How to run Wan 2.1 Model - Text to Video/ Image to Video VLM using Swarmui

Summary

In a comprehensive live session, the creator from AI ML Talks demonstrates the use of the Wan 2.1 Model in Swarmui to generate videos from both text and images. The tutorial goes through the installation process for Swarmui on different operating systems, detailing setup requirements like memory capacities and software dependencies. Throughout the session, the host tests various prompts to showcase the model's capabilities, emphasizing the importance of selecting suitable model sizes based on your hardware. Insights on utilizing both consumer-grade and high-performance computing resources are discussed, providing flexibility for different user needs.

Highlights

• Swarmui helps convert text and images to video using advanced VLM models.🖼️
• Mac users can install necessary tools using Brew for easy setup.🍏
• Different GPU models from WAN 2.1 cater to various computing capabilities.🚀
• The live demo showcases real-time video generation, including dogs walking on grass.🐕
• Testing videos on different hardware configurations provides practical insights.💡

Key Takeaways

• Wan 2.1 Model allows for both text-to-video and image-to-video conversions using Swarmui.🎥
• Installation for Swarmui varies across operating systems using platforms like Brew for Mac.💻
• Selecting the right model size is crucial depending on your available hardware.📊
• Swarmui supports open-source usage, making it accessible for creative projects.🌐
• The session highlights testing and troubleshooting for various system setups.🔍

Overview

The video explores how the Wan 2.1 Model, integrated into the Swarmui interface, can be leveraged for creative video generation tasks. The model supports generating videos from textual and image prompts, adapting well to different computing setups. Using Swarmui makes running these complex models user-friendly and accessible, even for those without deep technical expertise.

The presenter guides through the installation steps tailored for Mac, Windows, and Linux users to ensure widespread usability. Emphasizing software prerequisites like Homebrew, the host also navigates model selection based on system memory capabilities, notably highlighting the differences between handling 1.3 billion and 14 billion parameters models.

Throughout the session, live testing with real-time examples underlines practical strategies and considerations for achieving optimal results. Users can adapt their setup based on available resources, balancing between model complexity and hardware capacity for efficient video production. This accessible guide empowers users to explore and implement cutting-edge video generation technologies in their projects.

Chapters

• 00:00 - 03:00: Introduction I hope I'm able to see it yes let me check
• 03:00 - 09:00: Setting Up Swarm UI on Mac The chapter provides a tutorial on setting up Swarm UI on a Mac. It starts with a live session not being open, but the presenter checks the necessary configurations to proceed.
• 09:00 - 19:00: Running Swarm UI and Generating Video This brief chapter titled 'Running Swarm UI and Generating Video' primarily confirms the functionality of a system, evidenced by the simple transcript: 'yes it's working'. No further details are provided regarding the Swarm UI or video generation processes.
• 19:00 - 26:30: Using Different Models and Prompt Setup The chapter discusses the process of using different models and setting up prompts. The initial part involves troubleshooting audio issues, as indicated by phrases like 'am I audible' and mentions of delayed responses, but ultimately confirms that the setup is functional.
• 26:30 - 33:00: Detailed Explanation of Swarm UI This chapter provides a detailed explanation of the Swarm UI (User Interface). It covers various aspects and functionalities of the interface, helping users understand how to navigate and utilize the features effectively. The chapter likely includes discussions on the design principles, interactive elements, and customization options available within the Swarm UI, although the incomplete transcript limits a comprehensive overview.
• 33:00 - 47:30: Viewing Generated Video and Experimentation This chapter covers the process of viewing generated videos and conducting experiments to validate the results. The focus is on assessing the quality and accuracy of the generated content to ensure it meets the desired standards. Techniques and tools for effective experimentation and analysis are discussed, with a particular emphasis on iterative testing and refinement. The chapter also addresses potential challenges in the video generation process and offers solutions to overcome them.
• 47:30 - 58:00: Using Swarm UI with Comfy UI and Other Resources This chapter introduces the concept of Swarm UI and explores its integration with Comfy UI along with other resources. The speaker begins by engaging with the audience, asking about their online presence.
