• Unicorn Fuel
  • Posts
  • Abracadabra: AI-Generated Vectors are here!

Abracadabra: AI-Generated Vectors are here!

It just keeps getting better and better

Welcome back design friends!

Yet another couple weeks have gone by in the exhilarating, confounding experience that is the design industry in 2023 — which means even more AI bombshells have dropped.

But first, a quick recap about what's happening with AI in the world beyond design. The EU is making moves to regulate AI and label AI-generated content to fight disinformation. A report from McKinsey predicts that generative AI could add $4.4 trillion (!) annually to the global economy. And the Beatles are creating a one-song comeback — featuring the late John Lennon — courtesy of AI; it will be released later this year.

It’s kind of scary but exciting, because it’s the future. We’ll just have to see where that leads.

Paul McCartney

In the office, a BCG study of nearly 13,000 people from 18 countries revealed that “leaders are much more optimistic than frontline employees about AI (62% vs. 42%),” and those who use generative AI regularly were more optimistic than non-users (62% vs. 36%.) Then again, CNN claims that 42% of CEOs say AI might destroy humanity 😅, so it's clear nobody knows exactly what to expect. (And, apparently, if you listen to Stable Diffusion’s CEO you might be even more confused about what is true and what isn’t.)

The BCG study also revealed that leaders are being given AI training, while frontline employees are not. If this is happening in your company, be proactive. Ask for training, or take it upon yourself to train up ASAP. Developers, for example, aren't planning to be left behind: a Github survey found that 92% of devs are using AI tools.

Ok, now that you're caught up, let's dive into some new AI-powered design.

At a Glance

  • Oooo, did you say AI-generated vectors?

  • AI tools come to...uh, AI

  • Adobe is totally gunning for Canva

  • Harry Potter, reimagined in the worst way

AI Unveils Its Newest Trick

A much anticipated AI phenomenon just stormed into our quickly changing world — generative vector-based icons and illustrations via Recraft.ai. It’s an impressive new design platform that was released in the past couple weeks by Founder and CEO Anna Veronika Dorogush. 💪 

Recraft is the first Generative AI tool that creates complex vector art from text.

Anna Veronika Dorogush

Heavy-weights like Midjourney and DALL•E 2 have offered a pixel-party that frankly had us dancing in our seats, providing the ability to conjure stunning imagery in seconds. But deep down, we knew their reliance on pixels was their Achilles' heel.

For instance, while we could use those tools to generate logo ideas (as a form of brainstorming), it was much harder to produce usable logos. Since you can only get a high-res bitmap logo out of Midjourney, you'd have to recreate it as vector art to make it ready for prime time. And, despite meticulous prompt engineering, it was hard to avoid shadows and gradients that simply didn’t translate cleanly to a vector image.

A series of different illustrations of unicorn heads that resemble logos

Images on left generated with Midjourney. Vector logo on right finalized in Illustrator.

Recraft, on the other hand, generates vector-based imagery on an infinite artboard and allows you to download the results as SVG, PNG, JPG and even Lottie files (for animation).

Each of the logo styles shown below downloaded perfectly as a SVG, with no problems or glitches translating the onscreen image to a usable vector file. While we didn’t test a wide range of logo styles, from what we’ve seen so far, it’s a game-changer as far as AI-generated logos are concerned. We haven’t encountered another system that can compete with what we generated using Recraft.

A series of different illustrations of unicorn heads that resemble logos

Images generated with Recraft

The platform is in its infancy, and thus the interface isn’t polished and some of the images it generates can be a bit rough, especially the icons. But, it does show promise. Right now, it will let you specify color palettes, upload a style reference, and render your art as one of many preset styles: icon, illustration, pictogram, hand-drawn, photorealism, Flat 2.0, or 3D.

A screenshot of the Recraft.ai user interface showing the text prompt 'unicorn' and a series of images including vector and 3D-rendered unicorns

Screenshot of images generated in the Recraft interface

Recraft is currently free for everyone and allows for commercial use of its generated imagery, however we did not find any documentation about whether it was trained on royalty-free imagery or not.

Adobe brings AI to Illustrator

The graphic design mothership just pulled a new rabbit out of its hat, bringing generative vector art to the mainstream, announcing brand spanking new AI-driven features in Abobe Illustrator (Beta).

Behold “Generative Recolor”

An image showing an illustration of a long-haired girl wearing sunglasses surrounded by a halo of comic book-style shapes. Overlaid are a text field that reads 'neon spray paint' with a 'Generate' button and preview images of the girl rendered in different color schemes.

Image from Adobe

The Generative Recolor feature allows designers to use text prompts to radically shift the color palette of an existing artwork with one click.

You dream up a color inspiration, maybe "midnight in the jungle" or "sunrise on Mars," type it in, and voila! Your art auto-magically dances to a different color tune. It's an innovative approach to exploring design options.

If you have 1 minute to spare, take a peek at their promotional video to see it in action.

The video shows how you could use the Generative Recolor feature for product design, poster design, and illustrations. We could also imagine using it for identity design, running a logo-in-progress through a variety of different colors to explore different moods.

To get your hands on the new feature, you can update the app through Creative Cloud, if you have a license.

Once you've got the latest version, open an Illustrator file, select some vector artwork, and then use the Generative Recolor command found in the Edit menu under Edit Colors. Finally, add your desired “color theme,” and hit the Return key. The software will offer up 3 options to choose from.

