so hey guys, welcome back once again. In today's video, I'm just gonna talk about one of the interesting tricks that I was working on, stable diffusion with the help of ControlNet. And yes, you can definitely see noise offset. Laura, it's an amazing trick that I found. One of the users on the Reddit posted this trick. It's literally amazing that you can go and add details into your image. First of all, let me tell you, in order to work with this trick, you need to go and download this noise offset. Laura. You can go and download it from Civit AI website. And guys, after downloading this Laura, let me first of all tell you where to paste it. Just go on to your stable diffusion directory, go on to your models. As I'm already on stable diffusion directory, go on to the models, go on to the Laura, and just paste it over here. See, I have already pasted Epic Noise Offset Safe Tensor. So you can paste it over here. And with the help of that noise offset. Laura, you can go and experiment with this trick. Let me tell

you yes definitely it's amazing with the help of control net extension. As you can see, this was the image that I was using on control net extension, and I was using depth layer as and depth model over here. Yes, definitely it's an amazing trick that you can go and add details to your image if you have generated any image and if you want to add details to it. As you can see over here, I was just experimenting with like different weights of Laura with like epic noise offset zero weight with one weight and 1.5 width, and you can see the difference over here. Pretty amazing, the results are pretty much amazing. You can go and experiment with like different weights as well. You can go about 2.0 as well and about 2 weights as well. So guys, first of all, let me overall show you how to work with this trick. The first thing that you need to do is that just generate any of the image on text to image section. Just upload that image onto your control net model. You can definitely see this was

the original image. And these are the results with noise offset. And the next thing that you need to do is that you can select any preprocessor and model. The guy who posted this trick on Reddit used depth layers model, and yes, definitely the results with depth layers are amazing. So I was experimenting with this model only. The next and the most important thing is that just upload Laura prompt over here. If you have downloaded that Laura, then you can come on this section and click onto Lora. You will find this Lora file over here. If you just click onto disk, it will automatically update your Laura prompt over here. And yes, in order to show you what difference does it make, I was just using prompt Sr. You can definitely see with like 0.1 and 1.5. So this was the thing that I was using in order to show you the difference. And now let me show you how does it work. So guys, the first thing we will do is that we will just generate any of the image without using Laura. So we will delete

this Laura prompt. Let's delete this and let's turn off this. First of all, see I have turned off this and I have turned off like script as well. So let's generate and let's see what result do we get. So guys, this is the result that I got without Laura noise offset. And the next thing we will do is that we will just upload Laura over here. Just click on to this section. And after going on to Laura, just click on to this. And then after uploading the Laura prompt, just come on to bottom. And in control net model, let's upload the original image. So let's upload this image, original image. And then let's turn on control net model as well. And one more thing that I was doing was that I was just using XYZ plot with prompt Sr. You can definitely see I was using with like 0, 1, 1.5. So that's the thing that I was using. And the next thing what I will do is that I will just add some of the details on the prompt as

well, and now let's see what result do we get. So, guys, as you can see, that is the result that I got with like totally different pitch zero weight 1 and 1.5. If I show you the original image as well, this was the original image. And yes, there are a lot of changes as well. Just compared to like this part of the body, there is part missing over here. And on one weight as well, definitely there are a lot of changes going on. This part of the armor is missing as well. And if you see 1.5, definitely it's very close to this image. But you can definitely see the image just got a little darker and something more realistic lighting as well. So that's how, guys, you can go and experiment with like epic noise offset if you want to change the details of the image and get something like a realistic style. Then yes, guys, you can go and experiment with big noise offset and control net model as well. Now what we will do is that we will try to add some more details to The Prompt as well.

we will see what difference we will get. So let's change the prompt and let's see what result do we get. So guys, this time I went on with a different team and definitely you can see this time it's a totally different theme. First of all, let me show you. I just added snowfall at night, snowfall City street, and here as well young woman leaning against the wall in snow. So that's what I changed and this is the result that I got. And yes, if you see these guys, definitely amazing result with like control net and epic noise offset. Laura see the details going on. If you see on zero weight, yes, snow going on over here. And if you see on one weight as well, definitely amazing reason. I just loved each and every of the image. And if you see 1.5, it's a night image, but yes, there is a snowfall theme going on over here. So that's how you can go and add some details to your image. And if I show you the original image, see this is the original image and look at

this image is totally different and totally detailed image as well. So yes, it's a totally amazing trick that you can go and experiment with Control Net model and like Laura Epic Noise Offset. Now guys, let's jump on to like different image as well and let's see how does it work with like different theme as well. So guys, this time I went on with like different style and you can see this was the result that I got. First of all, let me show you this was the original image that I used and after updating with like Laura Epic Noise Offset, this was the result that I got. Let me open the original image. So this is the original image and with zero weight, you can definitely see yes, the result is pretty amazing. There are a lot of changes going on with this image as well. Yes, definitely we got a different image and if you compare it with one, you can definitely see it got darker and with 1.5, it got totally dark. So I think so my prompt was not good with this. So we will

try different prompts as well with this image. Only and let's see what result do we get with different prompts as well. So guys, as you can see, this is the result that I got. What I did this time was that I just deleted every night and dark word from the prompt. And after that, I generated this. And this is the result that I got. If I show you the original result, see, this is the original image. And if I compare it with like zero weight, you can definitely see yes changes going on as well. And with one weight, definitely it's darker, but yes, the results are pretty amazing. And this time on 1.5 as well, you can definitely see amazing result. This time we got amazing result on 1.5 as well. So what I did was that I just deleted night and dark word from the prompt. And this was the result compared to the early result. Yes, definitely this result is even great. That's how you can go and experiment with epic noise offset and control net so you can get some

amazing result. You can change the details of the image, and you can get some amazing result as well. Your prompt also matters in this. If you go on changing your prompt, then yes, definitely, you will get some different result as well. So, guys, isn't it amazing that you can go and add some details onto your images? And yes, definitely, it's an amazing trick to go and experiment with. So, that's it for today, guys. Yes, I will show you some more images that I generated using this trick, and then I will end the video as well. So, that's it for today, and I will be back with more and more videos tomorrow. Till then, have a good day. Bye, guys. Love you all, guys. Bye, guys. [Music]

Stable Diffusion Controlnet + Offset Noise + Lora Trick Explained | ControlNet Extension Tricks

Stable Diffusion Controlnet + Offset Noise + Lora Trick Explained | ControlNet Extension Tricks

스크립트