This Free AI Image Generator Beats the Paid Ones
HTML-код
- Опубликовано: 4 июн 2024
- My favorite AI image generator is totally free. Personally, I think it's better than Midjourney, DALL·E 3, Canva AI, Adobe Firefly, and even Leonardo AI. I mainly just use the text to image generator, but there are so many more features than that.
It's name is Fooocus - it's based on Stable Diffusion, but it has taken all the difficult parts, and made them so easy. With no tweaking at all, it produces amazing AI generated art and images.
I hope you learn something from this video. Comment with any questions, and I'll make sure to respond!
***
Links mentioned in the video:
- Google Colab file: colab.research.google.com/git...
- Community style examples spreadsheet: docs.google.com/spreadsheets/...
- Fooocus github page: github.com/lllyasviel/Fooocus
***
Attributions:
Morning Routine by Ghostrifter Official | / ghostrifter-official
When I Was A Boy by Tokyo Music Walker | / user-356546060
Music promoted by www.chosic.com/free-music/all/
Creative Commons CC BY-SA 3.0
creativecommons.org/licenses/...
Icons designed by Freepik, lutfix, Dewi Sari - Flaticon www.flaticon.com
***
0:00 - Intro
0:20 - What is it? Why is it free?
0:53 - Getting started
1:49 - Basic usage
2:31 - Advanced features
4:31 - Style examples 
Наука
Wow. This is awesome. You just earned a new subscriber
Dude this is awesome.
Thank you SO much! This will work perfectly for me.
Awesome! Glad you got something out of this video!
I was wondering what does the google collab file do, does it run the code on your own pc or on the cloud, and if it runs on the cloud who pays for the could.
It’s run on the cloud, on Google’s servers. Fooocus is just light enough to run on their free tier, so you can use it indefinitely for free! Apparently some features that I didn’t demo can use too many resources that Google will make you restart, but I’ve used them and never actually had it happen 🤷♂️
@@PatrickStorm_ I used it a few times earlier, but now I get "You cannot currently connect to a GPU due to usage limits in Colab." Are the limits on Colab associated w/ my account only, or does this mean there's too much usage of Colab overall. If I was to buy credits, would I still have this problem basically? Or not?
@@alexanderkuo1578 Hmm, I only hit that message when I was using colab for more than 10 hours a day. At that point, I upgraded to the $10/month subscription which allows you 24hr per day of the GPU instances. You might want to check if you have multiple colab files open, that can trigger that message too. Otherwise, I’m really not sure, maybe take a break for a day or two and see if the free usage limits reset.
@@PatrickStorm_ Now the whole story makes sense. Would it be better to just download the whole thing and run on our own machine. It seems like it can run with a regular graphics card machine of sorts - equivalent to running a valorant type game on steam. Is there something important I am missing there. complexity of building or additional steps beyond the apparent there.
Yeah, if you have a decent GPU, you will be able to run it on your machine. I have never done it, but that’s how most people use it, so the instructions on the GitHub page will have all the steps. github.com/lllyasviel/Fooocus
Is it possible to upload a subject, and then give a prompt so that it generates a photo with the subject in it?
Yes it is! That is a whole section I forgot to include in the video actually. At 2:34 you can see I click “advanced”, next to that is “input image”. Click that and you have a new set of options like in painting, out painting, removal of objects, etc. I haven’t tried to do exactly what you are talking about but I know people use fooocus for it.
@@PatrickStorm_ There are a bunch of capabilities with input image as well. But the user interface does not have enough directions or documentation to make it user friendly. Even regular things which are listed there, are impossible to do, take for instance background removal. One is supposed to paint the whole background to remove it. and then it takes everything down the pixel level granularity instead of a guidance, leaving ugly artefacts. At the end working with an image is just not intuitive or high quality result oriented.
Issue of 2 prompts and multiple input images is also very confusing. Product Packaging wise it seems more like a college project, than a usable high quality work, while it may have superior or state of the art features. Sad but true.
I might be missing the whole truth here, if I did. It would be a wonderful service to the project if you make a comprehensive tutorial for using input images in Foocus.
To me it seems that the features you use in your presentation are NOT in the free version of this tool!
There is only a free version of fooocus. To see all the extras I showed, you have to click the “advanced” button at the bottom. I have an arrow pointing to it around 2:34! Let me know if that works for you!
oh great, not looking forward to a tsunami of garbage AI videos and imagery from all these free/pay tools
You do know that it's YOU who decides what garbage you watch, right
Both of you make good points really. I think there will be a flood of low quality media. And hopefully the quality stuff will rise to the top.