Easy Mode Dreambooth - Google Colab Stable Diffusion

Easy Mode Dreambooth Stable Diffusion by geocine.
Really easy implementations, good for complete beginners.

Tips:
- Use 50 steps minimum when generating images if we got broken faces

Colab (latest update from geocine):
GitHub:

===================================

#dreambooth
#dreamboothstablediffusion
#stablediffusion
#stablediffusiontutorial
#stablediffusionai
#stablediffusioncolab
#stablediffusiongooglecolab
#stablediffusionaitutorial

=========================================================

Let's talk about AI on this channel

[VulcanDoodie]

I tryied tu change the MODEL_NAME: to "Linaqruf/anything-v3.0" or "prompthere/openjourney" the code sends an error that some file is not in the expected location, plz anylink to a tutorial? I just want to train specific models with my images and not just the SD1.5

[devaccount]

Wow, I've been looking for training methods, this one is amazing. Thank you for your work!

[Samosamossan]

You are amazing bro, thanks for your great work!

[andhykabayuprasetyo]

Sumpahh gilakkkk makasih banyak bangg tutorialnya sama promptnya 😭😭 anjirrr sumpahhh gilak bangettt gw cuman paham dari tutorial kamu aj bang gilakkkk maksih banyakkk tutorialnyaa 😭😭😭

[letitbleed]

hello. thank you so much! I did everything fine, but after adding another photo, I make a mistake. do i have to set it up from scratch?

[ff_sa]

very good tutorial, very easy to understand, but is there a way to convert the model in lora? or some method to create a lora?

[mixitnow]

Wow, this is amazing, for the training that would take for almost an hour, do we have to still remain active in the site to prevent time out? Or is it ok to just leave it until it finished, then we can proceed right away.

[relaxingparadise]

Hi, is the result obtained a checkpoint/model? To use it as LoRA, I need a software that does the conversion? Thank you.

[MMK]

looks awesome, i'll try it today👀

[sigma_osprey]

@nolanaatama hi. I tried this and it worked but my results looked creepy. May I know your exact prompts to achieve realistic photos? Or the guidance scale, inference steps, train steps that you used to generate realistic images?

[nolanaatama]

Issue has been fixed by @geocine! All is fine. Have fun!

[-dy]

thanks bro
edit: btw how do i use these in google colab after saving in my drive? hmm do i transfer the .ckpt file in the "models" folder after starting a model with one of ur models?

[ASonAcademy]

i've successyly traind model yesterday, now i just have ckpt file! how to run this file? my cpu do not support, now the creazt thing is i can not train my model everytime i run this colab right? please suggest.... sorry for bad eng

[hatuey]

one question : the model used is the 1.5 pruned or the 1.5 pruned ema only ? because i have somme issue when i want to use the checkpoint merger int that way A : my model B: model i want to merge ( f222 for instance) and C : 1.5 pruned ema only + the add difference . No matter if i put 0.75 or 0.8 or 0.95 in this way it's not working. I have to put in A : f222 (for instance) B: my model and summed it in 0.2 max to hope to see something is it ok ? Am i doing something wrong plz ?

[mixitnow]

Hi mate, can we change the base model or use other models like chilloutmix?

[samakrios]

Hello, is it possible to create multiple tags with multiple sets of images using the dreambooth and colab process? Like some models trained with multiple subjects and tags. Also, is it possible to create a Lora model using the Dreambooth and Colab template? Thank you

[gondrongkancong]

hi nice video tutorial, I have question two :
1. what if the object trained is not a person? lets say "a car"...so what will I text in SS-CLASS section?
2. is this setup will use our PC GPU or not? at 0:56 the video show your setup is using Tesla T4 @.@

[luckystrikehk]

kindly note to people about upload photos, first it's not working for upload the photos to /data/training_images, also the photos could not be bigger than 60kb for each photos you upload otherwise it won't work, you can keep resize the px from BIRME until upload pass.

[matoumatheux]

I would like to use my dreambooth trained model on protogen, is there a way to do it.

[andytangaming]

nice video! sorry, total noob here.. after i train my model, how do i actually use it? i mean where can i load it to and use it?