5-27-2024: Gradio service having issues with shared links. View status of Gradio service here: status.gradio.app/ ---- If you run into issues with failed username and password when downloading LoRAs or checkpoints you may need to get and add an API to the URL. On Civitai website in the account settings add an API key and then you can add that at the end of the download link with ?token=YOUR_API_KEY - If you find the video useful and would like to tip you can buy me some electricity for all these image generations. It is greatly appreciated! ko-fi.com/kleebztech
Civitai website. The first part is where it will save it and the second part is the link to download from which I got from Civitai although you can do what I do now and upload my models to Google Drive and share a link which I use. @@livealive7207
@@Zendarkdagger you can't add more than what already comes with Fooocus. In the Input Image section you have options like pyracanny and CPDS for example.
you are the man. Many people complicate tutorials instead of explaining them right at the beginning, you already taught everything right away, thank you very much!
@@KLEEBZTECH haha, yeah. I ended up getting an error here because the T4 gpu is not active and in the end an NVIDIA website appears to download drivers. I keep getting this same error, what do I do?
@@KLEEBZTECH Yes, I was functioning. So does that mean it can go back to normal the next day? And yes, I used it several times because I hadn't gotten a good image.The strange thing is that there seems to be no way to turn on the GPU.
you know how i know your video is the best? because you mention that you know that what youre doing may not be the best and if others know better then to mention. nobody knows "the best"of anything. collaboratively we all know things and the best is found when we all collaboratively share what the best is that we each have figured out. cant stand youtube videos that say "the best way to do xyz". nobody is all knowing. and someone willing to say that, is someone i want to learn from, becuz they are always willing to learn more. thank you :)
Thanks. That really made my day. I have been having difficulty getting videos out lately dues to some real world issues but also have felt I don't know enough about certain things to put the videos out. This really gave me a little boost.
i also really appreciate seeing you do the generations in real time showing mistakes as they happen. it helps bcuz i too will have mistakes and seeing how u go about solving them or rather troubleshooting them helps tremendously w real life applications. id love to see a less edited video of your workflow of "experiementing" to figure out how to accomplish some goal. for example, if u set out to create xyz doign xyz, then just show you experimenting until you reach that goal. or maybe you dont reach it but u learn stuff along the way. thatd be helpful i think, sort of teaching how to "experiment" w the settings, how u decide what setting to experiment w next.
@@KLEEBZTECH The basic point I learned from you, which I don't know, is that how can we change or edit the queries before run in google colab..... 👍🌹❤️
fooocus gives an error when I use another model, but if I use the default model: "juggernautXL_v8Rundiffusion.safetensors" then everything works without errors I select another model press "generate" and the generation hangs at the moment "Loading models ..." and then gives an error how to fix this?
I am still learning to use on Colab so hopefully didn't miss anything when it comes to the basics. Since I have issues recording when upscaling I am planning on using when doing videos that use any upscaling.
@@KLEEBZTECH Only the free version, I consider the paid version not worth it because I discovered other options available to run Stable Diffusion in the cloud, for example SageMaker Studio Lab and Kaggle. Both have similar limitations to Google Colab and I change from tool to tool every time it exceeds its use limit. I was focused on influencer modeling with A.I. for three months, the profits were minimal so now I will move on to Children's Videos with A.I., Stable Diffusion is a great tool
@@KLEEBZTECH Only the free version, I consider the paid version not worth it because I discovered other options available to run Stable Diffusion in the cloud, for example SageMaker Studio Lab and Kaggle. Both have similar limitations to Google Colab and I change from tool to tool every time it exceeds its use limit. I was focused on influencer modeling with A.I. for three months, the profits were minimal so now I will move on to Children's Videos with A.I., Stable Diffusion is a great tool
@@KLEEBZTECH Only the free version, I consider the paid version not worth it because I discovered other options available to run Stable Diffusion in the cloud, for example SageMaker and Kaggle. Both have similar limitations to Google Colab and change from tool to tool every time it exceeds its usage limit. I focused on influencer modeling with A.I. for three months the profits were minimal so now I will move to Videos with A.I. Stable Diffusion is a great tool
Wow. Didn't expect Fooocus can be used in Collab. Could you please tell me how do I know what console software can work with Collab? Something like rembg or like that. What should I consider before trying to run an app in Collab? Despite its resources limits, of course. Are all console software work in Collab that works fine under Linux, for example?
