Ideogram: A New AI Image Generator that Can Handle Texts

In this newsletter, read about:

• 🕵️‍♀️ Ideogram: Generating Texts Inside Images

• 🗞 News and Top Reads

• 📌 AI Art Tutorial: Midjourney Inpainting Deep Dive

• 🎨 Featured Artist: Blac

• 🖼 AI-Assisted Artwork of the Week

• 🤓 A Comprehensive Midjourney Guide

🕵️‍♀️ Ideogram: Generating Texts Inside Images

A new player has entered the AI text-to-image generation arena: Ideogram, a Toronto-based startup. Ideogram is the first image generator I've seen that outperforms Midjourney in some tasks, specifically incorporating text into AI-generated images. While I'm still a fan of Midjourney's performance and aesthetics in generating creative and realistic images, Ideogram is currently the best at typography.

What You Need to Know about Ideogram

Ideogram (pronounced eye-diogram) was co-founded by Mohammad Norouzi, Chitwan Saharia, William Chan, and Jonathan Ho, all former Google Brain researchers who co-authored the famous Imagen text-to-image generator.

The Ideogram team built the model for this AI image generator from scratch. The tool can produce beautiful typography in many fonts and text styles, which would otherwise be extremely time-consuming to create. People are using Ideogram to create logos, posters, graphic designs, and images that convey interesting and fun messages.

It is currently available for free at ideogram.ai. To experiment with Ideogram, you just need to log in using your Google account. All your generations will be publicly available.

Finally, it looks like currently Ideogram only allows non-commercial use of its services.

How to Use

To generate an image using Ideogram:

1. Describe what you want to see (i.e. write a text prompt).

2. Choose one or several styles, among photo, poster, 3D render, typography, cinematic, painting, fashion, product, illustration, anime, architecture, dark fantasy, vibrant, graphite, portrait photography, wildlife photography, conceptual art, ukiyo-e.

3. Choose an aspect ratio, among three options: vertical (10:16), square (1:1), and horizontal (16:10). Square image will be generated by default.

4. Hit Generate.

The website also features the Top, Trending, and Recent generations of other users that you can use for inspiration.

A Few Examples

Ideogram is a new thing, and I am not an expert yet to give you any great prompt tricks for beautiful and reliable results. So, let's experiment with creating a couple of images in different styles to see what we can get from this tool.

As you can see, Ideogram is still not perfect with texts. However, if you don’t get what you want from the first roll, you may choose to “Remix” the image that looks close, and the second attempt will usually generate much better results.

Now, let’s try creating a realistic portrait with Ideogram.

The images look interesting, but the quality is much behind Midjourney.

Finally, I want to see how Ideogram can handle product photoshoots.

Well, it seems that hands and faces are still a major challenge for this tool.

To Wrap Up

Based on my limited experience with Ideogram, I find it to be more business-oriented and practical than Midjourney, which is better at creating artwork, realistic portraits, and weird images 😂 

So, my recommendation is to:

• Use Midjourney, if you want to create original digital artwork with beautiful aesthetics, photorealistic portraits, or some edgy images.

• Use Ideogram, if you need texts or symbols inside the image, or if you need some illustrations for a business presentation or a blog post.

Obviously, Midjourney also gives much more flexibility and control over the generated images with their advanced tools like zooming out, panning, inpainting, etc. Still, Ideogram is very new and I believe with such a strong research team, they should be able to implement these tools sooner rather than later.

🗞 News and Top Reads

• OpenAI has partnered with Canva to allow ChatGPT Plus users to create logos, banners, and other visual content.

  • To use this feature, users will need to install the Canva plugin for ChatGPT.

  • It's important to note that this integration does not enable text-to-image generation like Midjourney. Instead, it uses Canva's pre-existing, wide range of templates that users can edit to their liking.

• X (former Twitter) has updated its Terms of Service to let it use all posts for AI training.

• Amazon has introduced Amazon One, a contactless palm-based payment and identity verification system that is more accurate than iris scanning.

  • The system uses a neural network trained on millions of artificial palms to create a unique palm signature for each user. This signature is then linked to their credit card or Amazon account.

  • Amazon One has a 99.99% accuracy rate on over 3 million uses and is being rolled out to more than 500 Whole Foods Market stores and dozens of third-party locations.

📌 AI Art Tutorial: Midjourney Inpainting Deep Dive

In this video, Christian Heidorn examines the strengths and weaknesses of the most recent Vary (Region) feature by Midjourney. He demonstrates that in certain instances, the Inpainting tool is prone to misinterpreting user prompts, leading to peculiar outputs.

🎨 Featured Artist: Blac

Blac has been an artist for a long time; as an abstract expressionist painter and designer, and now as a proud AI Infused Artist. He explores conceptual contrasts such as good vs. evil, freedom vs. ruin, and reality vs. unreal. His artwork has been exhibited in Rome at the Mecenate Fine Art Gallery, as well as Superchief LA, and included in several publications in the NFT & AI Art communities, including NFTnow, Aguilar, and more.

🖼 AI-Assisted Artwork of the Week

Meet Maya and her friend. Their character and worldview were formed on the endless expanses of the steppes. As the local animals dissolve among the inconspicuous grass, so people dissolve here in the endless space, where the boundaries are only imagination.

🤓 A Comprehensive Midjourney Guide

Share Kiki and Mozart

If you enjoy this newsletter and know someone who might also appreciate it, please feel free to share it with them. Let's spread the word about AI art and introduce more people to this fascinating field!