• 58:00 - 73:00: Memory and System Requirements The chapter titled 'Memory and System Requirements' discusses the basics of what memory and system requirements entail in the context of computer systems. However, from the provided transcript, it seems that the speaker is unsure or unclear about certain aspects ('I don't know I'll be see let me see how you can see this part so this is the first time'), which might imply that this is an introductory or exploratory discussion about memory and system requirements.
• 73:00 - 87:00: Using Swarm UI on Different Platforms The chapter "Using Swarm UI on Different Platforms" introduces the process of testing Swarm UI functionality across various platforms. It indicates the initial success of getting the system operational and sets the stage for practical applications of Swarm UI in diverse environments.
• 87:00 - 95:00: Using Models and Data Science Projects Discussion The chapter titled 'Using Models and Data Science Projects Discussion' covers the exploration of 'swam UI', an interface for testing image generation capabilities, particularly version 2.1, which is a repackaged version of comy UI. The discussion focuses on how swam UI facilitates the usage of this updated interface.
• 95:00 - 110:00: Conclusion and Next Steps The chapter provides a guide on testing software on different operating systems such as Mac, Windows, or Linux. It suggests using various platforms like Collab or Runpod for the tests. The choice is left to the user based on their access and preference. The chapter also includes a brief walkthrough on how to initiate the setup, although specifics are tailored to individual system choices.

How to run Wan 2.1 Model - Text to Video/ Image to Video VLM using Swarmui Transcription

• 00:00 - 00:30 I hope I'm able to see it yes let me check
• 00:30 - 01:00 I'm live it is not open so let me just check
• 01:00 - 01:30 yes it's working
• 01:30 - 02:00 am I audible audible hello hello hello it's working okay it is little bit delayed but yeah it's working now let me just test it out
• 02:00 - 02:30 sh for
• 02:30 - 03:00 yeah so let me see how much of
• 03:00 - 03:30 you how how many of you are online Okay so
• 03:30 - 04:00 I don't know I'll be see let me see how you can see this part so this is the first time
• 04:00 - 04:30 so okay now it is working it seems so today we are going to test swam
• 04:30 - 05:00 UI I hope this is visible somewhere yeah so we are going to test swam UI and uh using swam UI we will be testing how we can uh use when 2.1 so when 2.1 is uh like comy UI repackage sort of thing uh that you can use so it's a basically uh image generation uh interface out there uh that you can use
• 05:00 - 05:30 out there so I'll just I'll test it out for my Mac and you can do the same on your windows or Linux if you have the uh setup access okay so for this what we need to do is uh we will be using okay we can also use collab or runp pod okay so I'll not be using collab or runport but you can just do it so you'll just click here it will take
• 05:30 - 06:00 you to your collab and then from there you can just select the runtime okay uh Run 10 type select if you have a00 you can select a00 or you can use T4 GPU also uh depending upon your like a number of uh if you have Pro then like collab Pro or Pro Plus then you will be using this a800 GPU okay and then you can run it out okay so this is one way of doing it but I'll be using
• 06:00 - 06:30 the Mac version because I have 48gb MacBook out here uh so I like to test that out there itself okay so let me just check whether uh yeah stream is working perfectly so so this is how I'll share the link on the chat let me see how I can do that try to my
• 06:30 - 07:00 see live chat I can do that or not we see
• 07:00 - 07:30 so uh to install it in Mac you just need to uh install install net using um Brew so the first thing that it is saying that you know you need to update your uh Brew update so we will be doing Brew update so it is just updating home brew right now and we will see so it is already uh
• 07:30 - 08:00 updated okay there are 30 outdated formula instant we can remove those but I right now I I'm not going to do it so uh we'll see I have already installed b.net so let's see Sol okay yeah three of them are watching right now so I'll be using Brew install.net so yeah it is already installed so just
• 08:00 - 08:30 for those who have just joined right now so we are today we are trying to install uh a comy UI like basically swam UI U swam UI we will be using and uh we will be using diffusion models to create videos okay so we will see it takes a little bit of time so yeah so where were we we uh we had just uh installed Brew update uh so I'm
• 08:30 - 09:00 doing it on Mac so you can do it the same with like know windows or Linux whichever we uh that you have so for Mac you just need to do a a brew uh from Brew you do Brew update that we did right now and then uh you can uh install net so net is uh required uh if you don't have net you will not be able to uh use it and one more thing uh if you don't have python you need to install it so install