Actually just updated my pinned comment and my description to mention that Gradio which provides that link seems to be having issues today. Currently says working but if you look they have been having issues all day. Hopefully that is what the problem is. status.gradio.app/793595965
Thanks so much for the video! I am getting invalid username/password come up when trying to run the juggernaut model with the download path, any idea why?
I have added more info to the pinned comment about that. Might need API key but also have heard mention that it may not work directly now. I found a better option personally. I upload my checkpoints to Google Drive and then share a link and enter that link for the download. It then does it instantly. Usually even a free Google account should have enough space in Google Drive for at least one checkpoint.
I did just test with mine and the one I normally download is still working. Have you tried more than one? I think it could also be model dependent and the settings the person who uploaded it decide on. Also double check for typos. for example this is a link to one I just tested: civitai.com/api/download/models/454069?token=$XXXXXXXXXXXXXXXXXXXXXXXXXXXXX
What about refiners? I heard I can use SD 1.5 models as refiners but I can't get it to work. When I run the WebUI, it doesn't show in the list of refiners.
Everything runs, but when I'm in Foocus and want to switch to the checkpoint, in the Base Model bar, it doesn't show me the checkpoint I downloaded. Somebody help pls.
This might help. I have not tried using myself but appears the same as with LoRAs so the same method should work. github.com/lllyasviel/Fooocus/issues/297
Unless you just run the default each time. If you want to change anything it makes it so you don't have to make those changes each time. Yes the environment will be destroyed each time but anything custom like the commands to download the checkpoints or LoRAs will stay. So when I open mine and want to use the anime preset I don't need to change that line since I have the different options already set up.
Same as with the checkpoints. If you look you will see that it is saving it in the lora folder. But you may need to include an API key. Check out the pinned comment there should be a link to more on that. Apparently some you need an API key for. I had only run into one before and didn't realize it because the rest all worked for me.
You would need to have more than one model in Colab and then choose in the model tab at the top but on the free version of Colab you will possibly run out of VRAM.
can you run the flux model on this? I get an error on Foocus colab saying *"Username/Password Authentication Failed."* That's when i try and run the Flux model from civtai. But no problems with the model you used from the video
@KLEEBZTECH I'm trying to use a checkpoint in fooocus Google colab but civit ai recommends a VAE for the checkpoint so I tried using the API key method I've been using to import checkpoints and loras from civit ai to colab but it's still requires authentication even thought the API key is provided. I downloaded it and placed it in Google drive and tried to fetch it from there and it still request authentication. I need the VAE to get descent photos so Please what do I do?
Have you tried uploading it to the correct folder in Colab manually? Even LoRAs I have manually uploaded to the folder. Checkpoints don't seem to work because of size.
@@KLEEBZTECH I tried uploading it manually and after selecting it in the VAE section to use it?. It says the VAE file is corrupted. I don't know if it's because I manually uploaded it or something is wrong
I would possible go onto the Github discussion section for Fooocus and see if anyone might be able to help. I am not sure and I have not run into that.
"--theme dark" flag can be added to launch command or you can do what I prefer which is to change my browser to dark mode in the settings so everything is in dark mode. Only way to go in my opinion.
So is there any workaround if i am unable to access public link. I have tried ngrok and others but having issues. Help me out buddy, any quick solution
Heyy I was using it on my free colab notebook and my GPU hours got finished and could not use GUP anymore in this account. so can you tell me if it has any cooldown time for GPU so that i can wait or do i have to create new google acc to use GPU again.
