Home
Creators
Search Recent Missing 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

NEXT PATREON EXCLUSIVE LORA (Patreon)

Published:
2023-09-10 17:49:03
Imported:
2026-06
Tags:

Poll

NEXT PATREON EXCLUSIVE LORA

  • Movie Style LORA 254
  • Video Game Style LORA 138
  • Anime Style LORA 181
  • Other (Let me know in the comments) 54
  • 2023-09-10
  • 627 votes
{'title': 'NEXT PATREON EXCLUSIVE LORA', 'choices': [{'text': 'Movie Style LORA', 'votes': 254}, {'text': 'Video Game Style LORA', 'votes': 138}, {'text': 'Anime Style LORA', 'votes': 181}, {'text': 'Other (Let me know in the comments)', 'votes': 54}], 'closes_at': None, 'created_at': datetime.datetime(2023, 9, 10, 12, 28, 6, tzinfo=datetime.timezone.utc), 'description': None, 'allow_multiple': False, 'total_votes': 627}

Content

Let me know what LORA you wanna see!

Comments

Kynlif (edited)

Comment edits

2026-09-08 20:00 Thanks for the different shares today! It would be great to have some info on LoRa creation for example for clothes (only take the look of clothes from a video game or movie without taking all the visual style in the LoRa), or for specific objects or details (swords, shields, or everyday objects, create a LoRa to create Orc fangs on anything and everything, etc). Thx :D
2023-09-10 18:06

Alex Kuu

I like the idea of KL. That would be helpful. How to teach LoRA to complete drawing one element more or less getting into the right places in the image. How many single parts can be placed in one LoRA?

Nicholas

Other: using Kohya to train an SDXL model.

Ken Chapman

object, non human....Plane for example

Kendrick Lau

Outfit? maybe outfit of characters from video games or movies.

Hasan Als

a character lora settings with 10-20 mins training time

Donald Diffusion

Objects like sunglasses, clothing, a sword, guns etc.

baran mengül

SDXL TO SD VAE does not work If you have been encountering this situation, it automatically switches to default after selection. SDXL VAE does not work. Does anyone have experience ?

baran mengül

the possibility to modify a specific human illustrator, to modify our own ilistrator and photoshop designs, to see different variations of the creative without breaking the line

S.LANGEVIN

A object based on image sample!!

Zergits

Ben 10 omniverse style 🙏

Salvador Robles

Hello, of course training LORA is very interesting (I would like a realistic-anime style improving the style of this artist for instance: serafleur), also, It would be great begin with video AI supported as shown in this video: https://www.youtube.com/watch?v=FQ6z90MuURM thanks so much, great job Aitrepreneur

Aitrepreneur

Unfortunately I will not train models on an actual specific artist and distribute it on my patreon, that wouldn't be right, but if you want to, you can do it yourself following my latest style training lora video

Salvador Robles

Thanks, I will. Are you thinking about some video AI tutorials (warpfusion etc...)?thanks

Zac Doyle

Just getting around to watching your latest video and I think you might find a small side project I've been working on to be useful. https://github.com/Z-L-D/ImageTagJS It's a little rough still but it has certainly saved me a fair bit of time. Also, the preview image is pretty out of date. Its quite a bit further along than that looks.

Semon Xue

I want to know how to add a custom object into the lora, such as a new model car.

Doge Coin

Can you make a Google Colab SDXL Lora Training video

Doge Coin

Or a Google Colab Textual inversion SDXL Training video

Mike Menders

Pose Lora with special camera angle (from below or from top)

5iwy

Lora of object which is not well known by model. Object which is useful and usable by human (one or more at the same time) in various scenarios.

H. Wren

Photographic style : )

Aitrepreneur

I will be making more lora training videos in the future too for that as well

5iwy

Great! Couse not many vids are about objects, and how to cooperate with them .

omnidox

A Lora for Professional Photography would be interesting. Like National Geographic level photography.

Brad Taylor

how about a style that creates realistic reference photos that can be used for making 3D models in blender using Keentools Facebuilder or FaceGen for Daz3d. So ultrarealistic, hair up and not blocking any features, neutral expression, no makeup, posed in perfect angle from all common reference directions, even full lighting, same focal length, etc

335

Gundam Lora

brux Zhang

How about an anime character Lora?

Mantis

I would like anime LoRAs. However, most of them use the really generic modern anime artstyle. I'd love to see a LoRA based on anime from the 1980s or the 1970s. That would be so unique.

Javi

A Lora of a product

LW

In your video (https://www.youtube.com/watch?v=N_zhQSx2Q3c) {very useful, btw}, you mention that it is not necessary to crop training images to 1024x1024 for training a LoRa for SDXL. You have supplied a whole bunch of regularisation images for training which are cropped to 1024x1024. {thanks, these will come in very handy}

LW

oops... ends the input...

LW

Continuing.... My question is : Is it necessary to use cropped square images for the regularisation images?

LW

So, for example, if I want to make a LoRa for a specific giraffe, most images will not be square. Is it possible to use non-square images of giraffes as the regularisation images?

Aitrepreneur

I think you need to crop the images to the training resolution of the model you need to train but I haven't tried it without cropping them, maybe that will work. However in your case good news, if you are not training a person you don't need reg images for LORA, I only saw good results when using them for people training, the rest, the final results are negligible, they are not useless but they won't make or break your model.

baran mengül

comfy uı sdxl + controlnet sample on runpod

LW

I'll give it a try with non-square reg images.... SDXL may surprise me :)

Aitrepreneur

sure try it out, definitely let me know your findings if it worked well, I'm always interested to know more :)

Stephen Berry

Lora of Troye Sivan or Joe Locke

Greg Rink

Conceptual training

Erik Toscano

An object Lora, like a pair of Nike sneakers

ARAXA

Objects! Shoes, purses, etc