• 09:00 - 09:30 3.1 or uh 12 also is okay so I have 12 in my laptop uh so it's okay U otherwise 3.13 is not working okay uh so what we need to do and then you need to uh uh change the director SW UI and then like you know clone this repository and uh launch it out there okay so uh so what I have done here is uh I'll go to my code what so this
• 09:30 - 10:00 is yeah so I am already on the uh swam you like know swam we folder so this is what I have cloned it out there so uh from there uh or you can just do is like you know what you can do here is we you can simply uh take uh this part to close Don it out
• 10:00 - 10:30 there okay and wait a minute yeah and you can just paste it here let me see how much time it's taking there is a little bit delay here so I'm not able to see so you just need to copy it here so that like no just run it get clone and it it will automatically clone it out there okay uh
• 10:30 - 11:00 for yourself and then you can just uh run the command Okay so that it will uh work for you so what you need to do here is you just need to clone it so I had already done it from this this person I have done it and tested it out so that's why it's already there so then after cloning it so you need to clone it first okay and then you need to uh
• 11:00 - 11:30 CD to like you know swam UI okay so basically CD will like you know it will create uh directory out there and know change the directory to that portion and then you just need to uh run this part okay so uh we will just run it right now so let's run it so the moment uh it runs so you will get something like this
• 11:30 - 12:00 so this is your swam UI basically so what you need to do you want to do a text to uh uh video or uh image to text so that depends on you so you can just uh from this options there are Advanced options also you can when you click it here you'll get advanc options to have various values out there okay uh so what I will be doing today is like we'll be doing text to video U okay so this text to video you can increase the number of frames here so this will make it uh like
• 12:00 - 12:30 you know you can make it uh maybe last time I did uh 60 so I'll make it today 120 uh and it may it may increase and last time when I tried with 90 I got a video that was uh that was decent quality I'll show it to you right now after after this thing and then U it was around uh it was taking some time so it was uh I'll just I'll show it to you what was okay let me go to my
• 12:30 - 13:00 downloads okay so this is what I got it okay so this this was the video that we created so let me close it down there so see this is this was the like know uh this was the video that we created okay why all the these are openings okay yeah so this was a webp that we have created uh so see this was
• 13:00 - 13:30 an image of dog walking on the uh uh like grass so this is what we created so what we need to do here is uh we will be uh giving it a prompt so we'll let's give it a prompt uh uh because it's we will have to first we will see which model to use okay we have to like know so uh once this is opened you need to copy paste your uh this thing one more thing that we need to do is we need you need to go to swam UI and then copy copy
• 13:30 - 14:00 paste the models so the models is something that you need to download from this uh play place. comy org so when 2.1 comy oi repackage so you can install these things so what I have done is Ive downloaded this uh 14 uh 14b billion version also and then 1.3 version also so this is a smaller version so I downloaded it and then you in the swam UI part you need to uh put it there so I
• 14:00 - 14:30 I just like know because it was already set up in my system I forgot this part so what you can do is you will go there and again wherever we have put our swam UI okay this in the swam UI part you need to go to wait a minute the models yeah then you need to go to diffusion models and so what we need to do is in the swam UI models in the diffusion UI you just uh give a need to get download
• 14:30 - 15:00 it there and paste it there whatever you have downloaded so uh so I downloaded the 1.3 billion part also and 14 billion also okay the 14 billion will take some time uh but uh the 1.3 part uh will be something that uh it will be easier for you to like generate a video okay so let's uh stop it out there and what we'll go back to our uh image generation part
• 15:00 - 15:30 okay uh so let's me try a cat so if I do 14 so we'll see here yeah this is started now uh what will happen is you need to give cat and a dog [Music] fighting in a playground so you can give some negative prompt also but uh I'm not giving any negative prom prompt as of now and then
