If you're old enough, you'll remember going to your local strip-mall Blockbuster to load up on VHS tapes before the weekend. Or thumbing through a phonebook to get the number of a business that you hand-dialed to find out what time they closed. Or reversing the intricate origami of an interstate map, spreading it across your kitchen table, and using a Sharpie to plan a road trip.

You might also remember the moments you left those rituals behind, and how quickly you knew you were never going back.

Image editing is about to experience such a moment.

When we founded Reve, we did so with a single unifying principle: creative intent over technical proficiency. Humanity's digital transformation was clearly a democratizing moment, but if you trace the evolution of industry-standard creativity and productivity tools from their origins to where they are today, you will discover that, over the decades, they have reacquired most—if not more—of the complexity of their analog predecessors.

Reve's objective is not to remove experience, talent, or craft from the creative process, but rather to augment them. We seek to prevent the technical barriers and complexities of proficiency from extinguishing passion, expression, and ultimately, global communication.

Today, we are launching the first articulation of our original vision. Packaged into a single elegant and seamless user experience, the new Reve Image is really four products bundled into one:

  1. An image creator and remixer. Our image models are best-in-class at aesthetics, text rendering, and detail. Now you can edit both generated and uploaded images with simple natural language and combine anything you want from multiple images into one.

  2. A drag-and-drop editor (beta). No need to learn dozens of complex image-editing tools. For unprecedented, pixel-level control, open any image in the object-based, direct-manipulation editor, then simply click, drag, and render.

  3. Your own creative assistant. Riff with a tireless collaborator who can search the web, find inspiration, and translate any idea into beautiful artifacts.

  4. An API (beta). Third parties can now bring the power of Reve's image creation, editing, and remixing models into their own applications and workflows.

A New Visual Language For Content Creation

At the heart of Reve's transformation lies a breakthrough in how we represent visual information. Our new architecture analyzes and decomposes images into a code-like intermediate representation, what we call a layout representation—an internal visual language that captures the essence of any image with remarkable precision and efficiency.

This intermediate representation fundamentally changes how AI systems interact with visual content, offering three key advantages that unlock entirely new possibilities.

Precision Beyond Words

Consider the challenge of describing even the simplest photograph in text. To capture the exact position, size, shape, color gradients, lighting conditions, and subtle details of every object would require thousands of words, and still fall short of perfect accuracy. Our code representation solves this fundamental limitation. By encoding visual information in a rich, structured format, Reve can now generate images with complex spatial relationships, and make edits that preserve the unique likeness of individuals, and retains fine-grained details that would be lost in text descriptions.

Bridging Modalities: Where Language Meets Vision

Traditional AI systems operate in silos—language models understand text, vision models process images, but the connection between them remains tenuous. Our intermediate representation serves as a universal translator, seamlessly bridging the gap between human intent and machine understanding. This unified approach unlocks a new level of flexibility, seamlessly enabling conversational editing, creation from multiple visual references, or a combination of them in an iterative refinement workflow.

Direct Manipulation: Putting Users in Control

Our intermediate image representation is designed for transparency and control. Using existing AI content creation tools feels similar to vibe-coding, except these tools never give you the chance to tweak their outputs at a granular level, like you could with code. Rather than hiding the model's understanding behind a black box, we expose its internal representation, or code, to enable direct manipulation of visual elements. Users can now move, resize, add, remove or replace objects with granular control. This ensures users remain in the driver's seat, with Reve serving as a powerful but controllable tool rather than an opaque decision-maker.

What's next

The new Reve Image is already the simplest and most comprehensive image creation and editing platform in the world, and this is just the beginning of what is coming. Stay tuned as we continue to explore the frontiers of multimodal AI and build tools that empower creatives to bring their visions to life with unprecedented ease and precision.

Experience Reve for free today at reve.com.

Christian Cantrell, Michaël Gharbi, Taesung Park, Jon Watte, Mike Speiser

Co-founders