We are thrilled to announce the latest revolution in image generation, our newest model, Ikon-1.  One of the core challenges that exist today is that it’s hard to get the results you really want.  To put it simply, the image generators of today do not fully understand what you want to see. You find yourself entering a never ending cycle of generating images while staying unsatisfied with your result –settling for the output that is “most serviceable”. Today, that cycle is initiated because there’s really only one primary control: text. This is an extremely limiting prompting method that prevents the use of synthetic media in a wide range of use-cases.

Beyond Text: Unleashing New Levels of Creativity with Ikon-1

Ikon-1 offers a suite of multimodal controls –6 in total.  Today we offer pose, color, structure (depth), sketch (softedge), base-image (img2img), and of course, text. The controls can also be fused together! Of these, the most popular are pose image and sketch image.

In the pose example above, the generation is accurately controlled by my input asset.  The level of controllability is so thorough that details such as finger position and head tilt are controlled in the generation. This is all made possible with Ikon-1’s ability to extract the pose of an object and directly use that information to prompt a generation. 

In the sketch example, the shoe sketch distinctly controls the generation. Details like the shape of the collar and the angle of the tongue are precisely carried over from the drawing to the generation output. All of these details are visualized by the addition of a text prompt, bringing the sketch to life. 

With multimodal controls, like the ones discussed above, generations are more controllable than ever. An advertiser, who wants their bracelet displayed by a model in a certain hand position, can accomplish this using pose control. A storyboard artist can quickly visualize what multiple characters look like in a landscape –all assuming a specifically controlled pose. A product designer can quickly ideate upon their design using their own work as a sketch control. With these features, a first time visitor or professional designer can leverage multimodal controls to arrive at the precise outputs they want.

NEX for Enterprise

We’re also excited to roll out tailor-made solutions for enterprises. At its core, NEX Enterprise allows you to  create your very own custom AI models that capture the essence of your brand's IP.  Accompanied by the ability to collaborate on artboards, manage users in a workspace, and share assets, NEX empowers teams to do their best work.  

A common challenge faced by many image generation tools is the inconsistency in character depiction throughout generations. Prompts such as: “a cinematic photograph of a heroic robot fighting to reclaim its home in a vast futuristic cyberpunk landscape” and "a wide shot of a robot sailing the seas to find its home” would likely produce images of two distinctly different robots. However, with a NEX Enterprise custom model, companies can instantly create IP-consistent content. Imagine drafting a storyboard in minutes which includes a specific character in various scenes, poses, and styles with just the click of a button.

More to come

We’ve released a lot of new features and functionality this month but have much more in store. Fundamentally, we believe in the empowerment of creatives today with our core functionality revolving around controllability through assets of your choosing.  If this resonates with you, please send a message to us at outreach@nex.art!