• 15:30 - 16:00 we will generate it out there okay so now uh one generation uh one CU and one waiting on like model load so basically it is trying to load the model right now so it is trying to load this like know 1 14 billion model so as it it will take some time so it's better that we will move to uh uh 1.3 billion model
• 16:00 - 16:30 only so I'll just close it down and rerun it so it will open a new tab so this is gone so in the new tab again we'll go to uh first select the model so we'll select the 1.3 model okay and then this setup is already there uh from text to video okay so we'll uh uh this this prompt is there so what we need to do
• 16:30 - 17:00 now we will generate it out okay so let's generate this part so yeah so see it has started uh generating now okay so the mod this will this will take some time uh to generate the model and uh it will be like I think around an hour last time it took me for to generate the dog video so let's see how much time it has taking today okay so we'll see how it
• 17:00 - 17:30 works okay till the time I'll let me check me if there is any comment or anything no
• 17:30 - 18:00 how can I share this video let me see for
• 18:00 - 18:30 so let me
• 18:30 - 19:00 see yeah so we are we are generating this using swam UI so those who have joined late you can just go back
• 19:00 - 19:30 so this is a yeah like you know someone asked me in my chat that whether it is free or not yeah it's completely free you can generate it uh freely so what you need to do is you need to set up it out and then uh so I'm just testing it out I'm just seeing Oh many viewers are there okay five views are there there are two concurrent viewers nice
• 19:30 - 20:00 so see now my system uh it even if it is 48 GB to run the system uh it's it's taking time like you know we'll see the activity monitor so see physical memory is 48 even when it is 48 memory US 33 GB so that is gone right now okay so the python 3.12 is taking 20 GB by itself
• 20:00 - 20:30 so uh this is how like know if you are using a low version uh image like low version G or like know system it won't uh help you out so see uh this is started to generate something right after blue so it is started to generate something out of it so it will take some time uh it takes around an hour uh to have this thing uh so what you need to
• 20:30 - 21:00 do is you need to go to uh swam UI page so I'll provide that link of swam UI uh so uh I'll provide you the steps there so basically like you know I'll just open a text maybe open something yeah let me check it out yeah I I'll have this thing so uh you need to go to swam UI and and then from that one UI like you know install
• 21:00 - 21:30 it using Linux or uh Windows okay so uh you need to install swam UI then you can use it in collab also right so you can try it on collab and then you can uh see that part but I've have tested it on Mac uh so Mac it is running perfectly so it is generating as of now see uh this is
• 21:30 - 22:00 uh uh it is generating it is like a cat dog cat and cat and dog fighting in a playground so it is generating it will take some time uh it takes around like know when when I'm having a 48 GB uh MacBook it's taking time and it's like you know it's sounding like crazy right now you you might not be uh listening to the background noise but it's like you know it's giving some sound to me as of now so so you can use it like if you
• 22:00 - 22:30 wish to like you know use so basically uh The Comfy workflow is also there uh in it uh so these setups you can change the parameters uh so I have made it a 90 and like these setups you can like you know have it out so if you want to have boom brang and all you can have that format you can change the format part also you want G like GF format web P or like webm or MP4 format also as uh h264
• 22:30 - 23:00 so this is something that you can easily do it and this van 2.1 is a Alibaba uh large language model that you can use and so it is uh completely open source as of now and it's not that restrictive you can try to generate whatever thing you can imagine okay so that's something that uh you can try it out okay okay so uh uh let's uh let's see
• 23:00 - 23:30 where we are in our generation so it will take some time see some of uh it is are generating as of now uh so it will be around maybe 15 to 20 minutes we will be seeing the cat and to at least so see there are two figures I think you know it's showing somewhere so we'll see how that works so you can this this works in a St using a stable Fusion so you might need to you
• 23:30 - 24:00 might not need to if you just an user who is using it out for creating videos from text to video or like an image to video you can just directly use it as of now okay uh so you can just use it uh for your own use case let's say you want to have a flip cart video so like know you have the image that you use on in N number of platforms so what you can do is you can create uh these videos using a a image uh in these sort of platforms
