This new LTX tool helps you create your own AI video model

Despite recent advances in creative technology, AI video generators Still struggling with consistency. You can play the same prompt 100 times and you’ll probably get 100 clips that are slightly different. If you’re a filmmaker or studio trying to get AI content to have recognizable characters or mimic a distinct design style, you can’t have the AI ​​make the changes. That’s why AI company LTX built a framework, called LTX Trainer, to help you train models in your own style.

Instead of having an off-the-shelf AI model that replicates all kinds of artistic styles, you can use LTX Trainer to create your own AI model that is trained only on your stuff and creates content in your unique style. These custom, fine-tuned models can be built without the need for extensive coding, thanks to new updates rolled out today to GitHub, LTX co-founder Yaron Enger exclusively told CNET this week.

“It was very important for us to get this out there, so the community can police their own data,” Inger said in an interview. Creative companies that want to experiment with AI but need to protect their intellectual property can download the company’s publicly available tools and run models locally, avoiding the problems that come with uploading their IP to external networks or cloud servers, he said.

Watch this: WWDC26 Reality: Reactions to Siri AI

05:55

While AI video never really took off to the general public after that A crowd of AI descended — In peace, Sora -It is a booming business for professional creators and entertainment studios. Studio heads It was praised by Amnesty International Its ability to speed up workflow, while some creative people enjoy trying it. For professionals, the ability to customize an AI model opens new creative horizons beyond simply using mass-market regression generators that produce content that often looks the same.

But creative AI is controversial, as many artists are afraid of AI Their work is used without permission They can unethically replicate their methods and brands. That’s why companies like LTX and Adobe foot A way for companies to create custom AI models earlier this year, and they say they don’t train on their customers’ content. When open and trained LTX models run locally, it also keeps your data in your hands.

The trainer can now conduct audio-only training and combined audio-video training, expanding from its previous video-only capabilities. You can dial between methods, such as audio to video or image to video.

Creators can also customize their AI content to suit their unique style using adapters called LoRAs and IC-LoRAs. Essentially, these are your own custom, fine-tuned models that rely on LTX’s core global AI model to continually replicate your style across generations.

This process allows people to customize AI models without spending months and thousands of dollars, and companies like Google and Meta have the resources to do so. The new Proxy Assistant makes the creation process easier, so you can describe what you want to create in natural language.

You can also apply specific editing conditions to your custom forms. For example, you can create a model that reduces noise for each audio file it creates or upscales video clips to a higher resolution.

At the beginning of the year, LTX dropped its second-generation model Built in partnership with Nvidia. LTX-2 is different from other AI video models. First, it is designed to run on the device; Video creation using AI is a computing-intensive process It uses much more power than chatbotsThat’s why it usually runs in the cloud.

The LTX-2 is also an open weight model, which gives us an idea of ​​how it is manufactured and its functionality. (The open source AI will also share the content Experts say the model has been trained(which no major AI company wanted to disclose.)

Inger said the company’s ultimate goal with this update is to create a tool where you feel like your intellectual property is safe and you can experiment with artificial intelligence.

“These kinds of tools basically allow you as a creator to customize things exactly the way you want, while getting all the upsides of AI models that can generate things very, very quickly,” Inger said. “But then you can keep the IP address; it’s yours.”