Hey everyone, I've been trying out DALL-E 3 for a few weeks now, mostly for creating scenes to use in my project management guides and internal team announcements. Everyone keeps talking about the 'improved coherence' as this huge leap forward, and for simple stuff, I totally get it! The text rendering is amazing for a motivational quote graphic.
But when I ask for something with a few specific elements in a detailed scene, it feels like it falls apart? For example, yesterday I prompted for: "A diverse remote team collaborating in a digital workspace, with a virtual whiteboard covered in sticky notes showing project timelines, a chat window with Slack bubbles visible, and a calendar widget showing Q4 deadlines."
What I got was... a mess. The whiteboard was there, but the sticky notes were just blurry rectangles with no text. The 'Slack bubbles' were just colored ovals floating near a window that looked more like a browser. The team members looked great, but the specific details I asked for got totally lost or merged together weirdly.
I'm coming from using tools like Asana and Notion where specificity is key, so I thought being detailed in the prompt would help. But it seems like adding more elements just makes DALL-E 3 struggle to keep them all correct and logically arranged. Is this just me being a beginner with AI image gen? Are there tricks to get better coherence for busy scenes like this, or is this a known limitation?
I really want to use it for visualizing complex project workflows, but now I'm unsure. Thx!