• 24:00 - 24:30 and then you can use it everywhere okay so we will see uh where is my so code see it is like you know once it it is it will be finished so when you're doing it first time it will it will say that it is downloading that image also uh downloading this model so I'm doing it for the first time so that's why it is not giving me uh the final uh it is already giving me it is there because it has downloaded so what you need to do if
• 24:30 - 25:00 you have joined late so go to uh the first thing that you need to go is you need to go to swam UI in the swam UI part you if you want to uh install it in Windows or uh if you want to I'll just uh maybe I'll share the link here somewhere um wait a minute
• 25:00 - 25:30 this let see if I can do it from here let me see
• 25:30 - 26:00 it's not working from here so yeah so let me check that generation part still uh it's uh I think only 10% has been done so the Green P is here so once the green person comes here so you
• 26:00 - 26:30 will be able to see that you know cat and uh dog fighting in a playground so so let's see uh how much time it will take U we'll be waiting for it uh otherwise you can go to Swami if you're on your laptop go to Swami and then uh this com.org uh like know U hugging face page so so this is this is the page from where we will download it out okay so
• 26:30 - 27:00 maybe let me change it somewhere this YouTube description I will just copy paste it out
• 27:00 - 27:30 so
• 27:30 - 28:00 okay so see it's starting to get something okay so in the
• 28:00 - 28:30 meantime we'll be talking about our like you know how how can you how you can use it like you know you can use it from here also like you know but uh when you try using it so it's uh quite big uh like you know from hugging phas if you're using it so uh this safe tensor is given so you can try uh downloading this part uh but uh I have downloaded a comfy UI repackaged version so that is little bit smaller so you can directly
• 28:30 - 29:00 use this part also like diffusion P Touch model safe T So this model will also uh give you a decent uh uh like know decent resolution out there so and if you want to use it so if you want to see some demo so like you know so they have [Music]
• 29:00 - 29:30 w
• 29:30 - 30:00 so so you can like you know go through this
• 30:00 - 30:30 hugging pH also and from there also you can use it but uh like I would suggest going through swam UI but as Swami is like just one click thing that even the non Tei can do it so let's see where are we at our image generation part still like you know the video is not being generated uh as we like I think there are some fighting happening in background so one black IM and one white image is happening here but yeah so it
• 30:30 - 31:00 will take a bit of time see it's still going on so let's wait for it and then see so again so those who have uh joined late uh so today we are just focusing on swam UI so swam UI will give you access to create uh something uh that like you know you can use any
• 31:00 - 31:30 diffusion model out there to create videos or maybe images also so depending on like you know uh uh so for my current use case I'm using this thing um generating an image uh so it is a text to video thing okay that we have done and then you have like know image to video also available out there so that portion you need to test it out okay so I have tested text to image so see
• 31:30 - 32:00 now two figures have started coming up so see these two figures have started coming out so it has it has been around 30 minutes I think yeah uh not 30 minutes around 20 25 minutes so it will take some time to generate this portion so there are lots of like you know features that you can use U so that portion you need to like you know give
• 32:00 - 32:30 the uh give the whatever options you want you just select that part and give like a number out there so I have wanted one IM like only one image you can give another image you can give a different seed also you can check the uh steps also here so based on that it will be generated so here I think around 20% is generated and so background now you can see the tail wagging right so tail of the cat I think cat or dog I don't know which it is so
• 32:30 - 33:00 it's wagging right now and it's some sort of playground I don't know what sort of playground is this but it is it is showing some sort of playground right so we'll see now how it will come on later on for
• 33:00 - 33:30 [Music]
• 33:30 - 34:00 so you can see that like you know they have used 3D various Auto encoders and the efficiency of van is better than other models out there so opur plan there these are the open source models some Chinese models out there uh you and video and all those scenarios so it uh I
• 34:00 - 34:30 think you once you set up set this up in your local systems you don't need to use anything else or you can use it through another cloud provider also but uh if it is for a personal use you can always set up in your local systems okay
