Midjourney's newest feature introduces customizable mood boards that swiftly generate custom stylization across generations. This allows creative professionals to effortlessly transition between multiple styles with simplified prompts.

What is a Moodboard?

Midjourney Moodboards should not be confused with the task of using generative AI to create a moodbo…

Read more

This fall has been buzzing with video AI announcements, with companies like Meta teasing MovieGen and OpenAI previewing Sora. But for now, these tools remain more promise than reality—unavailable to the public and leaving creators eagerly anticipating their release. Meanwhile, discussions around generative AI’s impact on Hollywood are heating up, with v…

Read more

One of the biggest challenges with generative AI tools today is the sheer volume of content they allow us to create. For individuals, it’s difficult to manage; for teams, it can quickly become overwhelming. Now that content creation has become nearly limitless, we need new ways to curate and organize it effectively—a task current creative tools aren’t e…

Read more

Ideogram is not the first generator to create a canvas. In some ways, Adobe's Firefly has always been canvas-based, and Visual Electric was not far behind in emphasizing the canvas from its inception.

However, I’ve been keeping my eye on Ideogram as their changes have been coming quickly. Over a year ago, they were one of the first image generators to in…

Read more

After taking a brief summer break, it's been fascinating to re-immerse myself in all things work-related. Within the past month, there have been several major announcements and updates that have raised the bar yet again as we progress toward more sophisticated tooling. Naturally, this also means it's time to update the "Designing with AI" reviews. Since I hope to release an updated Playbook within the next month, I thought it would be helpful to at least release the visually generated updated reviews for those exploring and considering tool selection.

As a reminder, the DWAI ratings are based on five key areas with a rating of 1 to 4 stars:

• Subject accuracy refers to how closely the generated image aligns with the text description.

• Style accuracy refers to how well the AI model captures the desired style or aesthetic of the output.

• Cohesion is about the quality of blending and refers to the cohesiveness and resolution of the final content. (e.g. doesn’t include artifacts in final image)

• Ethics is about the approach to developing creative generative AI responsibly. This includes information about the training data, methods leveraged to avoid copyright infringement, and bias in output.

• Ease of use acknowledges that the way we access these technologies is changing. Many still require a lot of work, while others prioritize easy-to-use interfaces.

Let’s take a look at what’s shifted in the landscape:

Read more

Hate to hit you with another post this week, but I think it’s worth it!

As of yesterday, Midjourney has discontinued its alpha phase and introduced a new web app accessible to anyone who has created 100 or more images. In other words, it's now available to everyone.

As many of you are aware, I was initially hesitant to switch to the web app due to the lac…

Read more

About a year ago, many of us were in awe of the first AI text-to-image generator that could create words, not just images. Ideogram challenged our assumptions about what was possible and created a wave of excitement and concern simultaneously.

However, after experimenting with it for a while, the initial amusement eventually wore off because its integration into our work was limited. Frequently, words would be misspelled or disfigured. So, while the promise was compelling, the initial version felt incomplete.

But, today’s release of Ideogram 2.0 changes that trajectory. Let’s take a look.

New models

One of the most significant differences in 2.0 is the availability of five new, nuanced models. These models include: General, Realistic, Design, Anime, and 3D.

By default, you will be on the Automatic setting, where Ideogram makes a choice on the correct model, presumably based on your prompt.

Here’s an example of how the same prompt can be translated with each model:

Read more

As Midjourney transitions from Discord to the web, the first version of the web editor is now live for Alpha users. Personally, the inpainting and editing features in v6+ were the main reasons I continued using Discord. Let's see if the web editor is good enough to finally make me switch.

Finding editor
The new feature is unfortunately located in the bott…

Read more

One of the main challenges in using generative AI today is its limited ability to understand context. This hinders its potential to help us complete projects, rather than just generating isolated pieces of content.

In our latest AI Design Lab session, we reviewed the available features that help create consistency in AI-generated imagery.

Consistency can manifest in various forms, such as maintaining consistent styles across multiple team members, producing numerous images with the same style, or leveraging consistent characters across images. The possibilities are endless, but while these features are emerging, they are not yet perfect.

