Isometric 3D Icon Set Prompt
Create a set of [number] isometric icons for [subject, e.g. "a SaaS dashboard: cloud storage, analytics, settings, notifications"], using a consistent 30-degree isometric perspective with no vanishing-point perspective. Flat, saturated color blocking with minimal gradient -- [color palette, e.g. "navy and teal with a single orange accent"]. Each icon centered on the same background, matching scale and lighting across the full set.
Variations
Two more versions, ready to copy on their own.
Create one isometric icon of [specific subject], 30-degree isometric perspective, no vanishing-point perspective. [Color palette]. Flat color blocking with subtle edge shading for depth, more detail than a simple flat icon while keeping the isometric projection consistent.
Create a style reference for an isometric icon set: [subject, e.g. “5 icons representing project stages: planning, design, build, test, launch”], 30-degree isometric perspective, [color palette]. Describe the shared visual rules (angle, shading, color blocking) I should apply to each icon so I can generate them individually while keeping them consistent.
FAQ
Why do my icons look isometric individually but inconsistent as a set?
This usually means the projection angle wasn’t named explicitly — naming “30-degree isometric angle, no vanishing-point perspective” in every request keeps the angle consistent across separate generations.
Can I get more shading and depth without breaking the isometric look?
Yes — ask for “subtle edge shading for depth” rather than full gradient shading, which tends to preserve the flat, isometric read while adding some dimensionality.
Does this work for icons with text or numbers in them?
It can, but text rendering inside generated icons is unreliable across most models — for numbered or labeled icons, generating the icon shape alone and adding text separately in a design tool tends to be more reliable.
How many icons can I generate consistently in one request?
Around 4-5 tends to hold up well; larger sets are more consistent when generated in batches of 4-5, referencing the first batch’s established style.