Slave Leia Outfit now for FLUX! Still learning training for FLUX so consider this a V1 and its own seperate upload for now.
Suggested Strength: 0.8-1.5 - When nf4 patching may have to go up to 2.0
Suggested tags: wearing a slave leia outfit and loin cloth
Very flexible, may need to tag bikini top, should also specify the color of things otherwise it will do it's own thing (i.e.green slave leia outfit with red loin cloth)
Other outfit details: with a long braid, with matching choker and hair ornament
Description
Slave Leia Outfit now for FLUX! Still learning training for FLUX so consider this a V1 and its own seperate upload for now.
Suggested Strength: 0.8-1.5 - When nf4 patching may have to go up to 2.0
Suggested tags: wearing a slave leia outfit and loin cloth
Very flexible, may need to tag bikini top, should also specify the color of things otherwise it will do it's own thing (i.e.green slave leia outfit with red loin cloth)
Other outfit details: with a long braid, with matching collar and hair ornament
FAQ
Comments (5)
This LoRA is great and works well with the character LoRAs I've trained. I'm interested to know more about your training process for this. How many pics did you use? What was your captioning strategy? What training parameters did you use?
Thanks! It was trained on 34 images, each of different women, and trained using the Civit Lora trainer here on-site at 325 steps. I believe I used CogVLM chat captioning that I hand edited so it was consistent. This was before Joycaption or before Civit offered detailed captioning. An example caption would be "a photo of a woman wearing a slave leia outfit with loin cloth. The woman has long dark hair styled in a braid and she is wearing makeup that accentuates her features. the background of the image is blurred with bokeh but it seems to depict a cityscape during the evening or night with lights that add to the overall ambiance of the photo"
@cranpeach Awesome, thanks for the detailed reply - I really appreciate it 😊
please do it for zit
I have a dataset ready and have made a test LoRA, but I'm just hoping for better flexibility once the Z-image Base model drops. Most I've released is one image I used in an example in my Qwen Uncrop Lora: https://civitai.com/images/115634003
Details
Available On (2 platforms)
Same model published on other platforms. May have additional downloads or version variants.