INSANE FREE Outpainting Tool For Stable Diffusion! BEST ONE YET!

Key Takeaways

hello humans my name is k your a overlord and oh my god do I have an insane tool for you today now if you remember correctly a few weeks ago I created a video where I showed you a special website called pen.com that allowed you to connect to your local stable diffusion installation and use it for out painting but unfortunately a lot of people thought that this site was way too risky to use so today I'm bringing you a tool that is way way better than this one with a much better UI and this tool is called invoke AI and I think that this is by far the best in painting and out painting tool out there so in this video I will show you how you can install it and use that amazing tool are you ready then let's go and to download an install inkii is actually very very simple all you have to do is just click the link in description down below you're going to arrive on this page you're going to have to scroll down to the bottom and here you're going to select the installer for your computer so in my case since I have Windows I will click on this one so this will download a small zip file onto your computer then you can then extract either using WinRAR or seven zip so simply right click and then click on extract to invoke AI then you're going to double click on this new folder invoke Ai and here you're going to see three files now the first file that you want to double click is called win long paath enable Rag and this file will basically remove the 260 character path limit on windows by editing the registry this way info will be able to be installed on your computer correctly and then to begin the install all you have to do is just double click on the the install.bat file to start the installation but before you do and this is my little advice to you because I've had the issue you might need to disable your antivirus software otherwise this might block the installation process so once you've disabled your antivirus software you can simply double click on the install.bat file and then press any key to continue and then just wait until everything is installed so it might take a few minutes so be patient so after a while you will see this text that asks you to select a directory where you will install the models and some configuration files so by default it will be in C users your username invoke AI if you want another path you can change the path right here but in my case I will just choose the one by default and then press enter now it will ask you to theault directory for image outputs again I will choose the one by default and then press y to accept these locations and then press enter it will ask you if you want to enable the not save for workk Checker by default which I'm going to say no so I'm just going to put n and then press enter press enter again to accept the default Samplers press enter again for the default number of steps and then it will ask you which model you want to download so in my case I will simply use a so that it will download all the available models and then press enter press enter again to accept the license agreements and then wait until all the ckpt files are downloaded Okay so so this is me from the future I highly recommend not actually using the All model because this will basically download a bunch of models that you may not want to use and also it will take a very long time to download and install so I highly recommend that you just press R and it will download all the recommended models and as I said previously the default folder for the models is located in C users your username invoke AI models LD M stable diffusion V1 and here you will see all of your stable diffusion models so if you want to use your own model in invoke AI you're going to have to put it right here and then once the installation is finished you will see here A bunch of files but the two main files that you need to pay attention to is update dobat and invoke dobat so if you want to update the invoke AI software all you have to do is just double click on this update dobot file right here and as you can see it will automatically update the latest version of invoke Ai and the other file invoke dobat is the file that you need to use if you want to launch invoke Ai and here it will ask you if you want to generate images using the common line the browser based UI or the developer console and of course we want the browser based UI so just type two and press enter and you will know that everything was installed correctly once you see this local address right here exactly like we still table diffusion so exactly like stable diffusion you're going to select this URL contrl + C to copy it and then paste it into your browser and now we're finally inside invoke Ai and we can start to have some fun now the first thing that you're going to notice in invoke AI is the fact that the interface is actually pretty beautiful it is pretty streamlined and yet very complete now on the left side you will see here A bunch of icons and each icons actually represent present in different workspace so you start here in the text to image workspace then you have an image to image workspace you have a node workspace that is currently under development but that will allow you to generate images using a node-based system you have a postprocessing workspace that is also currently under development a training workspace where you'll be able to tr your own Cil diffusion model using dream Booth or textual inversion and I've left the best for the last and that is the unified canvas and this is where you will probably spend most of your time and this implementation is absolutely fantastic because in one single canvas in one single workspace you can use text to image image to image in painting and out painting on an infinite canvas that is absolutely fantastic okay so before I show you all the good stuff let's actually look at what we have over here here so here is where you're going to choose between your stable diffusion models so here I'm currently using the stable diffusion 1.5 and if you want to use the Imp painting 1.5 you can simply click on it exactly like in your normal stable diffusion installation here you have a bunch of hot keys in general shortcuts that you can use to make your workflow way faster here you can change the theme of your UI so I'm currently using the dark one but you can also use the green one or the light one all my God my my eyes oh no no not the light one do not use the light one just use the dark one here you have a report bug button a button to the official GitHub repository a button to the official Discord server so if you have any question you should definitely join their Discord server and here you have a settings button that allows you to change a few parameters and the settings that I recommend you to choose is in modles you can see here an option to load the in painting 1.5 that if you click on it will make loading between models way way faster so this is definitely something that you should try if you are someone who is constantly changing between models and as you can see now the inating model is now in the cache so loading it will be way faster next time that you want to choose it and everything else you can just leave by default okay so now let's see what kind of options we have to generate our images so obviously here is your prompt box this is where you're going to input your prompt so for example I'm going to copy and paste my usual prompt that I use all the time and also if you want to use negative prompts you need to put the aggrement between square brackets and if you want to use prompt waiting and put emphasis on some words just like stle diffusion you're going to put the words that you want to emphasize between brackets and then using plus to emphasize certain words or minus for the opposite so it's really up to you on how you want to use it so here you have all your basic parameters if you already used sa the fusion before the number of images you want to generate the steps the scale the resolution your sampler your image to image strength which is basically the same as the noising strength here you have the bounding box where you can manually change the size of the bounding box the seam correction which I highly recommend that you live at default because everything works really really well by default the infill in scaling that again you should leave everything by default default leave at Auto and use patch match which is by far the best infill method then you have the seed option where you can either randomize the seed or input a certain seed that you want to reuse and the variation seed option and if you're not sure what options actually do you can over your mouse over this little icon right here