It is a big step for AI community now AI understands images You could :

• ask questions about image
• ask to describe images
• extract a text or math formulas

DALEE and image generation

Another big update is a dalee model avaliable for chat GPT 4 users

Model is quite smart in generation. On your simple prompt model offer few more detailed prompts for imagest that you could reuse or corect or use for future inspirations

Here are the prompts:

Photo: A panoramic view of Berlin’s skyline with its signature buildings. In the center, a transparent digital screen floats, displaying intricate software architecture diagrams. Nearby, a steaming cup of tea is placed on a table with tarot cards spread out, and floating philosophical quotes relate to data privacy and self-sovereign identity.

Illustration: The Berlin Wall, but with a futuristic twist. Digital circuits and codes blend with graffiti art. People from diverse descents walk past, some engrossed in deep thought, others holding glowing tarot cards. At the base, tea leaves form patterns that mimic binary codes.

Watercolor Painting: A serene scene of Berlin’s Tiergarten park during autumn. Among the falling leaves, digital icons representing software components flutter down. On a park bench, a philosopher reads a book with a tarot card as a bookmark, and a teacup beside him emits steam shaped like data clouds.

Vector: A vibrant street in Berlin during a tech festival. Stalls are set up showcasing innovations in software architecture. People interact with holographic displays, while others sit in circles discussing philosophy. In the center, a large digital teapot pours out a stream of binary code into cups, and tarot card designs float above.

I’ll generate these images with the specified size for you. As you can see you could select size or continue play with prompts

Voice conversation

So one more update and feature is a voice conversation on mobile version. Now you dont need speech to tech module on your phone you could chat with an agent directly from your app!

Limitations

• you couldn’t use image to generate images and combine images for now
• query adout image and image generation is to separate modes in UI
• some times is better to prepare detailed prompt is separate chat and give a lot of details and only after use it in a chat
• it is hard to combine images and text so it is require multiple chats that create a problem of history and context passing between a chats