100% WORKED!!! Step-by-Step Guide: Install ComfyUi, Controlnet & Models | Beginner to Expert

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
second I'll never forget today uh and for the third time I'm going to install the comfy UI and control net but this time 100 I am sure that what version of The Comfy UI is working okay so for the last time again I am telling you that today yes I am sure that this version of the conf UI is working okay and I'm not sure about the outer version also this new version is not working well but about this version 0 4 underline zero six underline 2023 about this version I am sure that it is working you can download that from the confi UI GitHub page uh it's here go to the installing configure and click on release page and after that go to the assets area and download this version don't waste your time on other versions okay this is the file you should extract the version once that you the starting extraction and or finish the extraction you can go to the comfy UI after that go to the custom nodes here you can write CMD and after that the command prompt will be up here exactly in this address now I want to install the control net and confi UI and also this is the GitHub page of the control net scroll down at the end of the page you can see that how you can install the control net there is a comment git clone this repo select this URL copy the URL and paste the URL here okay it will going to download the necessary files inside the folder that we selected here and still the extracting is in process okay wait a minute to finishing the extraction okay so text section extraction is finished I'm going to open new folder inside the confi this is the folder that is extracted go to the patent embedded and here we have the control net that is downloaded in our system I'm going to drag the install.py onto the python.x this page is opened for me and it is going to download and install all necessary quotes and other things for the control Nets this process is very important and you have pay attention to the errors or other things that you can see in this page okay here you can see that the we have arrived to installing collected package everything okay successfully uninstall the old version successfully installed now it's going to download the models this is very important and I checked this installation with the new version of the configure it was not work but with this version it is working very very easy okay it's finished the installation of the control net is finished this time I am sure that the installation is working so now before running the confi UI because it is not the latest version I prefer go to the update folder and run the update underline config.bat it will try to update the config UI and also let me update the python and configure and python dependency also it is going to take about one or two minutes depending to your computer hardware and also your internet connection successfully installed accelerate Transformer Dash 4.30.2 okay uh before running the confi UI also I prefer to install outer extension that is very very necessary the extension name is from UI manager and also the installation process is very similar to control net I'm going to do the custom node folder click and write the CMT okay the installation is ready I have installed the confi UI manager and you can see the manager button here and if you click on that configure manager menu will be open also if you open go to the go to add nodes you can see the pre-processors so you can be sure that control net and also the configure manager both of them are installed on your machine okay so what are our also what should we do let me close the config UI and open the folder um here if we go to the models folder we can see some folder checkpoint clip clip Vision VA evae approx and some other names okay let me explain that to you I think in the checkpoint we have to copy and place the stable diffusion checkpoint models and you can download them from cv.ai from this side you can download lots of checkpoints okay each checkpoint is something like a database and it can provide some different style of the art I am suggesting to use the realistic vision and also rev animated these two checkpoints are very useful and are working very very good rev animated is some cartoonish style but it is very powerful to make um 3D models with some good consistency and I prefer to use that to use the Rev animated as the first processing image and the realistic vision for my second processing image because the realistic Vision can provide realistic rendering of course you can download them and after the downloading was finished you should place them here in the checkpoint previously I did that in another folder and just now I have to replace them oh I think I have deleted them so again I have to download the models or let me do a search over my computer no no still I have to okay realistic vision and rev animated and also there is a folder with the name of the vae also the VA is very important and for example if you go to the realistic Vision page after that you can see that the owner of the realistic region such as said to use this kind of vae and each engine is working with different vae to get some better results and let me explain what is the VA VA some encryption algorithms it will convert the image to the code and also it can convert the code to the image and each vae can provide some different style of the image for example we have some vae that can create an image with some glue angular style and and or this vae can provide image with some high contrast and some realistic appearance so it's also it is very important to download the vae and also place the necessary vae in the folder right now only I am using the orange mix and the vaef ftmse 84000. because this one is suggested to be the realistic vision this is the file and you can download that and in the Rev animated you can see that they have suggested us to use the orange mix or this phone and or this one okay now let me go to the control net folder okay this is the important file important part of the video and you should pay attention to that change once I have another VA previous before that let me copy that okay control net also the control net needs some database okay and manually we have to download and place them in default in the folder uh if you go to the control net page for the confi UI you can see that here we have some different processors for example there is a control net with the name of Kenny Edge and it's a preprocessor and this is the pro and this is the processor and the processor for example the name of the processor is the Kanye or T2i adapter can over here we have mlsd that can be used for straight lines and for example scribble and many many different um control