It should. I know some people do have more than one account so they can stretch more time but I would be careful about too many since Google might catch on.
But is that because of the usage limits of free colab? I ask because I know recently there were some changes that I thought made it work better on Colab. I know when I first tried it many months ago I had issues but lately I have not.
I don't know if you've noticed, but videos of trailers in 1950 super panavision are currently going viral. The pictures are all taken in Midjourney. I tried to do something like that in Fooocus, but it doesn't work. I couldn't find a model in civit either. Do you have any ideas?th-cam.com/video/DbEgnYy_W7g/w-d-xo.htmlsi=Y6GOQ0VlGVN373Oi
5-27-2024: Gradio service having issues with shared links. View status of Gradio service here: status.gradio.app/ ---- If you run into issues with failed username and password when downloading LoRAs or checkpoints you may need to get and add an API to the URL. On Civitai website in the account settings add an API key and then you can add that at the end of the download link with ?token=YOUR_API_KEY - If you find the video useful and would like to tip you can buy me some electricity for all these image generations. It is greatly appreciated! ko-fi.com/kleebztech
what should he do no? how can we run withouth the link ?
I still trying to figure, from where I can get the link you have paste !wget.
Civitai website. The first part is where it will save it and the second part is the link to download from which I got from Civitai although you can do what I do now and upload my models to Google Drive and share a link which I use. @@livealive7207
bro how to import control net in fooocus plz help
@@Zendarkdagger you can't add more than what already comes with Fooocus. In the Input Image section you have options like pyracanny and CPDS for example.
Introduction and Overview - 00:00:00
Limitations of Google Colab - 00:00:26
Setting Up Fooocus on Google Colab - 00:01:04
Copying Fooocus to Google Drive - 00:01:33
Running Fooocus in Colab - 00:02:07
Customizing Fooocus in Colab - 00:02:35
Understanding the Colab Environment Setup - 00:03:15
Using Built-in Presets - 00:03:59
Downloading Specific Checkpoints - 00:05:36
Cloning Fooocus and Folder Structure - 00:08:07
Changing Default Preset Models - 00:10:02
Stopping the Colab Session - 00:12:15
Installing LoRAs - 00:12:48
Using LoRAs and Custom Models - 00:15:31
Editing Presets and Custom Settings - 00:16:02
Downloading and Running Custom Checkpoints - 00:17:48
Monitoring Resource Usage in Colab - 00:18:54
Managing Session Limits and Resources - 00:19:29
Downloading Generated Images - 00:20:01
Conclusion and Additional Resources - 00:22:13
you are the man. Many people complicate tutorials instead of explaining them right at the beginning, you already taught everything right away, thank you very much!
That explains why people only watch the first few minutes. 😂 Although I do give some useful tips in the rest of the video. I think....
@@KLEEBZTECH haha, yeah.
I ended up getting an error here because the T4 gpu is not active and in the end an NVIDIA website appears to download drivers.
I keep getting this same error, what do I do?
Was it working originally? It sounds like you may have hit the max for your usage for the day.
You can see if it will allow you to change your runtime type in the runtime menu. If you hit your max it should tell you that.
@@KLEEBZTECH Yes, I was functioning.
So does that mean it can go back to normal the next day? And yes, I used it several times because I hadn't gotten a good image.The strange thing is that there seems to be no way to turn on the GPU.
