The model is trained using a dedicated dataset, focusing on black stockings with lace tops. The first aim was to improve the rendering of black stockings for the Realistic Vision models, where stockings show often problems with lace top appearance.
_____________________________________________________________________________
IMPORTANT: The serie RealStockings is made by 3 different Loras:
- BlackRealStockings, this Lora;
- WhiteRealStockings, specific for white lace top stockings, downloadable at:
https://civarchive.com/models/118574/whiterealstockings-lace-top-white-stockings
- TanRealStockings, specific for tan lace top stockings, downloadable at:
https://civarchive.com/models/111186/tanrealstockings-lace-top-sheer-stockings
All three Loras can be used together in the same image and make a sort of bundle, providing the same outfit for stockings of different colors.
_____________________________________________________________________________
BlackRealStockings is based on SD 1.5, is optimized for Realistic Vision (checked also with 5.0) but also gives good results with other models, by changing the weight of the model or the weight of the trigger word.
Activating Lora:
<lora:BlackRealStockings:1>Trigger word:
BlackRealStockings
To ensure that lace top bands are visible add in the prompt:
visibile BlackRealStockings top lace bands
To avoid the rendering of garter belt or suspenders write in the negative prompt:
(suspenders:1.3), (garter belt:1.3),Sampling steps recommended: 25-50
CFG Scale: 7
Sampler Suggested:
DPM ++ 2M SDE KARRAS for Realistic Vision
Euler a for non realistic models.
CHECKPOINTS TESTED
Model and trigger word weight
REALISTIC VISION V2.0, V3.0, V4.0 and V5.0
weight model suggested: 1
trigger word weight: 1-1.1
If prompt is very long and structured, inpainting is suggested to obtain a more effective result, especially on lace top bands.
Effectiveness on more checkpoints will be tested.
On Realistic vision this Lora is compatible with Opentoe heels/sandals with legwears. The use of Opentoe lora is reccomended with inpainting algorithm.
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.
Description
FAQ
Comments (2)
Periodically, stockings can disappear and leave only a strip. Besides visibile BlackRealStockings top lace bands used legs, sexy thighs,sexy hips,black stockings, ,large,fullbody,
Thanks for yor post and your test. I love it!!!
Concerning the stocking disappearing when I tested the lora I made many batch of images in very different poses and I noticed than in about 10-15% of images the rendering was not optimal at the first txt2txt try. In those cases firstly I tried to retrain th lora implementing dataset. I've reduced about 5% this problem, but occasionally it persists.
So i keep the implemented lora (this one) and in case of bad rendering I used another image in the same batch with good rendering to refine image or, if I liked exactly the image with bad rendering I used a single inpaint step and in all cases I gave back a good stocking rendering.
Another possibility may be linked to a use of certain checkpoints: in this case I reccomend to all users to change first the strength of tringger word until 1.3 and in case of need the strength of lora until 1.2 to check some valuable implementations of this lora for models i can't have possibility to test.
I hope I have been helpful to all the users that could appreciate this Loras.
To Fansay my congrats for images and accurate testing. Thank you!!!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.


