Hello and welcome to this video in which I would like to exchange a little bit of life time for knowledge again. Last week I was unfortunately flat with Corona for the first time for me. I was able to avoid all the years. Now it has caught me after all. And while I was gnawing in my fever juice in front of me and my circulatory system was apparently on vacation in Gran Canaria or something, two thoughts shot through my head. The first was, does this elephant always live in my bedroom? And the second thought was, why the hell do I keep re-uploading the same prompt parts over and over again? So what I like to use, for example, is, I'll build something together here. Schwupp Schwupp And where are you? There. I often do it so that I use things like Masterpiece, Absurd, Res, Intricate, something like that in the Art. 8K, UHD, HDR, High Detail. So style descriptions, or there are never even styles, they are actually more like keywords. Styles also work with what I'm about to show you, but that's what it's all about. Then I thought to myself, why the hell do I keep typing that in again and again, when I have such a pre-made set of words and prompts in my head, which I actually always want to pass on to the pictures in addition. And I've been thinking about it a bit and it's actually totally simple. And that's actually good for me to start with such a simple topic again in such a video recording. Because what I've been thinking about is the following. So let's leave that here. We can let a few things go through right away. I'll put it on save so I can do thumbnails too. I can do that, I'll call it prompting. Then I can use the photos better for thumbnails and stuff later. Well, what I've always liked to show you before, that's a bit of a sleep in the last few videos. But I like to store such prompts here. So it doesn't matter which loader or whatever you use. If it's not the normal text and code prompts from the ComfyUI, then I like to store them. And I like to say convert positive to input. So we can do that. We know that because we can put texts on it and then enter our prompt here. What I was thinking about now was the following. I wanted to store and save these prompts that I just showed you or these parts of prompts. And for that I created a folder somewhere on my hard drive. I called it Prompt Parts. Here I have differences in positives and negatives. And I created a standard.txt file here. Let's take a look at that. And that's basically what I just said. Absorbed Rest Masterpiece, High Detail, Intricate, blah, blah, blah, blah. But here also Cinematic Lighting and Depth of Field with a strength of 0.8. These are just such collections of words that I use again and again. And I would like to use that here. And that's totally simple in principle. We'll leave the text nodes here. We'll need them in a moment. You can also take any text node that you want to do. But it basically comes down to a node from the Was Node Suite. The link is in the description below. There I have listed all installed custom nodes. I've always installed a few more in between. But in the video description are the ones I use for the videos. Because I always test a little bit here for myself and so on. That's why it can always be that you see nodes here that I haven't linked below. But I haven't released them yet. In the Was Node Suite there is an I-O folder for input and output. So input, output. And here is a load text file. And that's actually the magic. We can now combine our prompt that we want to do with a text file. And for that we need a concat node. So to concatenate texts, to put them together. There are various nodes for this. In earlier videos, I always used the Tiny Terra Nodes concat for this. You have to store them. Then here are the texts, text 1 and 2. We only need two at the moment. But I don't think this node is so nice for it. I can just show you why. So. Because if we now hang a text debug in between. And I have to say something about that too. Do we get that right away? Ttntextdebug. Exactly, that's also from the Tiny Terra Nodes. The text debug node. I'll turn them both off once. So. I always blamed that this node would be broken. Because you can normally see the text displayed down here. And I always had to export it to the console in the last videos. But a viewer posted a comment. And I am very, very grateful for that. The node is not broken. That was an interaction with animate.div. I had installed that with me. To play around with it a bit and experiment. And that's just a side effect of animate.div. This node doesn't get the javascript events right anymore. And that's why it doesn't show the text. So everything is correct with the node. It was just a side effect. That you heard it too. Because I really like the text debug node. So. Good. We see now that we have downloaded text 1 and 2. And if I write here now. This is text 1. And I let the whole thing run. Then we see here that this concrete node. Always our delimiter. So the comma is still hanging on here. We now have two blank texts. So that's what we just entered. This is text 1. Comma blank text. I can't even mark here. Comma blank text. And that's why I don't like this node so much. What I prefer to take at this point is. We look for concat. And. These are not the underline o. They come from the quality of life suite. We take the text concatenate. And