LTX 2.3 Pro is one of only six models on Picasso IA confirmed to generate NSFW video with no content filter, and it renders straight from a text prompt, no starting image required, up to 4K at 50 frames per second. Vidu, the well regarded platform from Shengshu Technology, has real strengths of its own, multi reference character consistency and native lip sync, but its policy on mature prompts has never been tested. See what LTX 2.3 Pro actually generates, five steps to a finished clip, and an honest read on Vidu.
Type a suggestive scene into Vidu and the result is often softened, rewritten, or blocked outright, with no explanation of what in the prompt triggered it. LTX 2.3 Pro, built by Lightricks and sitting inside Picasso IA's own catalog, is one of only six models across the whole platform confirmed to generate adult video with no content filter standing between a written scene and the finished clip. It renders straight from a text prompt, no starting photo required, up to 4K resolution at 50 frames per second.
This page covers why LTX 2.3 Pro earns that confirmed status, what actually makes a cinematic NSFW clip work when there is no source image to lean on, a five step walkthrough to a finished video, an honest look at where Vidu stands on the same question, and the rules that apply no matter which platform renders the scene.
LTX 2.3 Pro is a text-to-video model listed at LTX 2.3 Pro. Describe a scene, a fictional character, and a camera move, and it renders the clip directly, up to 4K resolution at 50 frames per second, adult and suggestive scenes included, nothing rewritten or quietly softened before the render starts. A still image or an audio clip can also seed a generation when one already exists, but unlike an image-to-video model, LTX 2.3 Pro never requires one.
That confirmed status is not something LTX 2.3 Pro holds alone. Here is the complete list of six models Picasso IA has actually tested and confirmed run with no content filter, by what each one generates.
| Model | Generates | Link |
|---|---|---|
| Seedream 4.5 | Text-to-image | Seedream 4.5 |
| Qwen Image 2 | Text-to-image | Qwen Image 2 |
| Recraft V4.1 Pro | Text-to-image | Recraft V4.1 Pro |
| Picasso Image Editor Pro | Text-to-image and editing | Picasso Image Editor Pro |
| Picasso Video | Image-to-video | Picasso Video |
| LTX 2.3 Pro | Text-to-video | LTX 2.3 Pro |
For video specifically, that gives two confirmed starting points rather than one. LTX 2.3 Pro is the pick when nothing exists yet and a scene needs to be written straight from a blank page. Picasso Video is the pick when a still image is already in hand and only needs to move. Vidu, worth saying plainly, sits outside that list entirely: a separate product built and moderated by its own company, with its own account system, tested on its own platform, not on ours.
Because LTX 2.3 Pro renders a scene from nothing but text, the entire clip, the setting, the characters, the camera, the mood, has to live in one prompt. That is the opposite of an image-to-video model, where the composition is already locked in and only the motion needs describing. Here, the composition is exactly what needs describing.
Camera language earns its keep more than almost anything else. A wide establishing shot of a soundstage, a slow dolly toward a subject, a close-up on a hand fastening an earring: naming the shot type gives the model something concrete to frame, rather than leaving the camera position to chance. Setting does similar work, since a backstage dressing room, a lit vanity mirror, or a poolside deck on an outdoor set each tells the model where the light comes from and what fills the background, before a single character is even described.
Wardrobe and mood carry the rest. A silk robe, a black gown, a red dress catching sunlit window light: naming the fabric and the color gives the model something concrete to render, and it reads as intimate and cinematic rather than plain. A vague word like alluring describes an effect, not a garment, and leaves the actual look to chance.
Resolution and frame rate are both adjustable up to 4K at 50 frames per second, and credit cost scales with whichever combination gets picked. A shorter, lower resolution draft is the efficient way to confirm a scene and a camera move read the way intended, before spending more credits on the full resolution version of the same clip.
Building a finished, uncensored clip from a blank prompt follows the same five steps every time.
