Ideogram 2.0 Arrives

Well, well, well... What have we got here?

Ideogram just released version 2.0 of its text-to-image model, establishing the year-old startup as a serious competitor to established names like Midjourney, DALL-E 3 and Flux. This new version brings advanced features and competitive pricing that could change the AI-driven content creation game.

Enhanced Text Rendering

The standout feature of Ideogram 2.0 is obviously its improved text rendering, tackling an ongoing challenge in AI image creation. “The Design style significantly improves text rendering in Ideogram 2.0, and it enables you to create premium graphic designs including greeting cards, t-shirt designs, posters, illustrations with longer and more accurate text,” the company shared on X.com.

This improvement could be a game-changer for businesses and content creators who need clear, on-brand messaging in their visuals. AI generators have often struggled with producing clear, readable text, but Ideogram’s breakthrough might make rapid prototyping in advertising and branding quicker and more precise. Even FLUX, which is very good with text, is only able to deliver best quality when text is short. Once you start writing taglines, complete sentences, the model tends to mix up letters and generate some gibberish. With Ideogram's posted examples, even the longer text is perfectly readable.

Color Control and API

Another major update is the customizable color palette feature, which makes Ideogram even more appealing to professionals. “Choose from multiple color palettes for images, giving you precise control over the color scheme. This is useful for brand consistency or for capturing a specific vibe,” the company tweeted. This feature meets a key need for brands that want consistent visuals, potentially speeding up workflows and cutting down on post-generation editing.

Additionally, the launch of a public beta API is a smart move to broaden Ideogram’s impact. By allowing developers and businesses to integrate this technology into their own apps and workflows, Ideogram is setting itself up as a versatile tool for various industries. The company’s statement, “We offer superior image quality at a lower cost compared to other models,” underlines a pricing strategy that could appeal to budget-conscious enterprises looking to expand their AI-generated content.

Hands-on with Ideogram 2.0

In my tests, Ideogram 2.0 performed quite well so far.

I've used the same prompt I've used in FLUX:

A web banner for a dog food advertising campaign. Background: A chaotic scene of dogs in a fancy dining room, each sitting upright at a table set with fine china and silverware, with some dog food on the plates. The dogs are wearing tuxedos, bow ties, and pearls, looking extremely serious and refined. One dog is holding a knife and fork in its paws, another is delicately sipping from a wine glass filled with water, and a third has a napkin tucked into its collar. The background should be slightly blurred, focusing on the surreal elegance and absurdity. Foreground Text: Main Text is at the top of the image: "Gourmutt's Feast" in an elegant, cursive font that contrasts sharply with the absurdity of the scene. The text should be gold with a slight shine, like something you'd see in a high-end restaurant ad. Text at the bottom: "ORDER NOW" Add a small icon of a chef's hat on a dog bone next to it for an extra playful touch.

The model lost some letters and completely ignored a request for the 2nd line of text at the bottom. I ran the same exact prompt in FLUX and it understood the request well. The dogs also donn't look as realistic as in FLUX. Still great quality, but not a winner here.

Next up, I've tried another prompt to compare, one with poster text. FLUX couldn't get the text right in like 10 tries, so I ended up shortening my text for it, but lets see if Ideogram can digest the whole thing. This time I've left 'Magic Prompt' on. 

Era & Style: 1950s sci-fi B-movie style, with flat color shading. Subject: A dramatic scene of a giant robot towering over a city, with panicked citizens running in all directions. The robot has a retro, clunky design with blinking lights and antennas. There's a UFO hovering in the background. Text: "Attack of the Mecha-Menace!" in bold, retro-style block letters, with a smaller tagline underneath: “They came from the stars... to take over our toasters!”

This job was done much better and one in 4 images actually came out with text totally right. The illustration quality is also superb. At least 2D in Ideogram appears to be flawless.

Next prompt didn't involve text.

An awe-inspiring conceptual portrait of a woman's face, meticulously crafted to represent the vibrant essence of a night city. The woman's face is intricately designed, composed of a harmonious blend of illuminated skyscrapers, bustling streets, and a serene moonlit sky. Her piercing, luminescent eyes draw the viewer in, capturing the enigmatic spirit of the urban landscape. This photo-illustration masterpiece brilliantly merges the beauty of the night city with the essence of human expression, creating a captivating and memorable piece of art, portrait photography, illustration, conceptual art, photo.

And this is another user's generation featuring a magazine cover. You can see that the 'Magic prompt' attempted to include a coherent paragraph into it, but only the main short text came out flawless, the longer one still contains gibberigh. 

Stunning visual though. If you log in, you can see it by this url https://ideogram.ai/g/KxFkcHUkQdS0sBns1TVPww/1

My verdict so far is: platform is super easy to use, has a lot of handy features (some are locked on a free acount) like remix, image prompting with image weight, upscaling, color palette, prompt enhancer etc, generates decent quality images, very good with digital illustration style. Definitely should be on my radar. But it failed to impress with text rendering. As things stand right now, I guess running FLUX locally for free beats having to pay for Ideogram or Midjourney for most users, but it's great to have competiton and options and I wish success to every player in this booming sector.

BTW, Ideogram is currently uncensored in the celebrities/polititians depertment. People are having lots of fun with that.

Redditors Chatting About Ideogram

People are buzzing with excitement about Ideogram, especially the latest version, Ideogram 2.0, which is getting praise for its high-quality outputs and user-friendly features. Here’s what’s being talked about:

• Image Quality: Users are saying the image quality is better than other models like DALL-E, with results that feel more genuine and visually appealing. They’ve shared prompts that produced impressive images.
• Versatility: The ability to mix styles, like portraits and landscapes, is getting a lot of love. Users are trying out different prompts and are often surprised by how well they work, even for complex requests.

Overall, the community is excited about Ideogram's progress and is looking forward to more developments that could push the limits of text-to-image generation.

In conclusion, Ideogram 2.0 is making waves in the AI image generation world with its improved text rendering, customizable color palettes, and new API. These advancements make it a strong contender in a rapidly evolving industry.