Downloads · 30 days
350
5% of all-time downloads
TheLastBen/hrrzg-style-768px
hrrzg-style-768px is a text-to-image model from TheLastBen. Use it when you need an image from a text prompt. It is set up for diffusers. The card lists the license as creativeml-openrail-m.
This model is trained on Fred Herzog's work using 30 images 768x768, 12000 steps, 1k steps text encoder at the beginning and 500 steps at the end.
Downloads · 30 days
350
5% of all-time downloads
All-time downloads
7.1K
Public
Repo size
19.5 GB
Likes
89
Public
Click a slice to open those files.
.bin5.5 GB · 56%
From the Hugging Face model README
This model is trained on Fred Herzog's work using 30 images 768x768, 12000 steps, 1k steps text encoder at the beginning and 500 steps at the end.
close up portrait of beautiful young woman in a bus by hrrzg
Negative prompt: low quality, fake, painting, greyscale, night
Steps: 30, Sampler: DPM++ 2S a Karras, CFG scale: 7.5, Size: 896x768, Model hash: e93cb7f3, Denoising strength: 0.7, First pass size: 640x512
beautiful, ultra detailed, cinematic, ((sharp focus)), intricate details, micro details, city, street
Negative prompt: low quality, fake, painting, greyscale, painting, bokeh
Steps: 45, Sampler: DPM++ 2S a Karras, CFG scale: 8.5, Size: 768x768, Model hash: e93cb7f3, Denoising strength: 0.7, First pass size: 512x640
Always check the Highres.fix box and keep the resolution above 768x768, if you use the terms "city" "road" "street" "urban", you don't have to include the instance "hrrzg"
You can also train your own concepts and upload them to the library by using fast-DreamBooth.
Test the concept via A1111 Colab :fast-stable-diffusion-A1111
.png)