• 34:30 - 35:00 yes something is coming up but yeah it is taking time so so by the time like you know it is coming up I'll show you what I generated last time so so what I did was I generated this portion so this was something that I generated yeah it is not opening
• 35:00 - 35:30 here good man yeah you can see this so this was my prompt I create a dog walking on grass okay so like it took around uh it was yesterday like you know yesterday I did it so so see how good it is like know still it is it is on 1.3 uh 1.3 billion model so if you can if you're using any other model uh it will be coming uh if you if you use the
• 35:30 - 36:00 14b model that will be giving a better like you know results out there so I have not tried on 14b because it will take far more resources than like know it is currently there so see like the resource activity as of now is it is taking 19 uh GB for my 1.3 billion parameter model so and still some of it is there but I don't want to like you know extend it out so see the
• 36:00 - 36:30 cat and dog some of it is like you know now it is able to generate I think the white is the cat and the black is the uh White is the dog and black is the cat I think I guess so this is how it looks likes as of now we don't know how it will be and what sort of green background they are trying to create so let see let me see live chat here e
• 36:30 - 37:00 so van 2.1 has their own like you know
• 37:00 - 37:30 websites also so you can go there and check it out from there also so we'll see this is this is the model by Alibaba so I don't think this is their own website yes I think someone has created this part
• 37:30 - 38:00 yeah it's their own so let me check there think it will show okay because they have not given their website here so I'm not Lo yeah this is their website yeah so we were on their website itself so you can see like know this is what
• 38:00 - 38:30 they have created okay so you can see like you know what sort of prompt they have used a CP on Vintage photograph depicting a whim School bicycle race between several dogs wearing goles okay so these are on like know see the dancing part the boxing also cat boxing yeah so motorcy vehicle racing so I think these are on the 14
• 38:30 - 39:00 billion uh model not on the 1.3 model that we are trying today so still yeah it will take some more time I guess because it's able to generate a better version of it yeah now C cat and dog are still there but yeah until unless it is 50% sort of rendered uh it will take
• 39:00 - 39:30 time till then we will see this videos that is there so there hip hop also they have given a prompt and it is generated so if you want to use something you can use it uh so it's an open source model uh completely and I see so consumer grade GPU so 1.3 model requires only uh six 8.19 vram but like you know as I'm using using swam UI
• 39:30 - 40:00 I think that's why I'm getting like you know uh far more I have to use it directly so I'm not using it as of now because it's perfectly uh coming for me so yeah to see people have started using it and quantize it also so depending
• 40:00 - 40:30 upon your use case you can just quantize it but I would suggest you to use the M direct 1.3 version so that looks good only yeah see now the tail is wagging somewhere the dog but the cat still we are not able to see I don't know whether this is cat or the other one is cat so
• 40:30 - 41:00 still indistinguishable between both of them I'll
• 41:00 - 41:30 just paste it there will Super Chat will not be available until unless we have some views so let's see uh not right now but someday we will be doing it so see we are generating it out so we will see how much time it will take
• 41:30 - 42:00 someday I think I know I will take a end to endend machine learning project also uh during the similar timings maybe 9 to 10 and today I was just trying it out anyway so I thought let's share it on the live version instead of doing it a video because that takes time to record it and then upload it out there so it will be just direct uh scenario here so we'll see how this
• 42:00 - 42:30 works I hope it is visible let me see in the mobile yes looks good enough
• 42:30 - 43:00 and share let me share this in the chat I do I'm doing it for the first time so I don't know whether I can just click on my own live content or
• 43:00 - 43:30 not okay I will just post it here
• 43:30 - 44:00 what are we using we are using swam UI so you can just use it from here let me save it so now if you want to use it right away you can just go to like you know my description page and you'll be finding the links to download the model so from here you can just
• 44:00 - 44:30 download uh like you know the 1.3 either bf6 you take or floating. 16 both of them are same like know any one of it you use you download either of them and then you can download like know 14 billion model whichever will works for you so whichever Ram you have like or memory if there is a memory constraint for you uh so you can use 16.4 or otherwise you can use 32.8 so GB also okay so these are
