Skip to content
Notifications
Clear all

Beginner mistake I made: not using reference images

1 Posts
1 Users
0 Reactions
29 Views
(@grafana_guy_night)
Honorable Member
Joined: 6 months ago
Posts: 427
Topic starter   [#10230]

Hey everyone! Just started using Leonardo AI for generating images for my Grafana dashboard backgrounds and icons. I was struggling to get consistent styles across different prompts.

My big mistake? I wasn't using reference images at all. 😅 I was just typing prompts like "server rack line art" or "network graph icon" and getting wildly different results each time.

Then I tried uploading a simple sketch I made as a reference. The difference was huge!

**Before (just text prompt):**
```
Prompt: "a simple icon of a linux penguin, flat design"
```
Result: Got everything from 3D penguins to cartoonish mascots.

**After (with my rough sketch as reference image):**
Used the **Image-to-Image** feature with my basic outline, same prompt.
Result: Much closer to what I actually needed for my dashboard.

It seems so obvious now. The **Image Guidance** and **Prompt Magic** sliders next to the reference upload are key. Low Image Guidance lets the prompt do more, high Guidance sticks closer to your reference pic.

Anyone else have tips for keeping a consistent visual style across different generated assets?



   
Quote