Slowly, artificial image generators are overcoming classic challenges like human anatomy and text. Earlier this month we talked about Ideogram AI, a platform optimized for injecting text into our creations, but it's not the only one of its kind out there. Today we came across Glif, which offers a simplified version of the process. All we need to do is enter a prompt, the text to render, and a font size. The algorithm takes care of the rest.
Hands, feet, eyes, text. Nobody denies that generative models have trouble reproducing these elements, but as training methods improve, so do the results. Some platforms have decided to focus on text rendering, and one of the most interesting is Ideogram.ai, with a style that essentially reminds us of Midjourney. However, we also find Glif, a platform that has several generative models, among which stands out the creation of user "fab1an" for rendering text using different objects.
Glif: Render Text Using Scenes and Objects with AI
fab1an's Glif enables a total of three fields: the first is for entering our prompt (simple descriptions work, but it recommends a bit more work here), the second is for the text to render, and the third defines the font size. If it's a single word, the suggestion is 130, but the logical thing is to experiment until you find the exact size.
From what I've noticed so far, rendered text looks much better if we use short expressions... and we try more than once. Unfortunately, the generative models available in Glif do not have advanced parameters like guidance or intensity, so our main control is to roll the dice until the right image appears. The good news is that Glif has no generation limits in its alpha phase.
But the platform's story doesn't end there. A click on the Glifs section will take us to the full list of available Glifs (ranging from comics and memes to tarot cards and selfies), and beyond the occasional error, they work well. Finally, one warning: NSFW. Link below!
Access the Glif: Click here
