Step 1: Download the NEW 1.4 model from https://drive.yerf.org/wl/?id=EBfTrmcCCUAGaQBXVIj5lJmEhjoP1tgl
Torrent magnet: https://rentry.org/sdiffusionmagnet
Step 2: Download the cloned repo by going to https://github.com/basujindal/stable-diffusion and unzip it
Step 3: Go into the repo you downloaded and go to stable-diffusion-main/models/ldm. Create a folder called "stable-diffusion-v1". Rename the .ckpt file to "model.ckpt", and copy it into that folder you've made
Step 4: Under stable-diffusion-main/src, delete the empty folders named "clip" and "taming-transformers"
Step 5: Open environment.yaml in Notepad, and after the line saying "dependencies:", add
"- git"
The "-" should be lined up with the ones in the following lines.
Step 6: Download miniconda HERE: https://docs.conda.io/en/latest/miniconda.html. Download Miniconda 3 Windows
Step 7: Install miniconda. Install for all users. Uncheck "Register Miniconda as the system Python 3.9" unless you want to.
Step 8: Open Anaconda Prompt (miniconda3). Go to the stable-diffusion-main folder wherever you downloaded using "cd" to jump folders. or just type "cd" and then drag the folder into the Anaconda prompt.
Step 9: Run the following command: "conda env create -f environment.yaml". Make sure you are in the stable-diffusion-main folder with stuff in it. (i made that mistake lol)
Step 10: Run the following command "conda activate ldm". You'll need to do this every time you start makin' prompts (and Step 8 to get to the right folder!)
Step 11: congrats this is the gud part. to generate run
"python scripts/txt2img.py --prompt "promot" --H 512 --W 512 --seed 27 --n_iter 2 --ddim_steps 50".
(you might be able to get away with 512x640 if you have a 3080)
If you don't have a 3080 or better (10GB VRAM required), you will need to run
"python optimizedSD/optimized_txt2img.py --prompt "your prompt here" --H 512 --W 512 --seed 27 --n_iter 2 --ddim_steps 50"
You may need to lower the resolution with --W for width and --H for height if you don't have enough VRAM. Also it does not generate a random seed, so modify --seed and type random numbers if you want to get something different

tips:

-Got less than 10GB of VRAM, and want a higher resolution? You can do this if you accept half precision; saving over 2GB of vram for barely any perceptible difference
Replace your txt2img.py file in Scripts with this one: https://filebin.net/6jxnsnj3w7qsxr6j
-prompts fucking matter. I typed something bullshit "an anime gorilla" and got shit, tried the princess prompt and got kino.
-Make sure you follow this shit to the letter, it's very techy and easy to fuck up
-have fucking fun

Edit
Pub: 17 Aug 2022 03:44 UTC
Edit: 22 Aug 2022 22:26 UTC
Views: 55224