Review of PixaryAI

An important notice first: although PixaryAI advertises itself as an uncensored video generator, its own page expressly prohibits sexual content, celebrity faces, violence and political content. For the intent that brings readers to this catalogue, it does not apply.

Lawful use only with your own photos, or photos of adults who have given explicit consent. Responsible AI (in Spanish).

Every figure above comes from the provider’s own public pages on the date of the test. Prices change without notice, and the date is what makes the number readable.

Privacy and security (30%)

The page documents no privacy policy, no retention period and no deletion procedure for uploaded material, and it does not identify the operating company either.

Output quality (25%)

In its real use case — short concept clips, previews and stylised transitions — the result is respectable for open-source models.

Price and free tier (20%)

The free trial exists and is advertised as generous, but without numbers: no count of generations, no maximum resolution and no duration.

Speed (15%)

Generation in the browser with no local GPU worked with moderate queues in our session; clips came out in tens of seconds. There are no priority tiers.

Support and transparency (10%)

The seven-step flow is well explained and the input formats are documented. Points come off for the contradiction between the uncensored slogan and the stated restrictions.

The five weights are fixed across the whole catalogue and are set out in how we review. A tool is never marked against a rival, only against the same scale.

On 23/08/2026 the desk read the provider’s publicly visible pricing, payment and help pages, recorded every amount and limit into the table above, and ran the service through the same standardised scenarios used across the catalogue: well-lit frontal photographs at medium and high resolution, taken from our own test set.

What we record is what the provider states in public and what the interface does in front of us. On the free tier the first line we check is what the provider offers without payment, and here that reads: a browser-based ai video generator: text to video and image to video.

What we do not do: we never test these tools on photographs of third parties, we publish no results from such attempts if they reach us, and we do not estimate figures the provider leaves unpublished. An absence is recorded as an absence, which lowers the mark on the criterion concerned rather than being filled in with a guess.

Prices and limits change without notice. The date beside every figure is what makes it readable, and the desk re-checks the catalogue on a rolling basis — the next verification date will replace the one above.

• Rombach et al., High-Resolution Image Synthesis with Latent Diffusion Models — The diffusion inpainting this class of tool is built on.
• Kirillov et al., Segment Anything — The segmentation approach behind automatic garment masks.
• Diario Oficial de la Federación, Ley Olimpia reforms — The Mexican text on dissemination of intimate content, real or simulated, without consent.
• Regulation (EU) 2024/1689, the AI Act — Article 50 sets the disclosure duty for synthetic content.
• StopNCII.org — The international removal route for non-consensual intimate images.
• TAKE IT DOWN Act, S.146 — The United States takedown obligation, relevant where a provider is hosted there.

Prices, limits and retention come from the provider’s own public pages on the date of the test. Technical and legal claims rest on the published work above. Everything else is what the desk observed in its standardised scenarios.