Apple’s new Photoshop-like AI image editor is fascinating

Apple’s MGIE exciting new AI instruction-based image editor is fascinating
(Image credit: Future)

Apple has released MGIE, a new open-source AI model for image editing based on natural language instructions. As reported by VentureBeast, MGIE was developed in collaboration with researchers from the University of California, Santa Barbara, MGIE leverages multimodal large language models (MLLMs) to understand user commands and perform pixel-level manipulations on images.

The model exhibits versatility, handling tasks ranging from global photo optimization to specific local edits, including those resembling Photoshop modifications.

Presented at the esteemed International Conference on Learning Representations (ICLR) 2024, the research behind MGIE demonstrates its effectiveness in both objective, data-driven metrics and subjective human evaluation. Notably, the model achieves this while maintaining efficient processing times, minimizing potential latency issues.

While still in its early stages, MGIE represents a significant step forward in natural language-guided image editing. Plus, its open-source nature fosters further exploration and potential applications across various sectors.

Apple MGIE: What is it? 

Imagine telling your editing software, "Make the sky pop!", and witnessing vibrant clouds materialize. That's the magic behind MGIE, Apple's new AI model that transforms textual instructions into pixel-perfect edits. Buckle up, photo enthusiasts, because the way we interact with images is about to shift gears.

Unlike existing models, MGIE leverages powerful MLLMs – AI masters of both language and imagery. These super-brains analyze your words, deciphering not just their meaning but also the visual intent behind them. So, "Make the sky more blue" translates into explicit, actionable instructions like "Boost sky saturation by 20%."

And that's just the first step. MGIE then conjures a "visual imagination" – a mental picture of the desired edit. Think of it as an internal sketchpad translating your words into a precise vision. Armed with this vision, the model meticulously manipulates pixels, bringing your edits to life.

This intricate dance between language, imagination, and manipulation happens seamlessly, thanks to MGIE's innovative training program. By optimizing various steps simultaneously, the model learns to bridge the gap between words and visuals with remarkable accuracy.

So, while MGIE is still young, it paves the way for a future where photo editing transcends menus and sliders. Imagine instantly sculpting images with your words, crafting emotions and aesthetics with mere phrases. The future of image editing is bright, and MGIE's eloquent brushstrokes are a glimpse into its exciting possibilities.

Apple MGIE: What can it do?

Forget clunky menus and confusing tools - MGIE redefines image editing with the power of natural language. This AI marvel lets you sculpt photos with your words, handling everything from simple tweaks to complex object manipulations.

Global adjustments? Local edits? Photoshop-style transformations? Just tell MGIE what you want, and watch it work its pixel magic. Here's how it gets creative:

Clear & Concise Instructions: MGIE doesn't need cryptic commands. Say "Brighten the sky" or "Slim my waist," and the model translates your wishes into precise editing instructions.

Photoshop-Level Power: Crop, resize, rotate, add filters - MGIE's got it all. Want to change the background or add an object? No problem. It even handles advanced edits like blending images seamlessly.

Globally Gorgeous Photos: Need a quick quality boost? MGIE optimizes brightness, contrast, and color balance in a flash. Feeling artistic? Apply cool effects like sketching or cartooning.

Precision Local Edits: Focused on specific areas? MGIE lets you edit faces, objects, and even details like hair or clothes. Change their shape, size, color, or texture - all with your words.

MGIE is more than just an editing tool; it's a glimpse into the future. Imagine instantly bringing your creative vision to life, simply by describing it. With its versatility and power, MGIE promises to revolutionize the way we interact with images, making editing faster, easier, and more intuitive than ever before.

How to take advantage of MGIE

Say goodbye to complex editing software and hello to the intuitive power of words! MGIE, an open-source AI marvel, is here to revolutionize your photo editing experience. This groundbreaking model lets you sculpt and transform images simply by describing your vision.

Craving a vibrant sunset? Just whisper "Make the sky more dramatic," and MGIE will work its pixel magic. Want a complete makeover? Tell it to change the background, add objects, or blend images seamlessly. MGIE doesn't just mimic Photoshop; it empowers you with global adjustments, local edits, and artistic effects, all controlled by your natural language commands.

Dive deeper with the readily available demo notebook on GitHub, or experience MGIE firsthand through the online web demo. The open-source nature opens doors for customization and integration, making it perfect for individual exploration or pushing boundaries in research and development. So, unleash your inner artist and step into the future of photo editing with MGIE - where words become your creative wand.

Final Thoughts

Although the idea of Apple throwing its hat into the text-to-image ring excites me, in its current rudimentary form, I have found MGIE to be super slow, and not something I would use in its current form to help me be creative while trying to maintain the speed that my normal productivity requires. 

That said, I get the feeling, as with many products released these days, that it is far from a finished product and will end up becoming a slice of software befitting of Apple’s MacBooks in time. They are and have been the premier creative laptops for a while now, after all.

More from Laptop Mag

Mark Anthony Ramirez

Mark has spent 20 years headlining comedy shows around the country and made appearances on ABC, MTV, Comedy Central, Howard Stern, Food Network, and Sirius XM Radio. He has written about every topic imaginable, from dating, family, politics, social issues, and tech. He wrote his first tech articles for the now-defunct Dads On Tech 10 years ago, and his passion for combining humor and tech has grown under the tutelage of the Laptop Mag team. His penchant for tearing things down and rebuilding them did not make Mark popular at home, however, when he got his hands on the legendary Commodore 64, his passion for all things tech deepened. These days, when he is not filming, editing footage, tinkering with cameras and laptops, or on stage, he can be found at his desk snacking, writing about everything tech, new jokes, or scripts he dreams of filming.