• 44:30 - 45:00 several quanti version so every one is different see this is bf16 this is fp8 this is bf16 again and this is fp8 so this is for 480p and this is for 720p so both of them are different and again um this is bf16 fp16 for 1.3 so they have also given like bf16 version
• 45:00 - 45:30 of uh this is I don't know for which one they have used this is 420 this is nothing is mentioned so you can try downloading all of them and then make can try each uh one separately also so yeah something is coming up I I don't know what sort of playground
• 45:30 - 46:00 they using it out here so I hope we get a better playground
• 46:00 - 46:30 so you can go to the description of this video if if you want to try it right away and uh download uh swam UI so in the previous links I have told how you can download swam UI so just uh download the swam UI if you are trying it on collab uh you can uh try in collab also wait a minute if you want to try R and collab uh so you can use these links
• 46:30 - 47:00 there and if you're trying it on Windows you can just install using this uh instructions out here so you can you need to download the bat file and install it there okay so it is it is asked not to install in your program files you can install it anywhere else not in uh this poster okay or you can do the manual install from here
• 47:00 - 47:30 also I'm using it on Mac so in Mac you can uh use it as so I have seen you the like given you how you can use it so just sorry so just go to for Mac you just need to open your terminal and then from terminal you just we will do you will check whether Brew is there
• 47:30 - 48:00 or not so if you want to you update Brew so you can update brew and once it is update it is already updated so what you can do is you can do Brew install.net these are mandatory things that you need to
• 48:00 - 48:30 do yeah so it is already installed so once this this is installed uh uh what you will do uh so you you will have to go to a folder so let's say I want to go to coding part so see I will do CD coding from here okay this is my folder so coding I have gone now you just uh install it so get clone so either you can do it from here or you can just do it from your vs code so I
• 48:30 - 49:00 did it from vs code person so there have created it out and then just I have copied it there okay so you can do it from your terminal also so just do a get get clone uh that uh person and it will uh clone it out just need to copy this part and run it out nothing extra okay just copy this portion and run it and then uh you you
• 49:00 - 49:30 need to create once this is uh downloaded in that particular project then you need to do again do CD swam UI and then run this portion that we have done today so what we did was we did it and then we we ran it so once uh once we have done this uh code swam UI CD CD CD swarm UI then you need to run this SL uh launch Mac cox. sh so after that part
• 49:30 - 50:00 uh like know server will be open and then uh it will be coming in your Local Host SL uh the port ID so Port ID default is 7801 okay and then check you can do yes you can do like know if you want to use it in aw server you can do that uh also so depending on your budget so
• 50:00 - 50:30 you can just uh host it there not any you can because it's there is an option of uh option of installing using do Docker also so if you have like know the resources in AWS just like know go to Docker uh to check this like know file so uh you can install using Docker in AWS or any other like know cloud provider so that won't be a challenge out there okay so go to this swam UI and
• 50:30 - 51:00 then go to the docker part okay and let's see okay I think this one the first one is the cat but the the dog is the outside version of it or there are maybe multiple dogs that I need to see still generating
• 51:00 - 51:30 so yeah we were uh we were at like installing this part then in your terminal what you need to do is you just need to uh run it so like once you are into your uh swam UI folder uh then you just need to run this launch uh do/ launch Mac ossh and then if you are doing it for the first time it will just download the uh it will it will not download it basically it will open this
• 51:30 - 52:00 UI it will open this uh image generation UI so from here you can select the parameters that you want to have and here you can select the model So currently I'm using 1.3 model so once you are selecting any other model so I I also also downloaded the 14 million model also so but I using this potion and you can use uh these other things also okay so I have downloaded uh this model