_I have a mid-ranged laptop that runs official fooocus very slow that's why I switched to collab. Thanks for this helpful tutorial._
you know how i know your video is the best? because you mention that you know that what youre doing may not be the best and if others know better then to mention. nobody knows "the best"of anything. collaboratively we all know things and the best is found when we all collaboratively share what the best is that we each have figured out. cant stand youtube videos that say "the best way to do xyz". nobody is all knowing. and someone willing to say that, is someone i want to learn from, becuz they are always willing to learn more. thank you :)
Thanks. That really made my day. I have been having difficulty getting videos out lately dues to some real world issues but also have felt I don't know enough about certain things to put the videos out. This really gave me a little boost.
i also really appreciate seeing you do the generations in real time showing mistakes as they happen. it helps bcuz i too will have mistakes and seeing how u go about solving them or rather troubleshooting them helps tremendously w real life applications. id love to see a less edited video of your workflow of "experiementing" to figure out how to accomplish some goal. for example, if u set out to create xyz doign xyz, then just show you experimenting until you reach that goal. or maybe you dont reach it but u learn stuff along the way. thatd be helpful i think, sort of teaching how to "experiment" w the settings, how u decide what setting to experiment w next.
You are Genius۔۔۔۔۔۔ You have made a very good tutorial by explaining the basic point. Thanks a lot....
You are welcome
@@KLEEBZTECH The basic point I learned from you, which I don't know, is that how can we change or edit the queries before run in google colab..... 👍🌹❤️
I use Colab with the $10 per month subscription, and it works excellently. If you don't have a graphics card in your PC, it's an excellent option.
Thanks for the info! Do you find that gives you the ability to create plenty of images?
@@KLEEBZTECH Yes, I do
Have you any limitations by usage? By time per day, per week, per month?
I use free Kaggle. I have 30 hours of usage per week.
Thank you so much for this tutorial. May both sides of your pillow be cold every night.
You’re welcome 😂
fooocus gives an error when I use another model,
but if I use the default model: "juggernautXL_v8Rundiffusion.safetensors"
then everything works without errors
I select another model
press "generate"
and the generation hangs at the moment "Loading models ..."
and then gives an error
how to fix this?
Thank you, I don't have a powerful PC so I always use Google Colab, god bless you good man
I am still learning to use on Colab so hopefully didn't miss anything when it comes to the basics. Since I have issues recording when upscaling I am planning on using when doing videos that use any upscaling.
Do you use the free only or do you use the paid? If paid I am curious if you feel it is worth it.
@@KLEEBZTECH Only the free version, I consider the paid version not worth it because I discovered other options available to run Stable Diffusion in the cloud, for example SageMaker Studio Lab and Kaggle. Both have similar limitations to Google Colab and I change from tool to tool every time it exceeds its use limit. I was focused on influencer modeling with A.I. for three months, the profits were minimal so now I will move on to Children's Videos with A.I., Stable Diffusion is a great tool
@@KLEEBZTECH Only the free version, I consider the paid version not worth it because I discovered other options available to run Stable Diffusion in the cloud, for example SageMaker Studio Lab and Kaggle. Both have similar limitations to Google Colab and I change from tool to tool every time it exceeds its use limit. I was focused on influencer modeling with A.I. for three months, the profits were minimal so now I will move on to Children's Videos with A.I., Stable Diffusion is a great tool
@@KLEEBZTECH Only the free version, I consider the paid version not worth it because I discovered other options available to run Stable Diffusion in the cloud, for example SageMaker and Kaggle. Both have similar limitations to Google Colab and change from tool to tool every time it exceeds its usage limit. I focused on influencer modeling with A.I. for three months the profits were minimal so now I will move to Videos with A.I. Stable Diffusion is a great tool
thanks for every singl video you have made
Wow. Didn't expect Fooocus can be used in Collab. Could you please tell me how do I know what console software can work with Collab? Something like rembg or like that. What should I consider before trying to run an app in Collab? Despite its resources limits, of course. Are all console software work in Collab that works fine under Linux, for example?
22:33 i guess reason to copy is in case the original poster were to delete the notebook, you have it.
Hi, why i cant use Image Prompt in Input Image?
I'm almost certain you are allowed to use this Google feature just because they are training some sort of a model on you..
😆Thankfully I limit the data Google has on me. All my private stuff is mostly on Proton at this point.
you deserve way more likes
its only showing local url not the public url there is no gradio link , what is the solution for this?
I am not aware of a solution. I've never run into that.
