Adobe have released Firefly V2 some time ago. It’s very good, random sample from the net below. All my previous criticisms are void now. Two things I have noticed, though:

• the NLU (natural language understanding) capabilities of the input text prompt is not as strong as with OpenAI DALL-E 3.
• I may(!) have seen slight blurriness introduced when working on a delicate pencil-art image. Not entirely sure as it was 1) in the middle of the night and 2) observed on a very good monitor, but given that it was just minor edits at that point, I decided to just send “the other version” to the printer. Just look out for this, to be safe.

If someone is interested in a scrappy Google Colab Notebook for DALL-E 3, I’ll be happy to share. (Own OpenAI access key required as usual).

Kris Kashtanova on X

I can’t believe this is how far we’ve come with AI image generation since last year. This is Adobe Firefly Image 2.

prompt: a young woman floats in a gorgeous late summer sea. Red petals float in the water emphasizing the delicacy of her features. Her closed eyes reflect the…