How to Use Google Colab to Run Stable Diffusion (for free, without coding experience or a fancy GPU)

In this video I walk you through how to use the Stable Diffusion image generation model to generate images. You don't have to know how to code or own a fancy computer.
Here are some links that are relevant:
The Colab notebook: colab.research...
The usage license: huggingface.co...
Link to join Hugging Face: huggingface.co...
Where you can get your Hugging Face token: huggingface.co...
More info on the model: huggingface.co...
An article on setting up Stable Diffusion locally: / run-stable-diffusion-i...

[@lazydog6642]

After installing 20 gigs of Easy Diffusion and attempting to redownload the 1.5 gig model it comes with when it wouldn't run I figured out that my PC is too low spec to run it (after deleting redownloading and reinstalling because I thought it might just be corrupt). After all that I stumbled onto your video and in 5 minutes I accomplished what I was trying to do all day and I finally made an ai image. Thank you for the video

[@asalvani01]

none of the stuff you say comes up

[@serpaolo7413]

I'm getting error while running the code "TypeError: 'bool' object is not iterable"

[@GlaciusTS]

I’m not seeing the “access repository” thing.

[@hearinko]

not seeing it either

[@sparklee5044]

me neither

[@SimplyAIA]

hey guys i had the same problem. the video is a bit outdated so now there is no need for the "access repository" step. you're doing great! just keep going and follow along with the video! Everything works out nice at the end! Its a lot of fun

[@augustasmaniusis2999]

when I change safety checker to true I get error when generating image, can I fix that somehow?

[@digitalaiartist]

The link to the licence doesn`t seem to take you to agreement button ? thank you

[@cylisaether5077]

sir your link doesnt work, please fix

[@mindseyeproductions8798]

Thank you, relaxed and provided a good level of confidence in explaining the procedures.

[@mjfII]

Any way to use an image for helping the prompt?????

[@brosjay2131]

please how about image to image translation

[@mahmoud_ali_963]

Thanks for this great tutorial❤❤❤. I have a question: How can I enlarge the images I get?

[@thedigitaldilettante6681]

Glad you like it! Thanks for the question. I know there are other independent ways to upscale images you could look into. If you want to generate higher quality images you could check out this colab: colab.research.google.com/github//wakamenori/txt2imghd-colab/blob/master/txt2imghd.ipynb If it isn't clear, let me know and I will try to make a tutorial for the HD colab.

[@mahmoud_ali_963]

Thank you very very very much man . you are the best😄❤️❤️

[@lenovotablet2417]

Great explanation. Thanks!

[@rexmax2502]

Is there anyway to change the model?

[@paulmercy5455]

Tbe safety checker remove did not work for me but adding pipe.safety_checker = None pipe.requires_safety_checker = False Did:)

[@Revon-Feuer]

has not worked for me x3

[@every2007]

Thank you! This is fantastic!

[@peterusher2020]

I use several Colab notebooks that do not require Huggingface tokens. However, for them to run quickly and efficiently, I need a Google pro subscription. Your image generation was fast. Was it done with a free account or a paid subscription?

[@thedigitaldilettante6681]

It was done with a free account! I'm not sure but my guess is that the ones that avoid using tokens have to import and install a lot more stuff so they take more time. Using tokens makes sense unless you are worried about violating their use policy or something imo.

[@justaperson4809]

Thanks a lot!

[@PhylumVertebrae]

Can I use LoRA models on colab?

[@NineSeptims]

Should be able to.

[@FeralBoarding]

Awesome! Is there a way to use the image prompt like in the dream studio version?

[@thedigitaldilettante6681]

Hey Brye-life, do you mean using an image as input and having the model improve it? If so then yes there is a way, and I hope to post a video on that in the future!

[@ostyslav]

i try to change safe save from False to True. and i have error in 5 stroke, in promt window. but if start like false - all works (withouts nsfw) how to fix? *3*

[@cartergamegeek]

Each time I try local I get errors. I'm stuck on a torch GPU error. I feel like an idiot because I can't seem to fix it. My computer is likely just not built to host this.

[@bayuadrian1638]

Hello, can you spesificly how to get free license for model or how can i bring up a pages like this 3:59? Thank you

[@celatugamerboy4863]

How do you accept the terms on huggingface? I created an account but can't find any button.

[@komekko373]

Can I use my own Model cpkt

[@rdxpopex]

Please tell how to run other models like this??

[@shirohitsugi3809]

Is it possible to enter negative prompt in this notebook?

[@kerhabplays]

6:05 I did everything as you say but I'm struggling passing that error thing...HELP edit: it's fixed

[@criticalboot2567]

what did you do to accept the terms? its not showing up for me

[@kerhabplays]

@@criticalboot2567 Create complete account and scroll down then just use your young eye fr(I mean you ain't an elder) you'll find a check box

[@nishantmudgal4434]

Collab has too much errors bro....tired of it

[@NineSeptims]

What errors are you getting?

[@nishantmudgal4434]

@@NineSeptims it uses this "Gradio"....and after 5-6 minutes, it just says "No interface is running currently" and I have to connect the Google drive and all everything again, and this time, after 2-3 minutes, it again breaks off...

[@bambinoesu]

can't even connect to my google drive online to do this

[@Human_Evolution-]

Do you have an img2img video?

[@gins5774]

how do i add negative prompt, is possible?

[@hodl2123]

Thank you for the easy tutorial, I have 0 knowledge in programming and I was able to get it to work. Are we able to use this with other stable diffusion models like controlnet? Thank you

[@DONUTnews]

Incredibly helpful, thank you.

[@buttonsplaymusic4896]

I liked that you put a bit of an explanation of what Colab is in your video, which most haven't. I still don't understand some things about colabs though. Future video? Like I understand image generation requires VRAM, right? so where is the collab image generation actually taking place? A random or specific server somehwere? And how does that work with your machine connected just by a browser?

[@FeralBoarding]

I'm assuming it's in the google server dungeons.

[@thedigitaldilettante6681]

Yeah basically, google is hosting the compute power, which is why Colab is a good tool if you don't have a fancy GPU set up for machine learning. I might post a future video about that but I am thinking my next one will be on fine tuning stable diffusion using textual inversion to learn new concepts/objects/styles.

[@buttonsplaymusic4896]

@@thedigitaldilettante6681 That sounds really interesting

[@Ulexcool]

Are you deaf bro? 0:46

[@l0rd258]

​@@Ulexcool wow wow wow, calm down boy

[@zefilla4306]

good tutorial thanks alot

[@ak_mits]

Can we use this to traind SD for faces? If yes, where shall I start from? The menus seem limited on this colab page

[@samanthaunderhill9191]

I love it and your voice makes me happy 😊

[@sifu2u_now]

Don't we have to pay for the usage of Google GPU? Great Tutorial. 👍🏽

[@sildona]

thank you for this tutorial!!!

[@Human_Evolution-]

You are so straight edge. Gotta read docs to not harm people, gotta keep NSFW on. What is it like living like that? I am always the opposite.