and it will explain everything that this option actually do but essentially these options right here you should leave them by default because they work pretty well already out of the box but what's even more interesting is all the options that we have right here but let's first actually try to generate an artwork and I will explain every single options that you see right here along the way which I think will be a little bit more interesting so I'm going to try to generate four images of Christina hris and then click on invoke to start generating the images and as you can see it will slowly start to generate an image from noise and this process is actually pretty satisfying to watch and there you go we have now generated four different images then we can Now preview and select using this floating box right here so if I want to look at the previous image I can click on this previous button right here and for every image you have now here four different options either you accept this image and you continue with this image selected you have an show and hide button that basically show or hide the image you have the save to Gallery button that will basically save this current image to your gallery that is located right here or a discard all button that will basically delete all the images that you just generated but in my case I think I will select this image so I'm just going to click on this button right here to accept the image and now we can finally start the fun part because now we have a bunch of options that we can choose from so let's start from left to right here you have a choice of a different layer here we have our base image layer where we can paint using our brush tool or erase part of the image but we also have a mask layer and now our brush is now transformed into a mask brush that if you start painting on the image will create this new mask effect so next time that you want to generate an image this part of the image will be replaced by the next one so in my case since we have here A bunch of wide borders I will actually use the mask to delete these wide borders because I don't want them in my final image and if I did a mistake for example and I want to erase part of the Mask if I choose the Eraser tool this will not erase the image this will only erase part of the Mask which is like super super convenient because now if we choose the base layer and I have the Eraser tool selected if I try to erase the mask it will not erase the mask it will start erasing part of the image and I personally do not really like this workflow because this is a destructive workflow it is way better to use the mask layer and if I want to undo the changes I can simply click on this button right here or click contrl Z exactly like if it was Photoshop which again is super super useful so then we have also a masking option that allows you to change the color of our mask if you want to or a preserve masked area which is basically invert masking so everything that you selected with the mask will actually stay so for example if I only want to keep the face but not the rest of the image I can basically select the entire face and with The Preserve mask area option if I generate a new image only these areas will stay in the final image so this little section right here is everything linked to the brush tool in the brush to basically allows you to paint part of the image using a certain color so here you can use the Color Picker when you can choose a certain color here you have a brush options when you can change the size of your brush and the color of your brush and the opacity and then you can start painting on the image directly and what this will do is that the next time that you out paint the image in this area it will actually take into account the color that you just painted right here which is basically super useful because it gives you way more control over the final results and finally you have here the fill bounding box that will basically fill the entire box with a unique color and the erase bounding box area that will basically delete this entire box area so here you have the move tool that allow you to basically move this bounding box and also allows you to manually resize it to the size that you want and you can see here exactly what is the current size of your bounding box so the reset view basically allows you to recenter your view so if for example you were working on the side and you want to go back to the center you can click on this button right here or or on your keyboard it will basically reenter your image to the start the emerge visible option is basically only if you have performance issues on your computer and this will basically consolidate all the information currently being rendered by your browser and this should improve performance if your PC is struggling then you have to save to Gallery button that will basically save your image to the gallery a copy to clipboard that will basically copy your image into your clipboard then you can paste it somewhere else like in Photoshop for example and download as image that will basically download the entire canvas as one single image if the undo and redo if you want to undo or redo a certain action that you did previously and the pl button that allows you to upload an image from your computer directly into the canvas and the clear canvas button that basically delete everything in the canvas and then a bunch of canvas settings that you should probably leave by default unless you want to save the Box region only so if you have a big image and you don't necessarily want to save the entire image but only part of it you can activate this option then resize your bouncing box place it wherever you want and then press this Buton button right here to download as an image or to save to gallery and this will only save what's inside this bouncing box but in my case I will just leave it by default now the text to image workspace also has very interesting options first it allows you to use the face restoration technology by using either gfp Gan or code forer and I usually recommend using Code forer because it usually gives the best results you also have an upscaling option that allows you to upscale your image into higher resolution you can of course use face restoration and upscaling at the exact same time but you can also use them after the image is already generated so let's say for example I want to upscale this image I can simply over my mouse on this button right here and then click on upscale image and in a few seconds I have now an upscaled final image and same thing if I want to use the restore phas option as you can see right here I also have an option to have more info on the image by checking out the prompt the seed and basically all the parameters that were used to generate this particular image and once I have an image that I like I can click on this button and I have now plenty of options to choose from I can either send it to the image to image workspace I can send it to the unified canvas I can copy a local link to the image or download the image directly onto my computer and if you want to use any of these parameters in the unified canvas you can click on one of these three buttons right here you can either transfer the prompt the seed or everything at the same time and all the options that were used to create this image will be transferred to the unified canvas which again is really super useful so let's actually try to generate a new image and trying to merge these two images together so instead of Christina Hendrick I'm simply going to be using scy Johanson four different images and then I'm going to start generating these images so we have now generated four Images I think I'm going to choose maybe this one click on accept and now I will try to merge these two images together and to do this I'm going to choose the special in painting model which will usually give me a way better result and then again click on generate and this is the final result I mean come on now this is really good D two more like Deli who who needs Deli too when you have something like this for free on your computer come on now and if you want to hide the mask you can simply click on H on your keyboard so yes this tool is really super super cool and you should definitely try it out and there you have it folks now you have another brand new tool that you can use to inate and out Pate using stable diffusion and all of that for absolutely free so try it out because this is definitely worth it and there you go than thank you so much to my Patron supporters for supporting my videos you guys are absolutely awesome thank you so much for watching don't forget to subscribe and smash the like button for the YouTube algorithm and I'll see you guys next time bye-bye