FREE Ai Art / Stable Diffusion Photoshop Plugin! (SDXL 1.0 using ImageCreator)

if you want to be able to create air and images right within Photoshop you should check out image Creator by LK Vision it's a free plugin you just need to sign up for an account and download the plugin and install it in Photoshop and it's pretty straightforward now arcade Vision are the sponsor of this video if you get a chance check it out with the link in the description below if you want to learn more so to get started head to imagecreator.alcadevision.com and from there you can go in here to download and you can choose which format I actually recommend the Creative Cloud format if that doesn't work and you've got Mac or Windows and then when you're ready you can actually go to sign up I click sign in it says no account so I sign up enter my details and go from there now I have Photoshop open I go to plugins alcode vision and image creator a sign in gives me a little information about it and then I'm good to go but I do need to open a document first in order to get started it's now going to get straight into image generation within Photoshop using text to image so the idea here is if you've got Creative Image credit open here on the right uh basically I type in a prompt and I can create an image there's actually a whole bunch of different checkpoints here we can use we can add more models on the website which I'll go through later on but essentially we're going to go through and find a model to work with we've got things like stable Fusion 2.0 open Journey but if we keep scrolling we've got sdxl 1.0 which is the latest version of stable diffusion so if I click on that it says this particular model doesn't support control net we just tick to keep going the positive and the negative now these are just suggested prompts you can try out negative uh is words that it'll actually kind of describe what you don't want in the image the positive we can add in some kind of prompt and try and generate some kind of image so I've got something simple here an ancient warrior standing in front of Japanese Temple sdxl 1.0 and if I scrolling I've got some more options guidance scale so probably need to worry about this too much but you can adjust this up or down if you're going to try and change how much your prompt influences the image seven is a bit of a sweet spot that's why it's generally the default choose different Samplers if you're not sure about that I would leave that as it is images this will show us how many images will be generated I can do two one I think do down to one but we'll go four so we have a selection steps now the more steps you include the more it is included the more sort of generations it goes through to create your image so technically it would be better quality but it actually gets to a point where the quality isn't really improving the images changes and 20 again is a sweet spot for that we're going to leave everything the way it is we've got Laura's we can add in today but we're gonna check on that later but uh first now we're just gonna get into it and it says here an area must be selected do need to highlight and select the whole thing I'm going to click generate so these are our images straight out of the box I can click on one I double click and it shows up on my page but what if I wanted something more photo realistic for starters I could start by playing with the prompt I can add in some descriptive words like Photo realistic image cinematic movie still 8K High detail high resolution and that will help sort of guide uh stable diffusion to make the image I want so again I hit I come down I hit generate and while it's generating you can go to your history and it shows a history of the things you have generated here and you can see what we've been able to create I really I really like this one with the hat so I'm going to double click on that and now that image is on there and go over to my layers and we have the new image in the old image there as layers I can turn on or off so we can always come back to whatever we've basically added and go from there as I click on different areas I can do things like upscale so I can upscale an image I've double clicked and it's brought it in at a higher resolution so what I can do is go to layers and you can see how we've got this much higher resolution image here I can reposition it or I can actually just duplicate that put it onto a new layout or something like that but uh so you can actually upscale within the image Creator I'm going to remove that for now I'm going to keep going because we want to try and do a little bit more with this image what I'm going to do now for starters I can choose stable diffusion XL1 but there's also another dream shaper which is XL 1.0 which has another sort of cool uh sort of like style added to it these models have been trained a little bit differently so you get different results so if I put the exact same prompt in and I select the entire canvas hit generate now again a lot of similarities I can double click on this one to import it and you can see it's still it's very similar so the style is a little bit similar to the last one but just a little bit a little bit different because of the model so you can see how we can actually choose a whole bunch of different models to generate with I'm going to change to RPG because I want to see if we get completely different look so hit generate again and check out the difference we import this one we get a different look again this one different look yet again so you can see how it's how it's very powerful when we change from model to model but there's also Laura's these are these are models I sort of add on top of the mix to get a little bit more sort of variety so I can come in here and see if there's something that I think will work well such as moxin I can click on that hit generate you can see how it's influenced the style we're still using the RPG model but now we're added the Mox and Laura on top of it and it's created some very different looking imagery but it doesn't have to be quite this full-on I can go back up and where it has weight I can make that 0.5 so I can half the effect of that Laura I hit generate again and you can see it's definitely influenced the image but it hasn't gone for that full illustration look so we can