After ten months of silence, OpenAI has lastly launched Dall-E3, the most recent model of its standard AI picture generator. Dall-E3 is claimed to be a major enchancment with extra nuance and particulars over Dall-E2.
What’s new in Dall-E3?
- Better picture high quality and coherence than Dall-E2
- Dall-E3 can generate legible textual content
- Dall-E3 works inside ChatGPT
- Improved security
To higher illustrate the distinction, check out the instance under. The two photos used the identical immediate:
Prompt: A paper craft artwork depicting a woman giving her cat a delicate hug. Both sit amidst potted vegetation, with the cat purring contentedly whereas the woman smiles. The scene is adorned with handcrafted paper flowers and leaves.
This instance exhibits Dall-E3’s skill to grasp and generate complicated ideas. Among different issues, I observed the woman in the suitable picture is neither smiling nor hugging the cat.
Dall-E3 can generate textual content
Another important replace is the power to spell phrases accurately and legibly. For instance, you possibly can use Dall-E3 to generate a classic journey poster:
Prompt: A classic journey poster for Venus in portrait orientation. The scene portrays the thick, yellowish clouds of Venus with a silhouette of a classic rocket ship approaching. Mysterious shapes trace at mountains and valleys under the clouds. The backside textual content reads, ‘Explore Venus: Beauty Behind the Mist’. The colour scheme consists of golds, yellows, and delicate oranges, evoking a way of marvel.
I like how the AI intentionally selected the font and colours to match the general aesthetics of the poster. That’s one thing no different AI software is able to.