• 52:00 - 52:30 and I'm just selected it here so I have kept it as it is uh for 14 million also you can do the same but uh it will take a little bit of time of it so uh uh to download this model and if your memory is out I hope you don't get out of memory issue out there otherwise it will be a problem so what I have done is I have done I've selected 1.3 model and then I have just run it out there so once I've selected it out so what will happen in your terminal is it will
• 52:30 - 53:00 download it and then it will show that you are giving something okay so you once uh so once you need to download it here and then you need to give a prompt so prompt you need to give it here okay and then generate it so once you start generating this model download one it will it will giving one current generation one current training it is giving right now otherwise so uh what would what will
• 53:00 - 53:30 happen is uh it will give you uh the whole scenario okay so let's see how it is so it has come up to 50% and some cat and dog is there like Know video is starting to getting uh starting to get somewhat clear and we'll see uh how this comes up so for those of you tuning in late just go to the description and down down load uh this like know P install swam UI uh portion or like know you can
• 53:30 - 54:00 just not P install sorry like you can just clone it out there and I have given the link of conf like confy confy organization also from there you can just use this diffusion models like you know either one of them like you know whichever is comfortable with your system just copy it out there and then you can run it because how do you run it you just uh follow the steps given here uh so this is the step to install
• 54:00 - 54:30 in your Mac and then when you run it you just need to go to uh this portion to uh like you know uh you need to Once uh your sorry uh where is it yeah this coding portion so once you have installed swam UI okay and inside this form UI you need to go to the models and in the diffusion models you
• 54:30 - 55:00 need to paste the downloaded models so this is mandatory once once you do this then your models will be perfect and uh like you know you are ready to use it and just you will need a prompt a single prompt from here on uh will be helpful for you to generate a model so what you uh just copy this uh here and you need to go there and select your model from here select a
• 55:00 - 55:30 model OKAY select this model and give a prompt you can give a negative prompt also and then click on generate that's it and it will take some time to generate this thing okay so for me it is taking uh like you know we have been on it around I guess 40 minutes 40 to 45 minutes and some of it is generated uh still not uh that good but yeah I think it will take some more time so we'll
• 55:30 - 56:00 see I hope uh like know you can comment uh if you need any other data science video or anything that uh we can do a live session on it so uh we I'll be doing on a regular basis I'll be trying to set up the live uh scenario so da I will set up something uh something or or other so maybe uh one day we'll create a rack chatbot or PDF R chatbot or website chatboard that we can create
• 56:00 - 56:30 here it out there still still taking some time
• 56:30 - 57:00 let me Che for
• 57:00 - 57:30 image to video like you know I'll have
• 57:30 - 58:00 to try it out so I have not tried it as of now uh so we'll see uh once this is done maybe I'll do it out so we just need to go to image to video so so I just don't want to close it out there so this image portion we need to see how
• 58:00 - 58:30 this can be done okay let me check the like know okay so let me check the portion of it okay wait a minute
• 58:30 - 59:00 so I will just look into it let me check the dogs
• 59:00 - 59:30 this person I I'll have to check it out so I'm not getting it as of
• 59:30 - 60:00 now yeah still blurry even after uh guess 40
• 60:00 - 60:30 minutes but yeah it's I guess it should work so let's see e
• 60:30 - 61:00 so we try something else in the meantime
• 61:00 - 61:30 okay we can use the comfy work flow also
• 61:30 - 62:00 here so I'm not going into it as of
• 62:00 - 62:30 now let me check okay leave
• 62:30 - 63:00 it still generating so I'll see my but the meantime I'll just check something else
• 63:00 - 63:30 okay
• 63:30 - 64:00 so I think
• 64:00 - 64:30 so let me see think okay this is again [Music] diffus yeah so this is an image to a video version that is there so we need to use the 14 million model not the uh 1.3 one that I am currently
• 64:30 - 65:00 using so we need to use it and we will see how it works so not today maybe some other day I'll be using
• 65:00 - 65:30 I guess I'll just show it to like know how it is the coming to the final one so and then we'll just s it up because it is taking some time I think it will keep on generating so no need of waiting for it and
• 65:30 - 66:00 then so let's stop streaming so you can go and check the recording if you like uh if you want to install it and I'll just stop it out maybe and then so create a new video small video