slide that scale back and forth until we get a result that we really like so some of these have turned out really good and there's some really interesting images here so the next one we're going to look at is image to image which is basically using a reference image to create AI imagery from so I've got this guy pen photo this guy from pixels imported into my Photoshop document another function here I'm going to go image to image now what I also need to do is again select the whole canvas and I've got my image the strength is at five I've got no lures selected but I'm going to choose a checkpoint so a model to work from so let's find something like good with deliberate and we're going to add so I've got a man's face of the mean stair Electronic Component cyberpunk Style robotics just to add a little something different and scroll down here generate and just see what we get so if I click on this one here you can see how it's referenced the image if I switch between the two the layout is basically the same but it's added in some of the components I asked for and again if I click on this one same again we get that really cool look to it but what if we want to keep experimenting with this the first thing I need to do is actually just turn off the layers I'm not using and highlight my main layer again otherwise it'll actually reference the New Creations when I go to use image image again image to image again so what I'm going to do I'm going to go down to my Laura here and I'm going to add one in I'm going to say make selection come down I'm going to choose add more details I'm going to put the weight at 0.6 so it's not too overbearing and then see the difference in the results we get we've got this one here and this one and they do look I believe a bit more detailed if I go back to my first one I believe this one it looks the best and if I compare that to one of the other ones you'll see that the features are a little bit better and it is a little bit improved but if we come back to our original again I can actually continue to add more Laura's to this selection so I'm going to go down where it says add more details I'm going to click new to add in another and I can add in a different style altogether yet again I could even go for kids illustration which I don't know if that really applies we're going to go 0.6 again and let's see what it does with that image I do believe it's made a little bit more cartoony I believe we've lost some of that detail but we've gained a bit of a cartoon look overall I think it's a pretty cool effect so now we're going to check out control net which is kind of another way of using image damage but not it's not really the image to image function but the idea is you have different ways to let an image influence your Generations so the way it works is if I've got text to image here and I'm going to choose my model OKAY dream shape or version 8. I'm going to type in a prompt such as a robot looking at a techno-organic planet AKA realistic 3D render and what I want to do is come down to control net and open that one up and I've got a few different models there's different versions of canny depth is also open pose but what we're going to do is just have a look at a few of these very uh quickly first one we'll look at is depth if I go to depth version 11. what this does is it'll actually process the distance of certain objects in the image and the way it works if I go over here to process so under control net I've chosen depth I go to process I want to make sure I choose depth here and I'm going to select my image and apply and you'll notice on the right over here I'll input this as a layer so we can see it we have a rock here in the foreground which is white this person is into the image a little bit so they're not quite 100 white but still fairly bright in the background you can see how things are further away the darker they are the further away they are so what I'm essentially going to do I'm going to go back to my layers turn this one off and so I've got control net here I've got it ticked so now I can actually just generate and see what image I get so our images over here are done now these are square and they're actually a little bit squashed because I'm working on a more of a rectangular format but if I pop these ones in so you can see if I turn on our death map and then add this image you can see how it has mirrored the layout even here where the two sort of Peaks are in the mountains it's actually popped a couple of moons or planets in there so it has used that to create a sort of like a depth map of like okay these are the far away objects these are the closer objects even these rocks here are in pretty much the same position so if I actually make this semi-transparent you'll see exactly what I mean and again I can go through and I've got the same effect on all these other images so that gives you a little control over the layout of the image but there's actually more it's not just a depth map that you can use so I'm going to remove all this stuff and open up another image so now I'm going to look at another version called another model called canny under control net I have this drawing I made years ago of Batman which is probably a good example being a sketch doesn't have to be a sketch but this is going to be really a way to illustrate how this feature works so I'm basically just going to type in here cinematic render of Batman I'm going to keep it really simple to see what I come up with although we will put in there 3D render again I have dream shaper version 8 selected now I'll come down to control net where my control is and the weight is set to seven so we're going to experiment with this a bit because the weight will choose how much this influences the image so we'll go with 0.7 which is the default for now Kenny version 11 for governor process we can still choose a few different ones here to play with it's scribbles and different levels of it we'll stick with canny for now and I'll show you what it does I'm going to select the canvas so Ctrl a hit apply and I'll input that as a layer so you can see it's created a bit of an outline this is the basically the outline is going to work with for when it creates the image so I'll turn that layer off and I've got realistic 3D render here I'm actually going to