Fooocus is hands down the easiest way to get started with Stable diffusion and SDXL. The sleek interface and Style selectors make it easy for new users to make beautiful art without dealing with all the complicated Stable Diffusion requirements that were introduced with SDXL. Try it out today, It's a 1.8GB Download, then unzip some documents and click on run file. No python, GIT, or other prereqs necessary. Here is the Fooocus link to get your started - github.com/lllyasviel/Fooocus
Thanks for the detailed instructions! I'd love to try out Fooocus but minimum requirements show 16GB VRAM for AMD systems and I have an RX 7700 XT GPU (12GB). Does that mean it would not even run/generate images on my system at all, or just very slowly?
Invoke AI is the solution for people who wants functionality without sacrificing any new features combine with easy interface. Much more advanced that Fooocus yet easy to learn. In addition, as you progress, you can use more and more advanced functions such as nodes, just like in Comfy UI, all in one common environment.
this program is amazing. when i first loaded it i thought it was just overly simplified. but did not realize it has a LOT of nice uniqjue features built in as well as one of the BEST face swappers ive used def worth a try DONT be turned off by its simple looking interface at first!
I never bothered to try Fooocus cos I thought it wouldnt be as good as A1111 but your video convinced me to give it a try simply cos of how convenient it looks. Thanks for the video.
Ya it's much more convenient and faster. I also just link to my existing A1111 files so I can jump back into A1111 if i need the full control net suite. I'm not sure if all control net models are back up and working with SDXL though. Funny thing is Illyasville (the guy that made fooocus) is the control net guy so it's weird his tool doesn't have all of the control net pre processors.
Joined and like put. Could someone tell me why every time I launch the run batch file it downloads 6.62 Gb (of what?) instead of opening the program for me? Any help would be greatly appreciated. I've a Laptop with a Nvidia RTX 2060 installed on it. Thanks!
the face swapper in foocus is amazing way better then other options out there. but it does NOT JUST seem to copy the face but hair, skin and apparently maybe even body type. awesome!
Nah the face swap is decent for sure though reactor or roop are straight up better, if Fooocus get's dreambooth and reactor I'd purely use it and holy crap does it make things easier plus the results are waaaaaaaaaaaaaaay more consistent in my time using it while also looking a damn sight better
Do you know how to enable low VRAM mode? the Command prompt tells me i may need to enable it - hence why it doesn't even start to generate images, I have 4gb vram but it should work, at least slowly right?
The Config file didn't come with my download for the vanilla version. No mention of anything changing on the github page. EDIT: Didn't realize it gets installed when first launching it.
am using amd radeon graphics, which doesnt seem compatible with foocus installation, any advice?,,, also tried using it on google colab but no luck either, any advice? nb my pc cant support NVIDIA graphic drivers
so. i tried to make "someone" with robotic body. fooocus can't do it. whatever prompts you try - fcs doesn't understand what "cybernetic, robotic, cyberpunk, metall" means
i run a1111 on a gtx1070 SDXL1 is up to about 2min to render, newer sdxl1.5 up is faster, its a old 8gig card not issues at 1024 render size, i then do the upscale with gigapixle as its better than SD upscale where the vram can be a issue in that aspect. Easy diffusion is one click. I'll test Fooocus though. Easy Diffusion is definitely more user friendly than fooocus for people that are new..
Yup 8Gig is the key for A1111 and the min req. but if you start playing with the resolution sizes you might get an out of vram error. My 1660TI with 6GB of Vram can't even do the 512 render that gets upscaled to 1024. I could do it with Fooocus though. I'm starting to prefer Fooocus > Comfy UI then > A1111. Still love A1111 but it can use some optimization and simplification.
@@ArtificallyIntelligentAi made a video to help the issue you are having in a1111 will fix your memory errors, plus figure after you do it you can pass the information to others also. this works for 2gig and 4gig NVIDIA laptops too I have read. i deep dived forums to learn about, lot of gtx960 users told me about it. ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-TbFRXDdWssI.htmlsi=wYxXYcth-RBkthhn
i've been playing around with fOoOcus for a few month now, and i still have an issue i can't fix. i have 100s of food photos from back when i was a chef and want to use them in combination with my new table xD it sounds odd but basically, i want to use fOoOcus to add those foods to my table, as if i snap a photo of them ON MY TABLE. any idea on how i can make that haappen?
