Verified: 3 months ago
SafeTensor
Type | |
Stats | 115 179 45 |
Reviews | (21) |
Published | Mar 30, 2025 |
Base Model | |
Training | Steps: 2,035 Epochs: 9 |
Usage Tips | Strength: 0.9 |
Trigger Words | flowers flower field |
Hash | AutoV2 4F59EC9252 |
Trying to improve flowers in Illustrious. Just an excuse to post girls and flowers.
recommended weight: 0.8-0.9
trigger word:
flowers
training with words:
scenery, red flower, flower, yellow flower, red flower, purple flower, pink flower, white flower, cherry blossoms, branch, white rose, depth of field, orange flower, still life, plant, camellia, hydrangea, wisteria, lotus, leaf, nature, grass
you can use the main keyword "flowers_w@orld" but it's better not to
notes:
The flower names were not marked. Only what the autotagger could recognize.
Because the selection contains many images of flowers close-up, lora sometimes draws girls with a flower instead of a head or a flower with a girl's head.
The dataset was found on the Internet in 1 hour. The result is a lot of jpeg artifacts, as we all love.