@@KLEEBZTECH do you have discord?
@@KLEEBZTECH Lol same here Been searching for an hour now
I think gradio was down.
That happens occasionally for me also. Gradio has issues at times. I switch to Ngrok when this happens.
Hi, I have been using fooocus colab for a month. But itseems like that they arent generating the public url now. Can u help me with it?
Actually just updated my pinned comment and my description to mention that Gradio which provides that link seems to be having issues today. Currently says working but if you look they have been having issues all day. Hopefully that is what the problem is. status.gradio.app/793595965
Mine is powerful, but when Temperature rises to 90° i will prefer not to run it locally
At 9:11 I run it and says: Username/Password Authentication Failed.
Check pinned comment.
Thanks so much for the video! I am getting invalid username/password come up when trying to run the juggernaut model with the download path, any idea why?
I have added more info to the pinned comment about that. Might need API key but also have heard mention that it may not work directly now. I found a better option personally. I upload my checkpoints to Google Drive and then share a link and enter that link for the download. It then does it instantly. Usually even a free Google account should have enough space in Google Drive for at least one checkpoint.
Error
/content/fooocus/models/checkpoints/Juggernaut_X_RunDiffusion_Hyper.safetensors: No such file or directory
Can't download files from civitai, even with added API token in the url as: &token=...
Has anyone solved it?
I did just test with mine and the one I normally download is still working. Have you tried more than one? I think it could also be model dependent and the settings the person who uploaded it decide on. Also double check for typos. for example this is a link to one I just tested: civitai.com/api/download/models/454069?token=$XXXXXXXXXXXXXXXXXXXXXXXXXXXXX
Have u tried to use huggingface? Can find most on there
What about refiners? I heard I can use SD 1.5 models as refiners but I can't get it to work. When I run the WebUI, it doesn't show in the list of refiners.
You would need to configure it to get a 1.5 model by downloading it to colab so you can choose it.
Everything runs, but when I'm in Foocus and want to switch to the checkpoint, in the Base Model bar, it doesn't show me the checkpoint I downloaded.
Somebody help pls.
Very helpful 🙏 any idea on how to add Lycoris?
This might help. I have not tried using myself but appears the same as with LoRAs so the same method should work. github.com/lllyasviel/Fooocus/issues/297
hey, have you got it to work?
public url issue now resolved, its working
3:23 so why bother to copy to drive, that i gotta see. i have never done copy to drive. so if there is any benefit, will see what you show me.
Unless you just run the default each time. If you want to change anything it makes it so you don't have to make those changes each time. Yes the environment will be destroyed each time but anything custom like the commands to download the checkpoints or LoRAs will stay. So when I open mine and want to use the anime preset I don't need to change that line since I have the different options already set up.
@kleebztechai could you tell us where you copied the code for the LoRa and checkpoints you only mentioned to type in !wget
Same as with the checkpoints. If you look you will see that it is saving it in the lora folder. But you may need to include an API key. Check out the pinned comment there should be a link to more on that. Apparently some you need an API key for. I had only run into one before and didn't realize it because the rest all worked for me.
You can also download it and save it in Google Drive and have it download from there and directly.
i don't why but it doesn't work please could you help
Is it possible to use google colabs and therefore fooocus on a phone?
Yes. Fooocus has a decent mobile interface.
@@KLEEBZTECH thanks
how to enable refiner in colab??
You would need to have more than one model in Colab and then choose in the model tab at the top but on the free version of Colab you will possibly run out of VRAM.
still work ?
Спасибо! LoRA можно загрузить на гугл диск, и уже оттуда загружать - получается намного быстрее
my downloaded checkpoint not showing...Any idea why?
have you hit refresh all models?
@@KLEEBZTECH Yeah brother, i have did it, loras are working fine but checkpoint i which is downloaded in the folder is not showing in the generator
Found the solution! thanks mate, i didnt save in any file type thats the problem
can you run the flux model on this? I get an error on Foocus colab saying *"Username/Password Authentication Failed."*
That's when i try and run the Flux model from civtai. But no problems with the model you used from the video
Fooocus in Colab will not work with Flux.
