Mac users. Stable diffusion - Fooocus on MacBook step by step tutorial and performance

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
[Music] hello everyone welcome to my channel in this video I want to show you how to install focus on MacBook it's applies to MacBook with AMD GPU or the latest M1 M2 or M3 so let's uh get it started one thing you want to need to install to make or make sure you have it is the mini cond which is for python virtual environment if you are not sure how to install it please uh refer to my previous video I will have the link in description once you have the mini cond installed you will uh open the terminal so you will go to terminal yeah your terminal will have the base this indicates the cond environment so to make sure that you can type cond press enter it will show you the Avail options this means that you have the cond created all right so I will show you step by step how to install Focus so from here you want need to create a virtual environment cond create dasm this is the name of the virtual environment you can call it whatever you want so for me I will call it Focus EMV um yeah I was just just call it focus and uh then you want to specify the python version I what do 3.1.2 and you can press dasy yeah from here we can see that it install all the required packages and also shows you the instruction to activate you would need to do the counter activate focus and press enter okay so some pay attention that we see that here we change the from base to the focus looks good all right so next step we want to do is to uh install the py torch for Macbook so what you want need to do is to uh go to the developer app pro.com and then you copy this command and go back to your terminal paste that so this will install the py torch when complete we want to verify it we go back to the developer page we copy the script here switch to the terminal python press enter and paste the script we see that uh it indeed show us the device is MPS well looks good let's exit all right so let's uh move on to next step so next step we would need to follow the full curse official installation guide for Macbook so we would need to get a colum the Repository and uh press enter so because I already had it so I will skip it after that you would need to go change into that directory all right yeah and you can open that so from here you can see the focus scripts all the files here so next step what we want to do is to for here this include all the dependencies which we want to install switch to your terminal you want to do pip install dasr requirements on the score versions. txt and this way will install all the dependency press enter it will download some packages and install all of them okay we see that successfully installed all this packages all right we are ready to run our full Cur UI let's go to the instruction here we let's move on to the step seven so this one is all you need to start it also make sure that you also if you have a M2 MacBook you want to use this line here so for me because I don't have M2 I will just show you this one all right so let's uh Ty in that into the terminal and press enter the first first time you run that it will download some stable diffusion model files into the models directory so expect it to be long time if you don't have a fast internet connection so um because I already downloaded those so for my demo it will be quicker let's press enter to start it so this will be the user interface after the starting so you can see here there's a local URL by default it will open your favorite uh default browser but if not you can copy this and paste into your browser you can see the user interface all right so what I would like to do next is to do a quick demo R of it so for this meab book it has um AMD discrete GPU which has about 8 GB of vram and also 32 GB of RAM so I think it should be powerful enough to run it so let's see how it performs all right so for demo purose I would like to click the advanced and uh we can use image number go to one and uh use the speed and for the prompt let's do like a football all right so next I will also show you my activity monitor I will open the GPU history here and also my CPU usage [Music] here and also the memory here yeah so the memory may not be very accurate because I have like other programs opened so let's uh just to like show it here all right so back to our interface let's press generate yeah and let's see there's some information printed out on the terminal and we also can see our um dashboard graph for GPU for CPU and for memory all right let's uh just wait it yeah with see it needs about uh 5 minutes this this one is because I'm like doing some of the recording so it affects the performance but let's look at the GPU graph here we see that for some reason it's not reaching like 90 or 100% utilization so this is definitely some like a thing I think a future Improvement is needed with see it's uh General generated the image in about 334 seconds so this is the image I think uh quality wise looks really amazing okay so next what I want to demo is to the use the extreme speed so I think this one to use the LCM method talk about uh usually takes eight steps instead of 30 steps so it will be a really speed up so let's take a look golf all right let's generate also if you haven't used it previously you will need to download the LCM model okay we see that indeed this time to only talk eight steps about uh 45 seconds total time including the moving model part total time is 8576 seconds and uh indeed the golf ball is here looks also really nice yeah okay so for time purpose I will not demo the quality mod it will take much longer than the speed so feel free to experiment with that if you like the video please subscribe to my channel thank you for your support goodbye
Info
Channel: Tech-Practice
Views: 16,880
Rating: undefined out of 5
Keywords: ai, gpu, amd, hardwares, stable diffusion, fooocus, macbook, metal
Id: IebiL16lFyo
Channel Id: undefined
Length: 12min 46sec (766 seconds)
Published: Sat Jan 27 2024
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.