Luma Labs is making a major upgrade to its generative AI platform, Dream Machine. It is the largest since it was first announced earlier this year and includes a new image model called Photon and an entirely new way of working with AI video (more collaborative).

When the Dream Machine was launched in June, it was a significant step up from the previous generation model, offering faster generation, clips of 5 seconds minimum, and better motion. Later, features such as looping, prompt enhancement, and keyframing were added.

As competition intensifies and the release of OpenAI's Sora nears, Luma Labs is doubling down on Dream Machine, adding a variety of new features to make it easier to work with and turn ideas into reality. These include faster video generation and better understanding of natural language, allowing you to chat like ChatGPT.

The biggest change is Photon, a new text-to-image model built on Luma's architecture that is more personal, efficient, and creative. According to Luma, it is up to 800% faster than similar models and allows for accurate text rendering, easy character creation, and multiple image prompts.

Dream Machine is already an impressive AI video generator. It has several unique features that other platforms are only just beginning to catch up with, including the ability to set the first and last images that steer the video generation.

With the new update, Dream Machine gets a whole new user interface, and thanks to Photon, its text-to-video model is also much improved.

Available on both the web and iOS, Dream Machine will be more of a collaborator than a simple tool. It will be able to understand instructions and context, allowing people to work with the Dream Machine and bounce ideas off of it without having to learn prompt engineering.

Amit Jain, CEO and co-founder of Luma AI, says: “At Luma AI, we combine rigorous scientific research with an ambitious vision to help people think more visually, explore ideas more effectively, and share stories freely We help.”

The completely redesigned user interface makes it easy to unlock the full potential of the underlying AI video and image models. Instead of carefully crafting prompts for accurate output, users can select examples and converse with the models as if they were creative partners.

You can also give one or more images and use them to define the video output as well as set style, color, structure, and visual ideas. This improves character consistency and allows you to prompt with multiple character images to ensure they reflect what you have in mind.

According to Luma, Dream Machine currently has 25 million registered users, and the new update will also make it easier to see what other users are creating and remix their work.

One area where Luma Labs' Dream Machine struggles compared to other models is text-to-video conversion. The company states that it has solved this problem thanks to its all-new AI image model.

The Luma AI Photon will be built from scratch and will be available independently of the Dream Machine, allowing users to create impressive, realistic, and flexible images using AI. I have not tried it, but the samples the company shared seem to be as good as Flux, Midjourney, and Recraft.

A spokesperson explained that it is built on Luma's Universal Transformer architecture, which in tests runs 800% faster and cheaper than similarly sized models. The ability to generate images at high speed is important because in many cases multiple shots must be made to obtain the perfect image.

Anjney Midha, general partner of venture capital firm a16z and board member of Luma AI, said that design breakthroughs that enable new types of creative workflows are more than technical breakthroughs; they are real advantages in upgrading Anjney Midha, general partner and board member of Luma AI, stated that design breakthroughs that enable new types of creative workflows are more than technical breakthroughs; they are real advantages in upgrading.

“By combining fun product design, generative research, and software engineering, Dream Machine puts new creative power and capabilities into the hands of consumers and prosumers.

The Dream Machine will be offered at four different pricing levels. For hobbyists who enjoy a few videos and images per month, the price is $9.99. Each tier adds more generations each month, rising to $99.99 for those who use it at work.