For training a model with dreambooth in replicate, you can make one theoretically her but it's not work on this web UI.
So, i made my own dashboard for create a output.zip
and after use this output for create picture with Stable Diffusion.
You can see DEMO of this dashboard.
First, you need you clone this project in your computer and make following command :
npm i
npm run dev
# or
yarn add
yarn dev
# or
pnpm i
pnpm dev
Open http://localhost:3000 with your browser to see the result.
Now you can enter a replicate API key, select CKPT you like and put a direct link download .zip where you have picture for training your model.
You select after that all parameters you need ( Instance prompt... ) and click in Train model
and wait until in Link Model
you get a link where you will get your model !
I'll explain how to add a CKPT link using the example of OpenJourney V4, which has the link :
https://huggingface.co/prompthero/openjourney-v4/resolve/main/openjourney-v4.ckpt
Go to the following javascript file app/page.js
and at the beginning of the code from line 9 to 21 you will have 2 lists where you have to add the name and the link of your CKPT link like this :
const people = [
{ id: 1, name: 'Analog Diffusion' },
{ id: 2, name: 'Dreamlike Photoreal 2.0' },
{ id: 3, name: 'Realistic Vision 2.0' },
{ id: 4, name: 'Realistic Vision 3.0' },
{ id: 5, name: 'OpenJourney 4.0' },
]
const CKPT = [
{ id: 1, cpkt:'https://huggingface.co/wavymulder/Analog-Diffusion/resolve/main/analog-diffusion-1.0.ckpt' },
{ id: 2, cpkt:'https://huggingface.co/dreamlike-art/dreamlike-photoreal-2.0/resolve/main/dreamlike-photoreal-2.0.ckpt'},
{ id: 3, cpkt:'https://huggingface.co/SG161222/Realistic_Vision_V2.0/resolve/main/Realistic_Vision_V2.0.ckpt' },
{ id: 4, cpkt:'https://huggingface.co/SG161222/Realistic_Vision_V3.0_VAE/resolve/main/Realistic_Vision_V3.0.ckpt' },
{ id: 5, cpkt:'https://huggingface.co/prompthero/openjourney-v4/resolve/main/openjourney-v4.ckpt' }
]
Finally, go to code line 103 in the DataJson(event)
function, where you need to add your CKPT to the switch, which for me is CKPT[4].ckpt
. Here's how it works:
switch (event.target[1].value) {
case "Analog Diffusion":
ckptlink = CKPT[0].cpkt
break;
case "Dreamlike Photoreal 2.0":
ckptlink = CKPT[1].cpkt
break;
case "Realistic Vision 2.0":
ckptlink = CKPT[2].cpkt
break;
case "Realistic Vision 3.0":
ckptlink = CKPT[3].cpkt
break;
case "OpenJourney 4.0":
ckptlink = CKPT[4].cpkt
break;
default:
ckptlink = "ERREUR"
break;
}
And that's it, you can now use CKPT for your DreamBooth face training.
To learn more about this project, take a look at the following resources:
- Next.js Documentation - learn about Next.js features and API.
- Replicate Documentation - Create and use Replicate API.