Home
Creators
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register Favorites Keys Review DMs Logout
Community
Telegram ThePornDude
Home Creators Posts Import Register Favorites Logout

5 NEW LORA MODELS (Patreon)

Published:
2023-10-12 16:22:25
Imported:
2026-06
Tags:

Content

I've recently made 5 unique LORA models for both clothing and style!

The Collection features:

The David Martinez Cyberpunk 2077 Jacket

The Giant Butterly Wings

The Sin City Movie Style

The retro 80's Anime Style

Peaky Blinders Suit

What's Inside Each Folder?

With each LORA model, you're getting a complete behind-the-scenes look into the training process:

The Model

Preview png image (to place in the LORA folder with the model)

The Json text file (auto adds the token to the prompt, to place in the LORA folder)

Training Images

Image Captions

As always, supporting me on Patreon allows me to keep creating helpful resources like this for the AI art community. Thank you for your support - now go train some awesome LORAs!👉Download CYBERPUNK JACKET👈

👉Download Giant Butterfly Wings👈

👉Download Sin City Movie Style👈

👉Download Retro 80's Anime Style👈

👉Download Peaky Blinders Suit👈

password: PATREONLORA

Files

Comments

Mauro Prelli

Thank you! May I ask why your .safetensors file is only one and if I follow your technique to create lora models I get 9 different file? Is there a way to obtain a single file (or is that a merge?)

Aitrepreneur

watch my training video again, you get multiple files because you save every epoch, it's then up to your to choose the right trained model

Mauro Prelli

Thanks mate. Yes, I had seen it, but probably hadn't fully understood it until now. I realize that I can set it from the 'Save every N epochs' option, and if it's equal to the 'Epochs', then the file it generates will be just one. I'm not clear why it makes sense to create the intermediates then, but I'll experiment with it. Thanks again!

Aitrepreneur

Because each training is different sometimes the best model is obtained after 2 epochs while others are obtained after 4 or 7 epochs, you never know without trying them all side by side and chosing the right one

Mauro Prelli

So clear mate and thanks again. Also I've setup nvidia trick you've suggested and now it's about 50 minutes to a 24 images set with high resolution preset: wow! It remains a mystery to me why the model you created of Cillian is so large in megabytes, about 800 it seems, while those I create, even at high resolution, are around 50/60mb. Have a good day!

Aitrepreneur

It's the network rank, the higher the value the bigger the lora file

章章 许

Hello entrepreneur, I have seen your LORA training for cups, and coincidentally, I am also training a sports cup. I have two questions to ask. 1. Can the technique used for clothing LORA training, where the image is cut into two halves and stored separately in a newly created folder with Repeats set to 10, be used for training cups? The purpose is to allow LORA to learn more details. Can this technique be applied to cup training? 2. When working on the sports cup, I want to add some images of people holding the cup and sports scenes. Should I create a new folder for these images, and if so, what should the Repeats be set to? Thank you for taking the time to answer my questions.

章章 许

I used 40 pictures to train LORA with a file size of 6GB, OMG! How is this going?

Aitrepreneur

probably because you made a mistake and are training a full model with dreambooth, not lora