Vidu, the video platform from Shengshu Technology, deserves a fair hearing rather than a dismissal. Holding one character's identity across several reference images at once, and rendering dialogue, ambient sound, and lip sync together with the video in a single pass, are both real strengths, and its anime leaning, stylized motion is rated well by reviewers who test across platforms. For a recurring character built from an existing set of reference photos, that workflow does something LTX 2.3 Pro was not built to do.
None of that answers the question this page set out to cover, though. Vidu is a separate platform, built and moderated by its own company, and nothing here can honestly confirm how its policy treats a mature or suggestive prompt, since that has not been tested on Vidu's own service. Assuming it behaves one way or another based on reputation alone is a guess, not a fact, and a guess is a poor thing to spend a generation on. Vidu is genuinely worth trying for its reference-driven strengths; it is simply not, on the record here, a confirmed uncensored pick.
That is exactly the gap LTX 2.3 Pro closes for a scene written from scratch. There is nothing to test and nothing to guess at, since it is already one of the six models Picasso IA has confirmed generates a mature scene directly, at up to 4K resolution, straight from a prompt.
Everything on this page is for adults, eighteen and over, and every character in a generated clip, on LTX 2.3 Pro or anywhere else in the catalog, has to be a fictional adult. No real person's likeness, no public figure, and no one resembling a real individual used without that person's consent. A minor, or anyone made to look like one, is never an acceptable subject, in any style, and a prompt attempting either is refused and treated as a violation of the platform's terms.
Privacy holds by default, too. A clip generated on LTX 2.3 Pro stays tied to the account that made it and is never shown to another user or published anywhere unless that account chooses to share it. Confirming a model has no content filter changes what kind of scene it will render, not who it will render or who else gets to see the result.
LTX 2.3 Pro is priced dynamically, by the resolution and duration actually chosen for a given clip, rather than a single flat fee that applies no matter what settings are picked. Two people writing the same scene can land on two different costs simply because one picked a longer, higher resolution pass and the other did not, which is expected rather than a pricing error.
New accounts start with free credits, enough to run the five steps above and confirm a scene works at a lower resolution before spending on a longer or sharper version. Current numbers, plan tiers, and anything active for new signups live on the pricing page, kept current there rather than printed here where a figure would eventually go stale.
Yes. LTX 2.3 Pro is one of six models in Picasso IA's catalog confirmed to generate adult video content directly, with nothing blocking or softening a mature prompt before the clip renders.
No. LTX 2.3 Pro renders straight from a text prompt with no source image required, though a still image or an audio clip can also seed a generation when one already exists.
No. Vidu is a separate video platform built by Shengshu Technology, with its own app, its own account system, and its own moderation. It is not one of the models inside Picasso IA, which is why this page compares the two products rather than describing a switch between them.
That is genuinely not confirmed here. Vidu's moderation policy is set and enforced on Vidu's own platform, and Picasso IA has not tested a mature prompt against it. Whether it filters, softens, or allows a given scene is a question only Vidu's own service can answer.
Up to 4K resolution at 50 frames per second, adjustable down for a cheaper, faster draft. Credit cost scales with whichever combination gets chosen for the final render.
Retake redoes one part of an already generated clip, a hand, a turn, a piece of fabric, without starting the whole scene over. Extend lengthens a clip that already reads correctly, so a finished scene can run longer without a second, separately written prompt.
Yes. Alongside text and image starting points, LTX 2.3 Pro supports audio-to-video generation, turning an existing audio clip into a matching video.
No, on any model, confirmed or not. Every character has to be fictional and adult, eighteen and over. Real people, public figures, and anyone resembling a real individual without their consent are never acceptable subjects.
It is priced dynamically by the resolution and duration chosen, so there is no single flat number. New accounts get free credits to start, and current, exact pricing lives on the pricing page.
Five others: Seedream 4.5, Qwen Image 2, Recraft V4.1 Pro, and Picasso Image Editor Pro for images, and Picasso Video for image-to-video. Together with LTX 2.3 Pro, that is the complete list of six models Picasso IA currently confirms run with no content filter.
Choose your language