Of course, each feature comes with its own advantages and drawbacks in terms of implementation. If you intend to use these features to achieve a certain level of consistency, it's crucial to understand the trade-offs involved.

So here’s an exhaustive review of the options you have to create consistency in your work utilizing generative AI:

• Training your own model

• Flash or character sheets

• Prompt layering within a single model (often to generate prompt libraries)

• Firefly Reference Image

Read more

Despite my skepticism, I am continuously impressed with Adobe's advancements in AI tooling. Today, they made an announcement regarding the integration of AI features and third-party models into Adobe Premiere Pro.

This development surpasses my previous prediction that video models would be incorporated into our tools by 2025.

Read more

You may be wondering why I haven't posted about Sora this week. To be honest, there is nothing new for me to share that you haven't already seen in the 100 videos you've probably already watched. Once we are able to compare its features and performance with other tools, we will delve deeper into it.

In the meantime, let's take a look at a tool that is currently accessible to design teams: Visual Electric.

Visual Electric is not the first tool specifically designed for designers (Oh hey, Adobe), but it definitely sets itself apart with its user-friendly approach compared to other high-quality image generators like Stable Diffusion or Midjourney. When it launched in late 2023, it made me think that we may be entering a new era of products, as they made some significant choices in their prioritization. Only time will tell how the industry will evolve in the coming months. In the meantime, here are a few noteworthy features:

The generative canvas
When Visual Electric first launched, the generative canvas immediately caught my attention. Unlike other generative AI tools, this one focused on creation rather than just generation. While the canvas itself is not a new concept, the fact that it was prioritized over pure technology shows that the creators are aiming for more than just a race to produce the most advanced tool. This makes it much more enjoyable and feasible to incorporate generative AI into the creative process, whether it's for brainstorming or refining ideas. Will this be a long-term differentiator? Probably not, but it's definitely a reason to consider using this tool right now.

Read more

Hopefully, you were too busy relaxing to even notice, but right before the holidays, Midjourney released two major Alphas. First, there's the updated model, V6 in Discord. Secondly, there's the new website at https://alpha.midjourney.com, which allows you to prompt from a browser. Let’s take a look at both:

Midjourney V6 (Alpha)

The key changes to a sophi…

Read more

Over the past few weeks, we've seen the emergence of new versions of image generators— Firefly 2 and DALLE-3. Naturally, I felt compelled to assess their advancements and compare them to my current benchmark, Midjourney 5.2.

In every assessment, I’m looking at several things:

• ★ Subject accuracy refers to how closely the generated image aligns with the tex…

Read more

I hate to say it, but Adobe is on fire right now.

During the Adobe Max 2023 event on Tuesday, they announced (only seven months after its beta debut) Adobe's Firefly generative AI will receive a trio of new models. And over 100 new features and capabilities.

I think four of the AI-related ones are going to change the way we create with Adobe:

• Firefly Vec…

Read more

Today, Adobe Firefly and it’s related features in Photoshop, Express and other Creative Cloud apps are available for commercial use. This includes:

• Adobe Firefly (http://firefly.adobe.com)

• Generative Fill and Generative Expand in Photoshop

• Generative Recolor in Illustration

• Text to Image and Text Effects in Adobe Express

What does commercial use mean?
Any usa…

Read more

If you've taken any of my recent AI courses, you might recall me saying that generating typography in your images was not possible. Well, guess what? It is now!

Today, Ideogram.ai has made their tool available to everyone for free generation. Founded by former Google Brain researchers and launched with $16.5 million in…

Read more

Read more

They're boasting about cooking up images with fewer prompts and even conjuring words within these images. The first part, I'm a bit skeptical about—but the l…

Read more

The highly anticipated 5.2 was finally released this afternoon, and I couldn’t help but dig in. There’s definitely a shift in generative output. Photography feels sharper, more saturated, and more realistic. But I’ll be honest, when it comes to illustration I’m not sure if it’s for the better. A lot of my flat illustrations are no longer flat, but I’ll …

Read more

I don't often cover launch day news or announcements here, but I had to share this because not only is it an example of a powerful, generative tool being integrated into your everyday tools (yes, it's happening), but it's also something I bet you've wished for throughout your career. 

Read more