Now for a totally practical use of AI

Retype, another new Illustrator (Beta) feature, promises to identify any font in an Illustrator file — even if it's part of a bitmapped image, like a PNG. (Finally!)

Once you've identified a font, you can apply it across your design instantly.

This is less of a mind-blowing AI moment, than an "it's about time" moment. Similar tools have been around for years, but we sincerely appreciate having it baked right into the Adobe platform.

Adobe Express Strikes Back: Canva in the Crosshairs

Last week, Adobe Express (Beta) came at us, guns blazing, with a major upgrade and a host of AI features powered by Adobe's Firefly engine.

AI-powered text-to-image and text effect tools are now included in Adobe Express—a platform that brings all the Adobe superpowers into one platform, including background removal, image scaling, and more.

Unlike many popular image generation tools, such as Stable Diffusion or DALL•E 2, Adobe Express offers up dozens of preset "Styles" — like Pop Art, Cubism, Clay, or Synthwave.

We don't always love style shortcuts as a rule, but unlike some AI-driven apps, Adobe Express lets you stack several of them up at once. We've had some fun combining various presets to arrive at some experimental styles, like "Cyberpunk Layered Paper."

A split-screen image showing a user interface on the left and an illustration of paper butterflies in an origmi street surrounded by skyscrapers. The user interface includes the text prompt 'A swarm of butterflies in a downtown New York City alley'. Two preset styles, called Cyberpunk and Layered Paper, are also selected.

Image generated with Adobe Express (Beta)

Adobe also claims the image-generation engine is trained only on licensed stock, asserting that its outputs will be good to go in commercial settings. Finally, agencies can heave a sigh of relief, knowing it will be kosher to incorporate AI into their design flow.

(While the press release didn't make it obvious, we assume you should still hold off on using these tools for client work until they are out of beta, as mentioned in our article on Photoshop's AI Leap: The Next Phase of Image Generation.)

You're totally my type

Adobe Express also adds a new text-related AI-powered feature: your plain language prompts now hold the power to morph your type into artwork!

A split-screen image showing a user interface on the left, with a text prompt that reads, 'Swirling blue and purple and green fur'. On the right, the word 'Fuzzy!' appears in plain gray type, and below it, the same word appears to be made from colorful fur.

Image generated in Adobe Express (Beta)

If you grew up creating graphics of your name in glitter to adorn your MySpace page back in the day, playing with the new type feature can feel like a nostalgic trip to a simpler time.

And just like with those cheesy web-based tools of yore, the results can be a bit hit-and-miss. But Express does let you keep loading additional iterations until you find one you like. Plus, the ability to keep creating iterations until you strike gold is a refreshing change from the pay-per-generation model we're used to (cough Midjourney cough).

With animation and video, shared brand asset libraries, and the ability to export at various sizes — and publish straight to Instagram — Adobe is clearly positioning Express as a Canva-killer.

We'll likely put it to use for creating social posts and stories. It doesn't scream revolutionary to us, but we suspect that it will make the process easier and perhaps more fun. We did waste spend a couple enjoyable hours playing with the text effect tool. (Who doesn’t want to see their name spelled in liquid rainbow colors??)

Generated in Adobe Express

Check out the promo video for a quick taste of the new Adobe Express (beta) feature set.

If you want to give it a go, you don't even need a paid Creative Cloud account. Just hop over to the Adobe Express (Beta), sign up for a free account, and try it right in the browser.

Our take on the new tools

It has been a curious experience for us, checking out these new tools.

In comparison to our adventure of discovering and experimenting with AI generative platforms for the past two years, Adobe's ventures into AI feel a bit like bopping up and down around a track on a highly-trained pony — rather than riding a wild stallion bareback through the woods. It's still fun, but it has been so tamed and stylized that the experience is completely different.

That said, the more tamed the tools are, the more they will be of use to designers who want to keep working the way they have always worked — with some neat opportunities for styling.

The future path that Adobe is betting on is the same one many designers are hoping for — a world where the creative process remains relatively untouched, and we just gain efficiencies and some expanded capabilities.

It makes sense: Adobe's business model is heavily invested in the creative industry remaining roughly the same.

On the other hand, the future that Google seems to be shooting for with Performance Max could prove to be more disruptive. It's the first example of a multi-modal creative platform that we've seen thus far: the single app includes a chatbot, text generation and image generation.

Even though it's exclusively designed for making ads, we can envision other uses for that type of creative platform, as we discussed in Disrupting the Creative Process: New AI Tools from Big Players.

We'll dive into the tantalizing possibilities for multi-modal platforms in a future issue.

In the meantime, if there are tools you would like us to review and/or give our opinion on, hit reply and send us a message! 🚀

Get Inspired

After all that heady talk, we thought you might appreciate taking a break with our favourite lost sitcom, The Wizards Next Door.

A series of 3 images side by side. The first shows a building that resembles a mix between a castle and a suburban home with the text, 'The Wizards Next Door'. The middle image shows a rendering of actor Daniel Radcliffe as Harry Potter in a suburban living room, wearing a collared shirt and sweater. The final image appears to show actor Tom Felton smiling kindly while leaning on a patterned sofa.

Images generated in Midjourney

A creative Redditor used the magic of AI to reimagine Harry Potter as a mediocre TV show from the 1990s, and we think they nailed it.

See the whole image series over on Reddit.

Until next time, happy generating! 💗
s&k

P.S. Know a designer you care about? Consider sharing our newsletter with them so they don’t have to ride the AI rollercoaster alone either.

(You can earn some cool prizes too!)

Join the conversation

or to participate.