LoRA trained on 386 Images of Diffuse Texture from polyhaven and manually download and describe tags for each one of them... while respect polyhaven tags.
I shared my dreambooth config and the Training Data so that you can reproduce this LoRA, sharing also my process to train my model BerryGold on Large amount of anime images.
Note
SD1.5 can already create texture but not without shadow/reflection.
Tips
Related models
Checkpoint base on SD2.1 Texture Hell, Texture Diffusion. (withpoly, Text2Tex 1.), ESRGAN
Description
Initial release.
360 Images from Diffuse Texture from polyhaven while respect polyhaven tags.
v1.0 tags
FAQ
Comments (24)
question : how to create ORM or normal map from the result?
@niardordenravt look in description "Tips">"create AI PBR"
@DigitalDreamer ah,thank you sire
Materialize is also a great option
You might want to add diffusers or dif_textures or something to the download name. currently it is just v1.
Sure, changed name.
Does anyone know if it understands plaid? I haven't been getting any luck making SD understand I want it to give me plaid textures for a skirt, and it's very frustrating.
I wonder if controlnet could help guide it here? A few options there to set pattern and colors.
@creeduk I wouldn't know at all because I've never had luck generating nice looking patterns/textures. Especially getting textures that are the perfect blend of detailed, yet still fit in a very flat colored anime scene.
@xSatouMatsuzakax I just got my CN 1.1 all set up clean so i will try it and see if I can figure anything
@creeduk Sweet. What is CN 1.1?
@xSatouMatsuzakax ControlNet 1.1
@olivetty Ohhhh yeah makes sense
somebody made this , i dont know if this is any good for your project https://civitai.com/models/68942/felt
@miette tyvm for recommending it t me. I don't think it's what i'm looking for, however.
@xSatouMatsuzakax typically if you want https://polyhaven.com/a/fabric_pattern_07 tags will be fabric, man made, clean, green, pattern and more book, leather, brown, denim
or you can try jaks-woolitize-image-pack
@DigitalDreamer that's closer to what I'm looking for, but still incorrect. I need it to be an anime style so I can more easily put it on a 3d model without having to follow a tutorial
@xSatouMatsuzakax you could try making it more vector with vector lora, or more_details:0
I think a checkpoint would be really helpful when there's this many training images.
@A3tr Yes i know a LoRA of 5mb can't fit as much information, i may increase size but not switching on full checkpoint as i want use it to add texture detail on any image type of any checkpoint.
but feel free of anyone to train a specialized checkpoint Texture with my image and text caption in "Training Data" and share it, just mention me.
Does this model actually create tileable textures? If so, that's amazing. Yeah a checkpoint would probably be good.
@FoxDude Yes tileable work fine, all preview images are seamless texture.
There's an extension for that :)
@olivetty Do you remember the name..?
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.