change that because it's actually an outlier let me go I want to go with an anime style illustration manga so gonna create an illustration of this I'll just say bright colors which is not really a Batman thing but something to sort of play with this time let's generate and see what results it gives us so we've got our drawing here and check it out when I activate each of these four images you can see how the pose is identical even down to where the cape sits it's use the outline as a reference I can come up the bright pink is sort of that bright color it's been added and even the buildings you can see that they're a little bit flatter because we haven't actually drawn a background in but it's really adhered to my to my drawing pretty well and so what we're going to do now is actually turn all these off and experiment with the strength of the control net so I'm gonna go back over to control net and control and the weight is 0.7 I'm not expecting this to look great I'm going to pop it up to one which is a much higher strength to generate and see how much closer it has gone in a relation to my image okay so I turn this one on and you can see it's it's actually adhered a lot more closely and it actually looks better than I thought it would but I think it's just a little bit too close and therefore any of the sort of some of the features of my drawing that might not work are kind of not not really affecting it as well but you can see how useful this actually is but let's try it in the opposite direction let's give so this is actually running stable diffusion let's give it a little bit more control over the layout so instead of having one or 0.7 let's go 0.3 okay so once again we have my image and you can see the layout is slightly different however the image is much nicer smoother and sharper because we've given stable diffusion a little bit more sort of freedom to work with the image and not follow my outline to the T so we get a more adhesive image this is still fairly close so if I actually turn the center off you can see it's slightly different but still fairly close so if you're looking to smooth things out a little bit by bringing the strength down you give stable diffusion a little bit more sort of freedom to work so that is another thing that's worth pointing out when it comes to using control net so just to show you another sort of control net feature you can use I've got this guy here staying in front of a mountain and I want to capture it's a very basic layout and I want to capture that for my character which is a side Punk businessman a few other details to try and bring up the quality I'm going to come down and control that instead of canny version 10 I'm going to come down to segmentation version 11. good process I'm going to click segmentation and apply so you can see how it's used segmentation to create a bunch of colors if you have a quick look for some colors where the man is and like the sky and the mountains so it gives us an idea of the layout of the image so now I can use that to generate so I'm going to go generate my image or my cyberpunk businessman now check out how close the layout matches it's almost in the it's in the exact same place and one is facing the other way this way he's facing forward so we've got some pretty cool images here not to mention the image quality is really cool there are reflections in the middle I think it's really great but just the way it's you're able to take control of the layout with this uh is a it's a really cool feature so do you play around with controlling it there's actually a whole bunch of different models you can use so uh it's a pretty handy feature to play with and so this time we'll look at fill and how that works which is basically a function where it allows you to select certain areas of the image and just fill in areas of it without actually changing the whole image itself so I'm going to go into my layers first I'm going to create a blank layer above my image I found that gives me the best results and this bit in the sky here looks a bit funny so I'm just going to select above Maybe deselect some of the hair and none of my functions you're going to go across to fill and I've got a checkpoint here a model so this was created this image was created with controller earlier and I'll use dream shaper so I'm going to come down and there's a dream shaper fill model I can choose and what I can do is cyberpunk helicopter in the sky dark highly detailed 3D I've got my area I select my area again and so now I can fill this area with my prompt I just hit generate and you'll notice I have some helicopter images here I can double click on one and it pops that helicopter image in the sky within the image now it's really cool because I can also come down here sub pump sub pump cyborg robot matching the theme of the image highly detailed 3D cinematic and see what it pops in that little space once again I have some options and I'm going to go with say this bottom right one here and let's Pop My Robot into the scene which is pretty cool so you can see how this is a handy way to sort of change some of the areas of your images and improve certain features so Phil is a pretty powerful tool when you combine that with photos or air to sort of further tweak from there so next I'm going to show you is actually how you can manage your different models loras and sort of options through model Hub so you can see when we're creating images sort of text image here we got all these different checkpoints we can use and you see here there's add more models and click on that or go to the website but also the same applies for Laura's we've got these different lauras here that we can play with we can add more lures so you can actually change the order upload your own or explore other models on the website itself so if I go here and go add more model it will take me to my page if you're not logged into your account you will have to log in but I've got all these models here that I can use and I can even upload my own model but we'll get get to that in a minute what I want to do is I've got a filter here and I can go from things like checkpoints or Laura's or just a combination of both I can also choose which base model so if we're going to look at stable diffusion XL 1.0 we can actually filter it down to those two so we