@@KLEEBZTECH thank you for this reply. I figured, but just didn't know for sure
@KLEEBZTECH I'm trying to use a checkpoint in fooocus Google colab but civit ai recommends a VAE for the checkpoint so I tried using the API key method I've been using to import checkpoints and loras from civit ai to colab but it's still requires authentication even thought the API key is provided. I downloaded it and placed it in Google drive and tried to fetch it from there and it still request authentication. I need the VAE to get descent photos so Please what do I do?
Have you tried uploading it to the correct folder in Colab manually? Even LoRAs I have manually uploaded to the folder. Checkpoints don't seem to work because of size.
@@KLEEBZTECH I tried uploading it manually and after selecting it in the VAE section to use it?. It says the VAE file is corrupted. I don't know if it's because I manually uploaded it or something is wrong
I would possible go onto the Github discussion section for Fooocus and see if anyone might be able to help. I am not sure and I have not run into that.
@@KLEEBZTECH alright man
how do you set fooocus on dark mode?
"--theme dark" flag can be added to launch command or you can do what I prefer which is to change my browser to dark mode in the settings so everything is in dark mode. Only way to go in my opinion.
Does it work if you have 15 GB free in drive?
It does for me since I do not use it on my primary account and only use with my one without any extra space besides the free.
Not getting public link how to resolve it after runn8ng foocuss please help
It is an issue with public Gradio Share API that has been a problem the last few days. Usually try again a little later and might work.
plsss tutorial on how to use fooocus api???
I do not expect to do one. Sorry. I have no experience with the api for Fooocus.
@@KLEEBZTECH yeah from what ive seen its very poorly documented and hard to use
Not sure if you checked on Github already but might find some info there. github.com/lllyasviel/Fooocus/discussions/3361
@@KLEEBZTECH mmm might check it, thx
is lightning Ai better than google collab?
Have not used it so could not say.
So is there any workaround if i am unable to access public link. I have tried ngrok and others but having issues. Help me out buddy, any quick solution
I wish I had a solution. Last I had seen it sometimes is working but they still seem to be having issues.
Heyy I was using it on my free colab notebook and my GPU hours got finished and could not use GUP anymore in this account. so can you tell me if it has any cooldown time for GPU so that i can wait or do i have to create new google acc to use GPU again.
Should be good tomorrow again. Think it is about 24 hours.
@@KLEEBZTECH okay cool. I'll confirm if it would get fixed tomm. Or else have to create a new acc...
It should. I know some people do have more than one account so they can stretch more time but I would be careful about too many since Google might catch on.
@@KLEEBZTECH Got another 1 hour 30 mins free gpu for that acc. Cool down was about less than 2 days for me.
But it can't handle heavy tasks, like multiple image prompts and remixing...
I've used it for three image prompts without issue
@@KLEEBZTECH but after sometime you can't ...
But is that because of the usage limits of free colab? I ask because I know recently there were some changes that I thought made it work better on Colab. I know when I first tried it many months ago I had issues but lately I have not.
when i run the checkpoint it says Username/Password Authentication Failed. please help???
@@winter6278 check the pinned comment. You may need to set up and use an api key for certain checkpoints.
it is working thank you bro😘
@@winter6278 you are very welcome. 👍
how to fix public issue?
Right now sometimes it works sometimes it doesn't. No known fix.
I don't know if you've noticed, but videos of trailers in 1950 super panavision are currently going viral. The pictures are all taken in Midjourney. I tried to do something like that in Fooocus, but it doesn't work. I couldn't find a model in civit either. Do you have any ideas?th-cam.com/video/DbEgnYy_W7g/w-d-xo.htmlsi=Y6GOQ0VlGVN373Oi
i guess this code is not malicious
It is not.