net models each control net model can do some specific work for us and let me show you here this is the official page of the control net and here you can see a very very good documentation about the control net and their ability for example MLS th Canyon H okay for example um with the candy Edge the Kenny Edge is using these type of lines and after that it can convert these lines to an image and also this is a this is some kind of canning Edge processors M Dash LSD lines is for processing the straight lines and it can convert the straight line drawing to photo for example you can see here we have the window and in the image you can see the window position and also it is very good for the building for most of the modern architecture architectural form you can see the mlsd processor within in your controller HED is a working with some smooth lines you can see here and also you can see from this sample a scribble can work very very good with your hand drawing for example sometimes you have drawn a light a line with many errors okay the best processor for your hand throwing is scribbled and uh it's amazing just I can see it's amazing and it's very very powerful for example it generates these image from this funny driving [Music] and it's really working very very good scribble uh we have the human pose okay it can work from these lines and it is going to convert the position of the arms head neck body foods and other things and each line has a different color okay and each color is defining the position and orientation of that part of the body so with open pose yes with human pose or open pose you can transfer opposing a style for example here there is an input image in computer and after that the computer analyze it that's okay this is the head position this is the rotation of the arm of the chest another thing and when it is uh going to process and creating an image it is using this information so with human pose model we can process the position and posing of the body uh there is one other powerful model with the name of segmentic semantic segmentation also it is very very useful and powerful and it will try to convert input image to some color code and after that each color code will be Translate to some object for example this color code will be trans translate as a sky this color code will translate as grass this color code will transfer as well this color code will trans translate as building and uh it is very very useful for example here from this input image you can see that the system is analyzed at the image and convert the image to this segmentation with these colors and after that from reprocessing this segmentation it got these amazing results then another processor is dips the depth processor is working based on the depth information for example if you have a camera in your scene the camera has a specific distance with your subject and also with your background and it is trying to estimate the third dimension from extracting the position and the distance of the camera with your object with your subject and with your background also it is very useful and if you are working with 3D modeling software most of them can export zdeps information during the rendering also this model is very useful here we have another model with the name of normal map the normal map can process the image Dimension actually and it can it can understand that which polygon is orientated in which direction okay let me explain that to you uh for example let me see my video oh okay this is a Surface okay and we have three colors red green and blue okay red green blue is equal with x y z okay and for example if the rotation of this surface is an X Aziz after that the red color will be go up if it is on egreg for example it will be different color it will be green and for example if it is on Z position it will be blue okay so and also if this is the and also if the rotation is some average something like that it will be a color between this red green and blue um it's uh it's very very simple just imagine that if you want to send the information of orientation of each surface you have a map here that can export the position orientation of the of of each phase based on red green and red green and blue colors uh it's very very useful and again in your 3D modeling application easily you can export normal Maps also it can reprocess and extract the normal map from an image it's very useful uh also we have some new models and processors with the name of animal line drawing and if you pay attention to the line drawing you can see that and the lines are very clean and most of them are very sharp okay we we don't have a lots of running traffic okay uh so we have some different models for different processes and it's the main thing that you should know okay here I have a list of the control net processors and you can download them from here let me share if you go to the control net I will add the URL in the YouTube [Music] they are available in again face okay okay you can download the latest version of the control net version 1. 1 or 1.5 1.1 from this repo each model has a file near that with the extension of yum Health it's very important to download both of them and place at them in the folder already I have them and I'm going to place all the files in the folder in which folder coming for UI model control it okay and here it is also I have another uh control net models with the name of T2i let me show you also I'm going to share this control net with you and also these are very very similar with the control net models but in some case they can work better for example we have an Style we have a model here with the name of a style and it is very very fun for example here you can see that they use this sketch as input and this salad photo as a style and after that they transfer this motorcycle to a salad and it's very very fun and it's very useful and I'll also I have to place the control net T2i adapters in in my folder and you can download them from again face again here yes fine and version you should go to the models and here you can see all of the processors but if you pay attention you cannot see any dot uml file okay and we have to create them manually and this was the way that I did that because for example here I could make a copy from this file and please set them here and rename that for example it was in another name tcml and for example now the dips I need yum L file for the tips I am going to copy this name and place on the yaml file it's on just you have to make sure that all of your processors and all of your models with the extension of pth also have yaml file and we have to copy and paste the models here uh let me do that I'm going to select all of them and move them here okay uh okay but there is there are two files that should