Type | |
Stats | 2,787 |
Reviews | (260) |
Published | Nov 12, 2023 |
Base Model | |
Trigger Words | RealNylonFeets |
Hash | AutoV2 88C92257FC |
This Lora was originally trained for SDXL, due to the improved performance of XL release in rendering very well nyloned feet, with different nylon clothes (pantyhose, stockings), and colors (black, tan, white). If you use SDXL please look at this Lora (RealNylonFeets for SDXL)
Due to many requests received, several training runs were performed to assess if an SD 1.5 Lora should be set with a good performance.
Taking into account that with SD 1.5 the AI feet rendering is still challenging, this Lora is at the moment my best result:
please read the instructions in the GENERATION FLOW paragraph to maximize results!
Triggering prompt keyword:
RealNylonFeets
Lora weight:
A value of 0.6 to 0.9 is highly recommended. Higher values can trigger artifacts.
A base activation keyword for lora is:
<lora:RealNylonFeets:0.8>
This Lora could be used both with realistic or manga checkpoints. Realistic Vision and Henmix_real were used to generate the demo gallery.
Sampling steps:
About 40-60 sampling steps.
Additional options
- The triggering keyword should be preceded by a color. Lora was trained with black, tan and white colors, but a good prompt should render also many other colors.
Black RealNylonFeets
Tan RealNylonFeets
White RealNylonFeets
- There are some different particulars than can be rendered, like painted toenails, that can be triggered by using the following tokens:
Painted toenails
- Moreover is better to specify what kind of nylonwear is requested (Pantyhose, Stockings) adding a specific tokens:
pantyhose
stocking
For this purpose specific Loras for SD 1.5 can be added to the prompt, like:
GENERATION FLOW:
1) ControlNet is strongly recommended, during the first txt2img generation: canny or openpose approach is strongly recommended to avoid gross aberrations.
2) After a first round of txt2img, the best image should be refined using an img2img round, eventually adjusting the strength of some tokens.
3) Finally inpaint will be fundamental to refine feet details. If checkpoint include a specific inpaint model, its use is recommended.
A specific prompt is recommended for this step, focusing only on feet. For example:
<lora:RealNylonFeets:0.8> RAW Photo, black RealNylonFeets, RealNylonFeets painted toes
If toes part is affected by color aberration, in the negative prompt should be included the token:
reinforced toes
The details level of this Lora could drop in case of full body portraits (very little feet) or in case of many different Lora used together. Details can be rescued using inpaint algorithm only on feet, but in this case an upscale process is recommended if the original image is less than 1000 px.
Image dimensions
Better result are obtained with a base image of about 900 px dimension on the long image side with an image width/length ratio of 1.5. After the base generation upscaling is a good practice.
USE DEMO IMAGES PROMPTS AS REFERENCE
Final statements:
1. You are solely responsible for any creative work using this model.
2. You cannot use this model to intentionally produce or share illegal images or content.
3. Furthermore can't use this model for malicious, hurtful, defamatory, fraudulent or political purposes.