Photorealistic LoRA of an American actress Bryce Dallas Howard | 🇺🇸
Born 2 March 1981 in Los Angeles, California, USA.
Suggested LoRA weight: 0.8-1.0
Recommend using Adetailer for the better quality.
Some showcase images are generated using ControlNet.
Please share your review and results if you like this model.
It's not perfect but hope you like it.
Disclaimer: The AI image model provided here is not intended for the creation or generation of explicit or adult content, including but not limited to NSFW (Not Safe For Work) material. It is designed for general-purpose image-related tasks within ethical and appropriate boundaries. Any misuse or inappropriate use of the AI model for generating explicit content is strictly discouraged and goes against the intended purpose of this tool.
Description
New training workflow
Trained with 100 images of the actress
New trigger words
FAQ
Comments (3)
Hi, great lora. Question, what is your 'new' method? I noticed you split your dataset into 2 folders, is that to give weight (more repeats) to the better pictures? You only train with 512x512 pics? You don't move up to 768? Not worth it for SD 1.5? Sorry, one last question, you train against Serenity? Not the base SD 1.5? Sorry for all the questions but the quality of your Loras is quite amazing.
1. I just follow this guideline here. I'm not really that educated with this ai drawing stuff so I only follow some easy guidelines that someone posts and improvise them. Here's the link: Celebrity LoRA Training Guide | Civitai
2. I train my loras with 768x768 only for the faces and 512x768 for the body.
3. I train using Serenity because I think it gives me the most realistic skin texture for the results. But maybe there are better models to train with out there somewhere.
4. Thank you for the good comment ❤️
@y_k_0C Thank you for the answers, you've given me some new ideas when training my Loras. I never thought about using 2 different folders w/ the same trigger but with different number of repeats to give weight to the faces. I use Serenity all the time to test my Loras and generate but never thought about using it to train until looking at your config. Thanks again for all the info!