Thanks for your video, really useful! I have a question: is there a way to make an AI generate subject holding another object taken from a picture? For example, I've generated a character, now I want him to hold a particular videogame controller. Is there a way to make this happen using a controller photo as input? Thanks
Its pretty bad that, when you create a low quality image to see if the prompt works, but you like the image that much and click the "quality" check mark, that it just creates a whole new image instead of the image you first saw and liked in better quality... am i missing osmething?
Hey man, thank you so much for this!!!! Appreciate the help. Let me ask you, I downloaded the MRE version, and it's not appearing those Styles for me, do I have to download them from somewhere?? I'm still new to this, getting to read the Wiki now. I appreciate any help you can give me!
Hi, have you tried changing the source image ? I had something similar happen where the face of the output what just white and blurred. Also what the console say? If it says something about a failed wheel for insight face you the newest windows visual c++ . In the official install guide if you click on windows there is a Microsoft visual studios tools python command. You can just copy and paste that into python and press enter and choose desktop dev pkg c++ when prompted or asked
Hey could I get some advice from the OP or anyone who might know? I downloaded and unzipped everything, ran the run.bat and realistic.bat, but now how do I launch it?? In the video @ 7:35 it shows him clicking on the run folder next to the fooocus and python folders but mine is still the run.bat and just tries to install the batch file again.
@@v2iketiit165 you have to right click open the run.bat file and add this text to the path in the text editor: --theme dark so it will be like this: .\python_embeded\python.exe -s Fooocus\entry_with_update.py --theme dark
I've seen a couple in the past that were discontinued or not updated. Which one click installer were you thinking of and is it better than this one? I'll check it out.
I am very new to the AI image generation journey so it might sound silly. Can you let me know how can I use the image prompt to generate consistent characters or environment within fooocus. Can someone guide me to a step by step guide to get this, it will be a great help.
Im sure its great, ive seen it do amazing stuff. Unfortunately is just doesnt work on my RX 6700XT and i cant for the life of me get it to work. This is indeed a cry for help, anyone got any tips lol?
You can switch to the v1.5 model that uses a lot less vram if your not trying to use sdxl specifically. I think it downloads it automatically but let me know if you need help trying that I can walk u through it. The base model isn’t great for 1.5 but on the site civitai they have really nice custom models comparable to sdxl
Great video but I want to know how to use the same person (the same face) with different poses? I tried setting the seed but it doesn't work for some reason :(
I can never really tell on my end. I guess it's something I need to work on. Thanks for the feedback! Definitely going to keep it in mind for my next video.
its a 4070 some of it was cut but with very fast it was a 5-10 second render and an average of 10-30 for normal quality images. Depends on the type of image.
@@ArtificallyIntelligentAi Not gonna hold you to it, but do you think that a 3060 with 12gb would do pretty well? I'm not too worried about the time it takes, moreso running out of ram
@@mastabas yes, better than my more expensive 3070 which only has 8gb of vram and the 3060 should still be pretty quick. The biggest concern is not having enough. 12 is pretty good and it gets super expensive for the 24gb models 3090 and 4090
Anyone know how to uninstall this? It's... Alright. I mean I get why people would use it, but it's not for me. I vastly prefer MJ. However, I can't figure out how the hell to uninstall fooocus. Deleting the folder isn't enough, there's still 8gb+ of shit SOMEWHERE that I can't find.
I just left the link because the video was already too long and MRE deserves its own video with all the settings it has. TBH I need more time with MRE b4 I make a video on it.
How come people say this is good. Every image generation costs around at least 5 to 10 minutes and puts my GPU to 100% like a fighter jet. I got way enough vram and a decent AMD GPU with correct settings. Quickly dropped it, waste of time.
does this allow nsfw prompts? I create fantasy / dragon pieces and sometimes there's nudity. I don't want to waste my time downloading something that's going to have a filter. Thank you.
If you're going to provide an instructional video, you should be prepared to answer questions as well. What kind of teacher teaches, but doesn't answer questions?
