From 9c1d7205f4be4322dbdae4a9e5efbfc41056bccb Mon Sep 17 00:00:00 2001 From: JoePenna <100188076+JoePenna@users.noreply.github.com> Date: Mon, 26 Sep 2022 17:32:52 -0700 Subject: [PATCH 1/2] Update README.md --- README.md | 26 ++++++++++++++++++++++++-- 1 file changed, 24 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 5d89adb..f03f570 100644 --- a/README.md +++ b/README.md @@ -39,8 +39,30 @@ Now, if you wanna try to do this... please read the warnings below first: - There doesn't seem to be an easy way to train two subjects consecutively. You will end up with an 11-12GB. - I'm currently testing ways of compressing that down to ~2GB. -- You might have better luck training with `sd-v1-4-full-ema.ckpt` - - However, it's huge and it's annoying. +- ~~You might have better luck training with `sd-v1-4-full-ema.ckpt`~~ + - Further testing has shown that it doesn't help. + +- Best practice is to change the token to a celebrity name. Here's my wife trained with the exact same settings, except for the token: + +The `ground truth` (real picture, caution: very beautiful woman) +
+ +Same prompt for all of these images below: + +`sks` +
+ + +`woman` +
+ + +`Natalie Portman` +
+ +`Kate Mara` +
+ # Vast.AI Instructions - Sign up for [Vast.AI](https://vast.ai/) From f0684ab3796be1e518ff455b0e9b5057b7c6e047 Mon Sep 17 00:00:00 2001 From: JoePenna <100188076+JoePenna@users.noreply.github.com> Date: Mon, 26 Sep 2022 19:15:21 -0700 Subject: [PATCH 2/2] Update README.md --- README.md | 23 +++++++++++------------ 1 file changed, 11 insertions(+), 12 deletions(-) diff --git a/README.md b/README.md index f03f570..16daf2e 100644 --- a/README.md +++ b/README.md @@ -63,6 +63,17 @@ Same prompt for all of these images below: `Kate Mara`
+# RunPod Instructions +- Sign up for RunPod. Feel free to use my [referral link here](https://runpod.io?ref=n8yfwyum), so that I don't have to pay for it (but you do). +- Click **Deploy** on either `SECURE CLOUD` or `COMMUNITY CLOUD` +- Click `Select` on a GPU with at least 35 GB of VRAM (e.g. A100, A40, A6000, etc) +- Select a template > `Runpod / Stable Diffusion` +- Click `Connect` and choose `Jupyter Lab` +- Make a new notebook (it's just like Google Colab) and run the code below +```python +!git clone https://github.com/JoePenna/Dreambooth-Stable-Diffusion/ +``` +- With the file navigator on the left, `/workspace/Dreambooth-Stable-Diffusion/dreambooth_runpod_joepenna.ipynb` -- follow the instructions in there. # Vast.AI Instructions - Sign up for [Vast.AI](https://vast.ai/) @@ -86,18 +97,6 @@ Same prompt for all of these images below: - ![img.png](readme-images/vast-ai-step6-open-notebook.png) - Follow the instructions in the workbook and start training -# RunPod Instructions -- Sign up for RunPod. Feel free to use my [referral link here](https://runpod.io?ref=n8yfwyum), so that I don't have to pay for it (but you do). -- Click **Deploy** on either `SECURE CLOUD` or `COMMUNITY CLOUD` -- Click `Select` on a GPU with at least 35 GB of VRAM (e.g. A100, A40, A6000, etc) -- Select a template > `Runpod / Stable Diffusion` -- Click `Connect` and choose `Jupyter Lab` -- Make a new notebook (it's just like Google Colab) and run the code below -```python -!git clone https://github.com/JoePenna/Dreambooth-Stable-Diffusion/ -``` -- With the file navigator on the left, `/workspace/Dreambooth-Stable-Diffusion/dreambooth_runpod_joepenna.ipynb` -- follow the instructions in there. - # Original Readme from XavierXiao This is an implementtaion of Google's [Dreambooth](https://arxiv.org/abs/2208.12242) with [Stable Diffusion](https://github.com/CompVis/stable-diffusion). The original Dreambooth is based on [Imagen](https://imagen.research.google/) text-to-image model. However, neither the model nor the pre-trained weights of Imagen is available. To enable people to fine-tune a text-to-image model with a few examples, I implemented the idea of Dreambooth on Stable diffusion.