Introducing Texture Hell
This is my first publicly released sd model. It is meant specifically to create albedo/diffuse textures for use in video games and animations. It has been fully finetuned on the sd2.1 768 base model to provide you with high resolution results. Let's dive into the details.
Please read my description first before using this model, as it is rather diffrent to use than the usual sd models.
Training and prompt tips
The dataset consists of 350 high-quality images sourced from Poly Heaven, ensuring a diverse and representative range of albedo textures.
Clip 1 was used throughout training.
Prompts where comma seperated tag lists (no clip/bert full text descriptions). The most import one is "texture", followed by any material you are interested. Though because of the limited dataset, there are lacking domains (e.g. flesh and cloth). I plan to improve on diversity on subsequent versions. A full list of tags used on images with a frequency usage statistic is available -here- to help you identify potentially interesting tags. But there is probably a lot of room to experiment with other words, i was only able to do rudimentary testing.
Very Important: there are some aerial textures inside the dataset (ie. beach taken by drone from very far up) for landscape textures. If you don't want this type of texture, put "aerial" into negative prompt.
All example images have full prompt information inside them for more information about samplers, steps etc.
Tiling and WebUI
Please note that the textures generated do not perfectly tile. While there is a tiling option available in the WebUI, i do not recommend using it. Instead, i have found a "reasonable" manual workflow to turn them into seamless textures. A tutorial can be found on my Huggingface page:
https://huggingface.co/Nazzaroth2/texture-hell
I hope to improve the inherent tiling ability of the model in further versions, especially a bigger dataset and good tagging should help here.
Other Textures
My first try involded generating the full set of textures needed for the normal pbr-based render pipeline(normal, height, roughness, etc). The initial results are promising, though the reduction in resolution and limited dataset made me change my plans. This might change in the future.
In the meantime you can use other software to create these textures from the albedo image. I personally used the free option Materialize https://boundingboxsoftware.com/materialize . But there are of course also paid versions like Substance Designer.
I am curious to see what you all are able to make with this model and hope it can help you in your projects. If you have any ideas for improvements or a specific material type you think is lacking inside the model, i am happy to hear from you.
Happy texturing!
P.S. Why Texture Hell? Because hell is way more interesting than boring old heaven :3
Description
FAQ
Comments (23)
Great job!
This looks awesome! I can't wait to try it out. Could be incredibly useful
I'm really looking forward to seeing how this develops!
No config file for 2.1?
I didn't need one anymore for my "somewhat recent" version of webui (something around march was the last time i updated i think). It can recognize 2.1 v-parameterization automatically now and load it correctly. Maybe you need to keep the original 2.1 yaml file inside the models folder though.
incredibly useful with a tiling scipt i have :o ty!
Could you please share the tiling script? Really need it! THANKS a lot!! Wish you have a good day~
@KernelFunct This is one thats in the webui available extensions https://github.com/brick2face/seamless-tile-inpainting it works well, but you'll still probably want to do some more manual editing afterwards. I suggest Quixel Mixer, its free & can do all sorts of great stuff, and also whatever regular paint program you use, like Krita or whatever. The script basically does the offset technique for you, then inpaints the area internally instead of fussing with all edges, does a great job, might want to adjust its settings depending on your image size, im starting with 1024x1024 imates so I had to up it to 150px for this scripts Masked area width setting. Best to upscale after that then go tinker more in apps I mentioned above to fine tune.
nice for blender :D thank u
Look forward to future iterations
can we get a .ckpt version, i cant run the .safetensor one and i cant convert it to .ckpt for some reason
Nice job, my 1st time testing a 2.1 model, performs great & makes great diffuse textures. I had to disable all my webui extensions & add --no-half to my webui-user.bat to get it to work. My gpu is kinda old now lol, so I was pleasantly surprised & happy it works great.
I'm not really getting any results >-<
Why? It just cancels creation process
Oh, I had to use "Upcast cross attention layer to float32" in settings, nevermind.
Hey this is great, can I pay you for some work?
It's great.
Is it possible to generate correct normal maps for these textures? Can you recommend an extension?
Hey I haven't tried my self but ControlNet has NormalMap models. Wonder if you can gen a texture with this, then feed the texture through ControlNet and save the PreProcess NormalMap image. It won't be at the same resolution but maybe that can work.
Materialize does this accurately already not only that but you have full freedom in customizing the normals with Materialize; you don't need an AI model for additional textures!
RuntimeWarning: invalid value encountered in cast
x_sample = x_sample.astype(np.uint8) ??
The best one. Thank You
I've got some problems with a weird glitchy effect in the results. Do I need some specific vae or smth?
I just keep getting weird abstract results.
seamless ?
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.









