THIS AI CAN DRAW ME!!! - Dreambooth & Stable Diffusion

These pictures aren’t photoshopped pictures of me, no one created them. They come from two artificial intelligence tools, called Stable diffusion and Dreambooth. Let me explain.

Stable Diffusion is a machine learning tool that takes a noise pattern and uses a neural network to pull an image from the noise. It’s pretty magical really, and allows you to create some amazing shots. Dreambooth is another machine learning tool that takes in images of a person or even a whole art style and trains a model to recognise and recreate the person or style. Combining the two means you can put yourself into the stable diffusion images - but to make it clear, this isn’t just fancy photoshop. This isn’t taking an actual picture of me and morphing it onto something else - this is a neural network that actively understands my face and can recreate it in any style or form I ask it to.

In this video, I want to give a run down of how you can set this up on your own and talk about my experience of playing around with the software for the last week or so. This is very much what you’d call “bleeding edge” tech, so anything I show here will be out of date by some time next week so it’s not worth going too deep into a how-to, but here’s a brief overview.

Use referral code "techteamgb20" when signing up!

As an Amazon Associate I earn from qualifying purchases, using the links below or other Amazon affiliate links here.

Bitcoin donations: 1PqsJeJsDbNEECjCKbQ2DsQxJWYqmTvt4E

- My PC - AMAZON AFFILIATE LINKS -

About TechteamGB:

TechteamGB is a long-running tech channel focused on high quality videos on PC Hardware and Consumer Electronics. We work with nearly all of the big (and small) names in technology, from Intel and AMD, to MSI, Asus, Gigabyte and more! If you have any questions, feel free to leave a comment! If you are a tech brand with some new and exciting products, email Andrew at the address below - please include details of the products you would like reviewed/showcased. Also, if you are a brand looking to sponsor/advertise, please make this clear in your email. Otherwise, enjoy the videos! Oh, and don't forget to leave a comment on the first video of mine you watch and say hello - I'd love to chat with you!

THIS AI CAN DRAW ME!!! - Dreambooth & Stable Diffusion

TechteamGB

[BenKDesigns]

Extension developer here!

A few points of note:

1. You can now train 768 size images using the V2x checkpoints.
2. If it's broken on install, restart the whole UI to make it install dependencies.
3. Instead of training steps - use epochs. Epoch=number of steps per input image. 100 epochs is a good place to start.
4. Caption your input images, and give it a wider variety of images. Not just face, but a nice mix of face/torso+face/fullbody.
5. Don't just crop images to square. Padding them so they are in their respective formats will help with generating images of that orientation.

[Avenger]

For inputted pictures, it definitely helps to have different lighting settings, different angles, taken from different distances and different backgrounds (also different clothing). Results are typically quite poor when there aren't too much variation between the pictures.

[Avenger]

If your name is generic, you may run into issues. Generally people recommend that you use a rare token like "zqz" or something that the model doesn't know at all. If you use something like "John", then the AI won't really know what "John" you're referring to and your results might be less consistent or just overall worse.

[shadesofbeige]

Going to try it with my arc A770... Pray for me.

[MrEMeat-kktc]

5:30 No, Dreambooth! I said “TechteamGB”, not “Dawid Does Tech Stuff”!

[shadesofbeige]

I think this is a good use case for a Linode free trial credit 😉