AP Profile picture
Dec 5, 2022 15 tweets 4 min read Read on X
Stop overpaying for Dreambooth training!

Here's a low-code tutorial for training your own Dreambooth models for under 50 cents 👇🏾
1/ Create a Hugging Face account! This is the ML Ops tool that we will be using 🤗
2/ Head to this link: bit.ly/3FAseN3

Click on Duplicate Space.
3/ Give your space a name, set to private, and duplicate.
4/ After a few minutes, your space will have duplicated!

Now we need to attribute a GPU. Click on the circled link.
5/ We're going to go with a T4 medium. Click on T4 medium -- here you will enter your payment details.
6/ Now we do a bit more waiting. Click on the "App" icon on the top left. This will take you back to your work space!
7/ We're running!

Now decide what you want to train. Here we will select a person. Choose which base model you want to use.

Upload the photos for training (ex. 8 photos of yourself). Make sure all images are 512x512.

Now assign a token name, use something unique like "tstr1".
8/ We can see the estimate, cheaper than 50 cents!

Beginners: I do not recommend messing with custom settings.

Make sure to check the "remove GPU". This makes sure you aren't charged after training.

Name your model, save to profile, add your HF write token, and start training.
9/ Time for a walk! Training can take anywhere from 20 minutes to hours depending on the image data set.
10/ Click on the top left HuggingFace logo to enter your activity.

Here you will see the model you just built! Exciting!

Click on the model to proceed.
11/ Now we can test our model! Type your prompt along with your instance token here:
12/ Within files, you can download the .cpkt file to upload into Automatic1111 or your UI of choice.

Now you can get to experimenting! All for cheaper than a $1!!

Compare that to the $10+ that Dreambooth services charge per model 😱
13/ If you found this useful, please help share by RTing or liking!
14/ Are you looking to implement generative AI tooling into your workflow? Fill out this form to get started today: bit.ly/3gWpRdU

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with AP

AP Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @angrypenguinPNG

Sep 19
🚨 Wan 2.2 Animate is now open-source!

If Instagirl Wan LoRA shocked you… this pushes motion + open-source video into new territory.

examples 👇
1/n: Character Animation
2/n: Character Animation
Read 6 tweets
Dec 28, 2022
✨ 3D Assets using Generative AI ✨

Here's a no-code tutorial on how you can create 3D assets using Generative AI technology 👇🏾
1/ There are two ways to do this.

To get the best accuracy, we are first going to generate an image of what we want in DreamStudio.

I used the following prompt: "3D Voxel, Rubber Duck, cute, white background"
3/ What is important is the white background. We did not get it with the first generation, but that is ok. We will use the tool remove (dot) bg

Upload the photo, and select the white background.

Then, download!
Read 10 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(