If poor WALL-E had received as many updates as ChatGPT, maybe he would have found proof that life still remained on Earth a little sooner (although we all know who was to blame for there being nothing left).
OpenAI’s AI has launched ChatGPT 2.6 Images and, considering that, as we mentioned in this article, every week (yes, every week) 4 billion images are created, sooner or later they were going to have to update it. So we are assuming there will now be even more images generated by ChatGPT, because with this update it will generate them 50% faster.
According to OpenAI, Images 2.6 has been available globally since launch in ChatGPT and Codex, across all subscription tiers, on web, mobile, and desktop.
In addition, they have launched two models for developers, Flare and Sunburst, which are also available via the API:
One of the major new features is the ability to turn any image into a GIF using text instructions. It can animate brand mascots, characters, product photos, memes, or illustrations. It can also transform a simple drawing into an animated character.
We tested it with a drawing we had made of a giraffe (if you keep an eye on our social media, you will soon find out why), and this was the result:
Here is the original drawing:
All I had to do was add the image to the conversation and say, “please turn this drawing into an animation.” And its response, aside from the image, was the following: “Your giraffe has come to life! 🦒🎀 I animated it while preserving the drawing’s original style: it gently moves its head, winks, and smiles while showing off its pink bow.”
The model does a better job of preserving the features of characters, products, and brand elements when edits are made. This makes it easier to create multiple assets for a campaign without losing their visual identity. Along the same lines, it also improves the composition and arrangement of elements, especially in advertising posters, marketing creatives, and product images.
And something key is that you can now make localized changes without having to redo the entire image. In addition, the image markup feature (image markup), which was already available in ChatGPT, is now also coming to the API. It is ideal for developers and illustrators, who can now also export vector designs in SVG format for use on websites, apps, and printed materials.
They have also added a parameter that determines how much effort the model devotes to interpreting and executing a request. And it is the user who can choose whether they want a higher or lower reasoning effort. Low complexity favors quick edits, while medium and high levels are recommended for more demanding compositions.
Some companies such as PicsArt, Higgsfield, Figma, fal, and Sable were among the first to gain access to the tool.
Your email address will not be published. Required fields are marked *
Δ