that comes from the was no suite. And we are a bit. Equal here. So if you connect them. And push this over here. So. Then we see. I have to put the delimiter first. I say comma blank. Our trend sign for the different texts here. But we can already see here. That the node works cleaner. She does a comma. That is also justified at this point. Because. Text A and text B are. Obligatory entries. So we need at least one comma. If you send an empty text down here. Then you don't need a concatenate. That's why. That's okay. And if we are here again. For example, hang this on text C again. Then we see. That he's here again. Blah blah blah. Written again. The first comma is, as I said. The second not. But we also see here. He doesn't make a comma behind it. Because text D is not filled. And that. I think that's pretty good. So. But we just saw the text file. What I have here. Standard.txt. And everything we have to do here now. Is once right click. Copy as path. We click in here. Control V. Delete the quotation marks. Front and back. Press enter. And that's it. Because now we can see. That we now wonderfully. Once what we want to achieve. Or our motif. Have standing here at the beginning. And what we are very often. Very often. Enter again and again. Although it is always the same. We have now. Concatenated behind it. We can of course. Different. For example again. A low text file behind it. If it is necessary. To describe different exposures. Or something else. You can do that. In the file name. I could. Say again. I want a description. Cyberpunk City have. I put that behind here. I also say here. Copy as path. Control A. Control V. Delete the quotation marks. So. Hang that in text C. Then say queue prompt once. And now we see here too. That we have the whole bunch of words. Now get delivered automatically here. Without us. I delete it again. Now we have our description again. And I just give up here. Just once. Beautiful. Cyberpunk. Woman. So. Before we let go. Here we still have our. Sampler Setup. We can now simply here from the text. Debug node. And for videos. I always like predefined negatives. The one or the other. Viewer already determines. Here we do that too. I take here again. A text debug node. Just so we see what happens. I take one. Load text file node. Also here for the negatives. So. Go to my folder. Negatives. Say here for YouTube. Copy as path. Click here. Control A. Control V. Remove quotation marks. And the whole thing. Can start rattling. I just set the model right here. I just didn't do that. Oh no. We don't need VAE here. We take the epic realism now. 512 by 512. That fits. And. And. Why does he do nothing here? Haha. What's going on here? Does the console say something to us? God prompt? Ehh. What do we have here? Was that from before, right? Look, I can now too. I can show us the console. Is that not madness? Okay. Something seems to be smeared here. Something seems to be very badly smeared here. Should I restart Comfy? Path blah blah blah. Specified cannot be found. Oh, that was just. Somehow I entered something wrong here. We let that run. That is. Also correct. Last time there were even comments again that I should leave errors in it. No, apparently my ComfyUI doesn't do anything anymore. That's nice too. Really nice. Let's do it here. Restart it once. The console we can hide. I'll just cut. So, I have a new Comfy instance running here. Let's see if it does it now. Ah. Interesting. Exactly. Now it ran. No idea what he just had there. Here are my negatives, NSF, W, text and watermark. And now we can go here and let us create the pictures we are funny. We can now also pretty easy just say. I have here, you may have already seen it. In the Prompt Parts Positives I also have two here Anime presets prepared. I also say for Anime 1 copy as path. Say that now. I just want to have that. But we also switch to an Anime model and now we change the VAE. Let's take the Grapefruit. And we get here pictures in this style. Pretty clear. Now we can also pretty fast just change. I make Anime 2 as path copy. Add that. And that's it. And now we get a completely different style. Simply because we our predefined styles, which you experiment a little bit. Find your own style somehow in Prompting. Be it in the positives or negatives. Which you then find. And yes, we can just save anywhere and then reload again and again. Maybe you should then a little bit the file names, not Anime 1, Anime 2 name and so. That's clear, but also for me this is still pretty new here at this point. Therefore there will be a little bit of collect. On predefined Promptings between which you then can change again very quickly. You can of course still work on the motif up here. With pink color Tear. So that goes of course also on. As I said down here or here in these stylistic descriptions here, mainly only styles like Brush Strokes, Ultra Detail, Digital Painting and something. And describe the motif up here. Text C and Text D are optional with this node. That means you can even add you could add more text nodes, if there is a need. And I think with four entrances you can get through here quite well. So that you don't have to rebuild something like this over and over again, I'll show you something. Pay attention. So, let's do it first. Let's make it a little nice. I'll say now I can actually do it that way. I've tested it