Install StoryDiffusion Locally on Windows with 1 Click

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
just yesterday I created a video on story diffusion and showed you how exactly this works behind the scene from the research purposes but in this video I'm going to show you how you can get it installed with only one click on your windows local system before I do that let me refresh your memory as what exactly this story diffusion is story diffusion can create a magic Story by generating consistent images and videos story diffusion can not only use self attention for character consistent image generation but it can also cater for long range sequences it is hot plugable and compatible with all STD 1.5 and stxl based images diffusion models it also has got motion predictor for long range video generation which predicts motion between condition images and in a compressed image semantic space and that is that is how it achieves larger motion prediction in very simple words you can generate comics and you can generate long range videos with the help of story diffusion in fact you can create the whole visual story with it so enough talking let me show you how you can get it installed locally on the windows the tool which I'm going to use use it called as Pinocchio if you don't know what that is I have just did a video on my channel and I will quickly show you so that you would know how to install it sorry let me quickly search Pinocchio just type Pinocchio and search there you go so this one doesn't work but this one I just did it one hour ago so you can simply go and check this easiest tutorial to install Pinocchio okay so I have already launched Pinocchio now just click on discover the first time you'll install you'll be presented with this screen here simply type story and you go so you see there is story diffusion Comics click here click on download and the first time you'll do it there'll be it will take bit of a time to download so let's wait for it to download the GitHub repo of story diffusion is downloaded on the left hand side in Pinocchio just click on install and then it is going to install it on your local system and that might take bit of a time you see that it has git cloned the repo and now it is installing it so let's wait for it to finish and the great thing about this is that you don't have to worry about any of the get commands Linux and what to install what not to install all the prerequisite setting up virtual environment nothing so just let's wait for this one to finish and while that happens let me give you bit more information around some of the technicalities of it for example um you can also leverage the image produced through their consistent self attention mechanism and they can extend the process to create videos by seamlessly transitioning between these images so primarily this is a two-stage long video generation approach also they have their own project page which I can show you quickly so if you go to this this is the project page if I scroll down you see these are the comics they have generated like jungle Adventure Moon exploration and then you see this is a long video generation and story diffusion has generated them by image semantic motion predictor how good is that look at this one in different um consistent self attention there you go look at this human s of things amazing amazing results if I just make this one a bit bigger you will see the expression on it look at the Expressions simply V Factor amazing and then look at this enemy one and this is just full of examples upon examples and you'll be surprised to see the quality of it I'm very very impressed by the story diffusion and then you can even do cartoon character Generations multiple character Generations as you can dog and human and in different poses there you go and this is a reference image and when winning the tuning award and this is a Jeff Hinton the Garland squirrel and this is the whole architecture which I was trying to describe where it is showing the pipeline of story diffusion to generate subject consistent images and this is a self attention if if you want to know what exactly self attention is please refer to my other video which I just showed you at the start where I go into way more detail okay so I'm not going to bore you further with the detail of this but look at this amazing amazing model okay let's check out our uh Pinocchio so you can see Pinocchio is still working so let's wait for it to finish I'm just going to make it smaller so that you can also see what is happening okay so it is still installing so let's wait story diffusion is installed as you can see here it it ask you that click the start tab to get started so click on start Tab and it is going to take bit of a time to get started so let's wait for it to finish it is loading the model now and that is going to take bit of a Time it is still loading the model so let's hope that my 16 GP uh GB GPU card is able to sustain it because that is one limitation I'm bit anxious about but anyway now we know that how can we get it installed so let's wait for it to finish and then it is going to launch it in the browser and we will check it out okay thankfully um I'm so glad to report that it is working now so you can see that everything is loaded and uh running so in order to open it just simply uh click on this open web UI and it is open going to open it here as you can see here that is a gradio demo you can even if you go back to the terminal you can even access it locally on Port 7860 so I have just open it in the browser I will just make it little bit smallish there you go so this is running in Local Host 7860 let let me make it bit bigger again again so that you would see what I'm doing now in order to run it all you need to do is to just start filling up these values I'll go SD model I'm going to use unstable one let's try to give it the character description so textual description I'll say a man viering PL suit let's also give it some of the negative front which shouldn't be included so I'm just saying bad Anatomy bad hands missing fingers that sort of stuff and now let's give it uh our comic description now you can this is your own Innovation which you can use and I'm just going to use Japanese enemy and my comic description is this let me make it bit bigger for you sorry if you can't see it now I think you can see it at home read uh newspaper at home the newspaper says there is a treasure house in the forest blah blah blah so it's a Hu big whole story line here and then I'm I have selected Japanese enemy and then if you you just scroll down there are a lot of stuff which you can set but I'm just going to keep it as is and then let's click on generate so it is going to take bit of a time now this is where it is running so let's wait for it to finish and then we will check it out so there you go it has generated our comic and how nice it looks and you can SC scroll down to see all the images it has created and my system is very very slow just due to that look at that this is the whole enemy that's it guys I hope that you enjoyed it let me know what do you think about story diffusion I think this is simply best and I will drop the link to it in video's description if you like it please consider subscribing to the channel and if you're already subscribed then please share it among your network as it helps a lot thanks for watching
Info
Channel: Fahd Mirza
Views: 4,135
Rating: undefined out of 5
Keywords:
Id: grUTY6a88xo
Channel Id: undefined
Length: 8min 54sec (534 seconds)
Published: Sun May 05 2024
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.