not place that shouldn't be placed in control net folder it's very very important then the style core adapter estai okay and T2i Style these two filed these two models and with vitarium L file totally these four files should not be placed in the control net they should be placed in a style model it's very very important it's very important okay now I have placed all the necessary files and again let me run the confi okay now let me add a control net processor load controllers model and now if you click here you can see Elites list of the control net models okay uh and now let me make a sample with control net inside config UI okay really I like to start with this solid sample let me go to the asset example [Music] okay let me start with this two photo and let's try to make this test okay I'm going to save this image and also save this image and let's start and see how we can make something like this in configure with using of control okay here at first I want to use the realistic Vision model and with realistic Vision I have to load another vae and put I am double click on this on the screen and after that writing v a e and using the vae loader and from here I'm using the vae this this vae for realistic vision and going to connect the vae to the vaeed code cycle repeat okay this is my prompt and also let me adjust the resolution here and I prefer to start [Music] processing with these two sampler now I have to load the control net okay here I have two image let me open them one of them is sketch and one outer is the salad okay what can I do I need to processor one one of them should process the sketch drawing and one out there should start processing the style two loaded model for a start a sketch I'm using Co adapter A Sketch okay and I am going to export the conditioning ah not here let me also load another control net model net apply the conditioning go to the conditioning of the control of the app layer control net model and after that the conditioning goes back to the positive the apply control net also need a control net model and also it need image if you click on image and drag that you can load image and you can drag your image into the area or you can browse that so now we have the motorcycle with this shape let's skew the prompt I want to start processing okay I prefer to place the save image here let's increase the strange because I believe that if you increase the syringe after that the output will be more similar to your sketch let me open that just a little more um okay with amount of with this amount of 1.6 I think with S range it can analyze this sketch and convert this sketch to the motorcycle photo okay so now how we can load the solid photo as this type again I am double flick and type style here we have two options a style model apply and a style model loader okay this style model will be connected to the style model the conditioning that go out from the control net will should go to the conditioning to the style and after that conditioning from this type we'll go to the position and also here we need clip Vision output I will explain that but before that here we have two different style model I'm using Co adapter and for the clip Vision output and tracking and after that it will suggest to add clip Vision in quote and when I am adding the clip Vision encode it needs to input one of them is clip Vision that I have to load the clip vision and one other is image that I should load an image for that okay so uh for the image let me open the solid image but what should I do for the clip vision okay aha let me go to the comfy UI page and search for clip unclipped model let's be searching on Google [Music] I will place the link here and they told us that we have to download this file by touch model.bin from this URL okay and click on that and we have to download this file the downloading is sorted and we have to place the file inside the models inside clickfish still this is in downloading until one minute later the downloading will be ready and I can place that once that I place the clip Vision model here then we can use that inside the conf UI and it's very very important to do all this process hmm okay the download is ready okay so now we can go back to the config UI and just I I have to refresh refresh the page okay now inside the clip Vision I can see the name of the pytorch model dot pin it's very important it's very very important now I'm going to do the startup processing Q prompt and you can see it is going to start analyzing and processing okay we have the motorcycle but without any salad so what should we do this time I am going to decrease this range okay and one more time doing the Q Prime again decrease the strange okay now we can see the salad but there is not any motorcycle I'm going to increase the syringe and this time we can see both of them salad and also the motorcycle uh I prefer to use the batch size of eight and Q prompt so with this option eight different image or analyzing based on different seed with this strange and after that we can take the better decision that we have to change the syringe number or we have to change some other thing okay it is analyzed and they are all processed okay let me check my comment my prompt a motorcycle built with builds motorcycle solid very very simple not any other additional spinning and this time it's in vaet code processing okay we're changing the prompt and with this storage we cannot see any motorcycle in these images no problem I'm going to increase the strange because this is the power of the sketch and I think we have to find the right number again increase this range and start processing we cannot see the motorcycle again increase hmm I cannot see the motor and I'm going to increase the estrange The Prompt syringe and also the style model all of them are important ah okay here it is we have that it's working so with this prompt and with this strange to create this image [Music] and also this is the mod this is these are the wheels and these are the piece of salads that are trying to make the motor and if I again if I increase the strange it's a processing um okay great it's working yes and with number of 1.3 hmm okay okay it's working and also in this video you learned how to install the comfy UI from zero install the control net applied a style model installed and download the clip vision and all other things and now also you can run and make something like this on your computer I have lots of thing in my mind and I want to create and record many many different videos for you uh I hope you will like this video and
Info
Channel: Arch_AI_3D
Views: 247
Rating: undefined out of 5
Keywords:
Id: FJwcBTL_voc
Channel Id: undefined
Length: 51min 5sec (3065 seconds)
Published: Sun Jun 18 2023
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.