can actually look at those this doesn't change the plug-in this just filters on The View here but the changes you make here in the order or in the homology update will actually show in the plugin so there's also accesses such as public which are platform models community so members of the community have uploaded models so if I go to community you'll see there's actually no models here but uh it's personal I haven't uploaded any but there are all the public models here available for you to check out so you can actually just go through and filter the models that you want to look at so first thing I'm going to do is if I get choose checkpoints is if I got checkpoint what this checkpoint here for yes mix if I don't really want that here you can see first of all where it came from you can see the models the stable to Fusion 1.5 you can see a bit of information and some images that have been generated with that model I'm just going to head back and one thing I can do if I go up the top here I can edit or rearrange so if I go to edit here I can turn off certain models so if I know I'm never going to use this model or that model so let's turn off say ink Punk and yes mixing and on top of that I'm going to click save and then you can see they're disabled I'm also going to go to rearrange because what I want I want stable to Fusion XL to be at the top since that is sort of one of the more recent models dreamshaper for XL 1.0 and put that there and I click save so now if I head back into Photoshop and this time I'm going to exit Photoshop open it up again so I've reopened photoshopping you see here my checkpoint has defaulted to stablexl 1.0 and it's at the very top the next one dream shaver Excel 1.0 and I've changed the order and you notice that the ones we disabled aren't there either so that's how I can actually change the order of the models in model Hub and you can do the same for Laura like I said before I change from checkpoint to Laura and again we can rearrange edit and we can even delete models that are available to us but what if we want to add a model we can upload a model add a model if I click add I can add a link to one or I can upload so what we can do is go to a different website to find the model we want to download and then upload this is civet Ai and it's a place where you can go to download free models for stable to Fusion I'm going to click on models over here and find one that we can play with Juggernaut here looks like a pretty cool model I'm going to click on it this has what looks to be some really awesome images I'm going to download it so it's a big file I'm going to download this file keep in mind I want to keep this window open because I will need to reference the information when I upload so coming back to model Hub if I want to upload my model I click this button up here button says upload model and I pop the model details in here now keeping in mind all this stuff I will need to reference so I'm going to actually open up this page as well so I can find the information I need and I can start to reference this information across the checkpoint so I keep it on checkpoint it's a merge it's a checkpoint merge you see here it says base model so we choose base model you type in the description if you want to and I can make this uh sort of limited or open for General use for now I'm going to keep it private for the sake of the video because this is primarily to demonstrate how it works it is version 2.2 stable to Fusion 1.5 you see here SD 1.5 so stable Fusion 1.5 and this is an image to image and text to image so I'm going to put on that does support controller I'm actually not sure uh I couldn't find that information I'm just going to say no for now there are version change notes so you can actually find some of those down here and why the author I am most definitely not the author and the source link which actually popped up here I will pop down there but obviously you want to describe the model a bit more better a little bit more accurately so this is like a photo realism now the scissor more Lamb's already tagging so I'm just going to put Juggernaut private in here click next and now I'm on the next page I can upload my files so I'm going to drag and drop that in there is it big Falls it's going to take a little while to upload let's say model size full floating point if I come back over here under the information for safe tensor you can see it's a floating point of fp16 and it's actually pruned so I need to come back to pruned and fp16 it says the upload is successful I click next we do need to give the model an image so I'll save this image it does need to be a square image so I'm going to have to crop this image quickly I drag my image in and also asks for me to upload some images as a gallery so I will need to go back and just download a few of these images drag those over reorganize and publish and now Juggernaut private is at attached to my account and I can use that model in Photoshop so now I'm in Photoshop and you can see Juggernaut private has been added to my checkpoint but if I head back to civit AI I can actually click on this image and even test out the exact prompt in Photoshop [Music] I can even check out the details like the steps so we've got 30 steps on here so pumped the steps up to 30. select my canvas and I can generate an image using juggernaut and we've got some pretty cool images here I'm going to open this one up and check that out a really really cool high quality image I'm going to click I'm going to upscale and check out the quality of this particular model after I've upscaled it using Juggernaut also with the L with the alcade image Creator very cool very handy plugin to use in Photoshop so try that out but checking out the link in the description below otherwise if you like the video please consider giving it a like otherwise hope to see you again next time have a great day and thank you for watching

Channel: Wade McMaster

Views: 6,879

Keywords: Ai Art, alkaid art plugin, stable diffusion, Photoshop Plugin, ai, stable diffusion photoshop plugin, alkaid.art, stable diffusion tutorial, stable diffusion art, free stable diffusion photoshop plugin, stable diffusion photoshop, photoshop stable diffusion, photoshop ai

Id: tfXkni5wN0E

Length: 24min 18sec (1458 seconds)

Published: Mon Aug 21 2023