Trained on 46 flat color, minimalist vector graphic illustrations. The main purpose is to produce images suitable for boring business style images (aka "Corporate Memphis" style).
But the training material does contain some "fun" vector graphics as well, so the kind of result you will get is a bit unpredictable and heavily dependent on the prompt. Please browse the showcase gallery to get a better idea of what the LoRA is capable of. Have fun! 🎈😁👌.
If you are looking for a more "artistic/cartoon" style, check out this other LoRA: Minimalist Flat Color Illustration Style
Description
This is epoch 10, other epochs can be downloaded from tensor.art/models/843588640140637348/Flat-Color-Vector-Art-1-NcD8C5-2025-03-21-23:47:54-Ep-10
FLUX.1 - dev-fp8
Trigger: vector art
Repeat: 20 Epoch: 5 (Trained without caption on 46 512x512 image for 4600 total steps)
Then trained for another 5 epochs, for 10 total epochs (9200 total steps).
Unet LR: 0.0005 Scheduler: cosine Optimizer: AdamW
Network Dim: 8 Alpha: 4
Epoch Loss
1 0.321
2 0.319
3 0.300
4 0.26
5 0.280<
Prompt: vector art. A man and a woman dancing with a polar bear.
6 0.287
7 0.278
8 0.254
9 0.244
10 0.235< lowest
Prompt: vector art. A man and a woman boxing in a circus. A black bear is the referee.