0:00 [Music] now let's go for another thing we can do with the same technique to follow my workshop you can download my workflow on open art you see here there's a download button and above that you have a green button that says Lounge workflow where you can actually run my workflow in the cloud for free when you click on that button it takes a little while and then it's opening up this window as you can see here everything I've built for you is in here here you don't have to download or install anything this is completely prepared for you again we're going to the group we want to use here and we rightclick and then set group
0:45 notes to always so that everything is active in here and as you can already see because now you are experienced with these workflows We have basically the same build we have our checkpoint the two prompts in here we have here our our image that we are loading and then we are going into the case sampler and we're creating an image over here now what is the difference in this case well actually when we zoom in here you can see I load an image has a very long name here this is an image that I have created with mid Journey you can see that it looks pretty cool very nice robot here standing in a city but you can actually use any kind of image that you have created you find online any
1:31 kind of information even sketches Doodles whatever you want so let's scroll out here again and you can see here I'm using a different note this is called image resize in this case what I'm doing here is I'm resizing this to a width of 768 and a height of 512 I'm doing an interpolation here and then keep proportions in this case so that we have the same proportion so now you can see here again I'm doing some probing over here by sending this to a preview so I actually see what I get from that and then of course we are going to send this image into our vae en code and then it's going to go into our
2:17 K sampler as a latent input let's have a quick look here at our prompt I'm writing here giant battlemc City in background Masterpiece best quality so now let's click here on the Q prompt and see what happens so you can see here the image has been resized the rendering is going on and out comes this image here so that's already pretty cool you can see we get here a different robot but it's still similar to our input image of course the reason for that is that in this case we have set our D noise to only 0.3 and that means it takes a lot of the original image into consideration
3:03 but at the same time when we look here we still get a lot of variation from that so that means we can have the same artistic style and colors in here and the same composition but we can create variations from that let's set the Q prompt again and see what we are going to get next and there it is it's already finished you can see it has changed a little bit from our robot of course you can also set the D noise higher if you want to to get even more variation in here so here we have another variation that is also based on the same D noise I'm going again to our image resize now I'm setting this to 512 and now I'm setting this part to 768 and in this case we're also going to
3:49 turn off key proportion so click here so this says false and now we are getting an image that is vertical instead of horizontal so you can see here from our image this this has actually been squished into that space let's set the D noise value instead of 0.3 to 0.5 and as you can see the output has considerably changed now we have a battle Mac that is actually a lot more humanoid has a lot of like a backpack or a weapon on there we have nice details down on the street and the overall composition with the buildings is different but at the same time that is really beautiful but I also want to show
4:35 you an alternative workflow that is basically using the same technique but it adds an additional level of control and depth to it now this workflow is actually kind of a magical machine to get creative outputs from a single image input you don't even need to write a prompt so let's inspect this workflow closer what I have here again is my checkpoint that is loading but now you can see that over here for my prompt I actually have an empty box in here and then here I have my negative prompt now why is this box empty the reason for that is because I converted the text box which is called a videt inside of confu
5:22 I notes into an input the way you can do that is by right clicking on the note and then looking for the input you want to create you can see down here it says convert text to input if I click on that the output is going to be a note like this where we don't have a text field but instead we have here a text input now here's the reason why we need a text input here we are again loading the image and here I have a note that is called the V D14 tagger and this will actually analyze the image it will look at it and it will turn it into a text prompt and you can actually see the text prompt down here as a preview it says
6:07 outdoor multiple boys night glowing robot ground vehicle building scenery Mecca motor vehicle science fiction City car police lamp post lights search light so there's a lot of information in here about what is found in the image and this is created automatically for you now personally I want to have a little bit more control over how the output is going to look so I add another note here this is called a string function now what this is doing is it can combine a prompt that is coming in with other prompts that you're writing here usually you have here three text fields and a preview field but as you can see up here
6:54 I have converted this area for text a into an input again the way to do this is right click on the note look down here to convert and look for the input you want to create it says here text B and text C because I have already converted text a into an input so as you can see what's happening here is this output from the wd14 tagger is becoming our text a in here and then followed by text B in text B I'm writing style of concept art masterpiece digital painting facing viewer and down here in the preview you can actually see that this part that I'm suggesting is at the end
7:40 of our prompt so all of that information is now going into our positive prompt and then into the case sampler again as you can see also with the image I'm using the image resize to squish that together in a vertical format and then vae encode it into a latent image both the text that is Created from the image as well as the image itself is going into our K sampler here as information for the AI to render and then of course it's simply rendered via VA decode into the actual image and now look at this amazing output we have here a humanoid Mecca that is standing in a city street a really big Mecca we even have your
8:28 person walking down here on the ground and every time of course I click here on Q prompt this is creating a different variation of this image here so here we have another variation with the Mecca and here we have yet another variation of our Mecca again one thing to also point out here is that in this case because I want to give the AI a lot of freedom is that I'm using D noise of 0.78 this is also possible in that case because we also have the text input to give us additional information of what we want to have in our image now at the beginning of explaining this workflow I called it magical and here's the reason why I did that let's go to our image
9:13 loader click here and then I'm loading this image here where we have a guy with a brown jacket standing in a landscape I'm not going to change anything not the prompt not the settings nothing I'm only loading this image and then clicking on Q prompt and look at that we have as an output a guy in a brown jacket standing in a landscape but it is a different image we have created a variation of this in a very similar Style with the same elements in there how amazing is that thanks for watching this part of my workshop there's a lot more you can do with the latent input and I invite you to experiment a lot more with that also watch out for my other Workshop videos
9:59 oh you're still here so uh this is the end screen there's other stuff you can watch like this or that really cool and yeah I hope I see you soon uh leave a like if you haven't yet and well um yeah
