Controlling image attributes and positions
DALL·E can control object attributes and positions, but success depends on caption phrasing and object count.
DALL·E can modify several of an object’s attributes, as well as the number of times that it appears.
Simultaneously controlling multiple objects, their attributes, and their spatial relationships presents a challenge. The success rate can depend on how the caption is phrased, and DALL·E is prone to confusing associations and brittle with respect to rephrasing as more objects are introduced.
Consider the phrase “a hedgehog wearing a red hat, yellow gloves, blue shirt, and green pants.” DALL·E must correctly compose each piece of apparel with the animal and form the correct color associations without mixing them up.
DALL·E can control object attributes and positions, but its success rate decreases sharply with more objects. Caption phrasing is critical, and semantically equivalent rephrasing may yield no correct interpretations.
- DALL·E: Creating images from text ↗OpenAI News · OpenAI News · Jan 5, 2021 · paraphrased
First seen on Subvolts Jan 5, 2021. Updated Sep 11, 2026. Spotted a problem? Tell us.
- DALL·E: Creating images from textOpenAI · OpenAI News · Jan 5, 2021
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- PromptGenerate 8 minimalistic logos conveying word meaning visuallyCreate 8 minimalistic logos where letters visually express the meaning of a given word. · 5 min
- PromptGenerate city block with 'BERLIN' integrated into architectureCreate a cityscape where buildings are shaped like letters spelling 'BERLIN' subtly. · 3 min
- PromptCreate image of phrase made out of wood chucked by woodchuckVisualize the phrase 'How much wood would a woodchuck chuck' made from wood chucked by a woodchuck. · 3 min
- PromptCreate infographic about a plant with interesting informationGenerate an infographic about a specific plant, focusing on interesting details and key information. · 2 min
- PromptCreate infographic showing how to make elaichi chaiGenerate a step-by-step infographic illustrating the process of making elaichi chai (cardamom tea). · 2 min
- PromptDesign 'TYPOGRAPHY' with retro screen-printed textureCreate a vibrant 'TYPOGRAPHY' design with a retro, screen-printed look and 3D effect. · 4 min
Questions people ask
- Which AI tools does this work with?
- Any current model. It was seen with in the sources.
- Where did this come from?
- OpenAI News (OpenAI News). Subvolts collected, normalized and credited it; the link above opens the original.