before. These are our positives. These are our negatives. So, what we can do now in the ComfyUI and I think that's actually not that long yet. I think the update came in October. We can now mark all these nodes and say now I have to look for myself where it is. Ah, yes, of course not. Click on a node on the right, but here somewhere and then we can say add group for selected nodes. Now we've already created a group here where our whole construct is in it. And what we can do now is we can select the whole thing and then say with a right click somewhere save selected as template. For this we can enter a name. Let's say load promptings. Did I write that right? Load promptings, yes. Exactly. And we'll kill them all now. Away with it. The group too. Edit group. Remove. And what we only have to do now is we have now built our setup here. We have positive and negative stored. And now we can just right-click at this point go to node templates and say load promptings. Ah, the group is gone. Too bad. I thought it would save with. Apparently they haven't built that in yet. Okay, I learned something myself. You still know that you just right-click again here to get the group around. This is a new feature of ComfyUI. Apparently. Both are new features. To get the group around and also to save these templates. Although I mean I've seen that before. But it was in the October news nodes. So in the ComfyUI blog. Nevertheless. You see how it works here. We have the whole template just loaded again. We can also take any many of them again. Then we don't need to put together again and again. To build together. But you just load the whole thing as a template. You just have to connect positive and negative again. And then you can with any sampler setup or workflow that you want to stick to it work again. I find it very practical, very convenient. It's not a lot of effort just to copy the thread in here. Especially if you now of course also assume that you always expand the prompts a bit. Then it saves it into the file. That's a good keyword, by the way. I have to tell you something. But it's pretty good. So. What I wanted to say now is work with the prompts and save it in files. If we go there now and say, do I have the text editor open? No, I don't. I'll take my standard again. A standard description. I don't need to have it open twice. So. And we'll load it in now. I say copy thread. So. Pack it back into positive. Standard text. We'll take another model again. Let's take a quick look. Aha. What's going on? Why is my ComfyUI always smearing? Ah. F5 did it. There seems to be something not quite kosher to be. Some node seems to be moping a bit. Press F5 and it works again. It will probably be done with the next update. I pulled updates this morning. Maybe something broke again. And someone has to fix it. We see here that our text file was loaded. But if I go there and say test and CTRL-S to save the text file and press queue prompt it won't appear here. And that's unfortunately the fault of ComfyUI. Because ComfyUI doesn't load the nodes when nothing has changed. And our path is the same here. To trigger that you have to delete the T or something. We say run. We add the T again. And then we see that our prompting has been updated. So we can already file within the text file to the prompts. But we have to make a change down here. So this node has been refreshed. I haven't found a way to force the execution of a node. That would be pretty cool if you could say in ComfyUI don't cache this node but always run it. But I think the profit you get from this is better than having to refresh the file. You can also go and say I want to build a prompt like this. So let's turn it off with CTRL-M. Let's take what we got from the text file. Put it in there. And we can also expand the prompting here and write something in there. And when we're satisfied with it, we copy the text and throw it into the text file for further use. That works too. But my intention was to think about why I always have to refresh the file if there are terms I always have to use. I can also show you Cyberpunk City. I tried it with ROM. Here I say remove the quotation marks. We can see down here I have some descriptions of Cyberpunk City with neon lighting. I think Rainy was also in there. Rainy, reflections, high tech, sci-fi, augmentations, gear joints. Such stylistic descriptions can be used again and again. I hope it helps you at this point. I find it very practical and very convenient to be able to compile things from the text area and then expand them to a custom motif like up here. So I would like to have the motif, but with style settings that I represent my style. Without having to type a lot, just copy over the path, put the whole thing together and then save the construct here as a template where we can just say we want here, no matter where at the front, then clamp the two things to it. By the way, this works with the templates for every node setup. You can also use it for upscalings or whatever you use every now and then. Right click somewhere, add group. You don't need to, but I find it very pleasant. Yes, have fun playing around yourself, experimenting, finding out, tinkering with text files with prompts, combining these so that at some point you have a library with which you can express your style really well. Take care until the next video. Ciao.
