 2026, and the machine-readable copy lives at cloud-facts.json.
- A failed job can still cost you, and quoted maximums assume every run succeeds. Real months land under the table figures.
- Media jobs are limited to 30 requests a minute per account, so a batch queues rather than all starting at once.
- A hosted request leaves your machine. That is what hosted means, and no phrasing changes it. If it is unacceptable for your work, the local app is the answer.
- The catalog is curated. You cannot upload your own video checkpoint to our GPUs.
Online or on your own GPU
If you own the hardware, none of the above applies to you and none of it costs anything. The desktop app is AGPL-3.0, version 3.0.5, Windows and Linux, and it runs whatever video model you install with no credits and no policy setting. The tradeoff is VRAM and patience instead of money.
We wrote the local route up three times over, at different depths: uncensored AI video locally for the overview, how to generate AI videos locally for the step by step, and AI lipsync and video extend local for the follow-up operations. The hosted side is summarised on the cloud page, the text lane has its own page at uncensored AI chat online, and the endpoint behind it is on the API page. What uncensored AI means, local and online covers the word itself on both sides.
Compared with browser video tools
Venice.ai is the comparison people ask for most. On venice.ai/pricing, read on September 22, 2026, video is billed through their credit system, 100 credits are worth 1 USD, and clips are listed at under 80 credits. That is the whole of what we will claim about their video pricing, because it is the whole of what we read on the page. A fuller side by side of the two services, plans and catalogs included, is on our Venice comparison.
The structural difference is easier to state than the price one. Our uncensored models need an image, so the comparison is never purely clip against clip: a prompt-only tool does a job ours does not.
FAQ
Is the video generator uncensored?
The uncensored video models appear once you change the content policy setting in your own account. The default setting is Standard, and on Standard some prompts are refused. We do not describe hosted video the way we describe hosted text, where we add no moderation layer at all. Two lines never move at any setting: material involving minors, and photographs of real identifiable people uploaded without their consent.
Can I type a prompt and get a video, with no image?
Not on the uncensored models. Every one of them is image-to-video only, so the render needs a still to start from. LU has text-to-video models in the catalog, such as Wan 2.2 720p and LTX 2.3, but those are not the uncensored ones.
How long are the clips?
Five seconds. Wan 2.2 Uncensored Extend adds length to a clip you already rendered, and it costs 15,000 credits each time you run it.
Does it do uncensored video?
Yes, on the fourteen uncensored models listed above, and only after you set your account content policy to Off. All fourteen render a clip from an image you supply. Wan 2.2 Uncensored Extend is a separate tool that lengthens a clip you already made, and it is not one of the fourteen.
Can I animate a photo of a real person?
Only with that person's consent. Uploading a photograph of a real, identifiable person without their consent is refused on every request whatever the content policy is set to, and reported.
Do you keep the clips?
Cloud renders are held for seven days, then they are gone. Download anything you want to keep. There is no training on user data and it is never sold, and accounts and billing are in the EU.
Does it run on a Mac or a phone?
Yes, in the browser at lu-labs.ai. The free desktop app is Windows and Linux only, so on a Mac or a phone the hosted studio is the route.
What is blocked no matter what I set?
Material involving minors, on every request, on every model, with no exception. Photographs of real people uploaded without their consent, the same way. The Terms of Service apply on top of both.