Midjourney Panning + Dreamshaper Version 7

Key Takeaways

mid-journey panning and dream shaper version 7. hello my friends how are you doing today I am team Nostromo wow look at that pretty cool right okay so here we have the new panning function of mid-journey look at this amazing cyberpunk City pen and it's going and it's still going and it is still going because you can go as wide as you want with this so here we have the full width of my image and it is amazing how great this is adding parts to the city that really makes sense and really build up the scene now this is very easy to use the only thing you have to do here is to create an image and then you want to upscale the image by clicking on one of the U buttons and then afterwards when this is upscaled below the upscale you will have these arrows that are pointing left right up and down and when you choose one of those this is of course extending your image in that direction now after you've chosen one of these directions you of course can go as often as you want but the next time you will only be presented by the same choice you chose last time so in my case it is either left or right but no more up or down if this option does not show up for you you want to go on to mid-journey and type slash settings and then when the settings window is open you want to activate the remix mode Let's go to dream shaper version 7. so here we have the Civic AI page as you can see there is a new version here and as always I would suggest that you look at all the settings all the text down here from the description and the sample image of course I made a lot of sample renders for you that we are going to look at here you can see things that have been improved and you can also find out information on how to use this better for example especially this part here that we are going to look at in a second but first but before we go into that let's talk about what is the special thing about dream shaper 7. now this is a model that can do photorealism but at the same time also digital art and at the same time even anime images so when you click through these images here you will see a lot of different styles that are all executed in a pretty well style now because I know the realistic style the best this is what I tested the most and I will also show you some of my prompts but of course what you want to do here is when you check out the model is to click on one of these images you can see how the Beautiful the quality is it is very good with hair it's really very very nice but also here The Shining Armor is also very very good but more importantly on the right side you have either prompt that was used you have the negative prompt you can see here is two negative prompt Parts here that are both negative and betting so we have bad dream and unrealistic dream and then below that we have the sampler of course the model the CFG scale but also we have the clip Skip and as you can see here this model is using clip skip 2. that is very important to know about the model so you're actually using the right clip skip version so let's have a look first at the negative prompt here what are these negative embeddings actually this is from the same person so when you go back to the cvd iPad and you scroll down here you can see this text that I've highlighted for you you click here on BET dream and you get to this page these are both negative embeddings you can see here we have bad dream version one and we have unrealistic dream version one both of them are specifically made for this dream shaper model so that you can improve the image quality and this is a classic method to improve and get a better output from a model that still has some problems but they can be reigned in with these negative embeddings so where does all of this stuff go so with the model file you want to download this into your automatic 1111 folder in there into the models folder and in there into the stable diffusion folder for the negative embeddings on the other hand what you want to do here is to download this into your automatic 1111 folder but then not into the models folder but instead into the embeddings folder after all of this is done you want to start up your automatic 11 11 you want to choose the model up here and then for the negative embeddings you have on the right side here this pink button you want to click on that and when you go to your textual inversion tab you can see that here you have your bad dream negative active embedding and the unrealistic dream negative embedding the only thing you need to do to add this to your negative prompt is to click inside of the negative prompt field and then click on the negative embedding you want to use so that it is added to your negative prompts very easy to do now the next big question I get asked a lot is where do you get the clip skip Choice up here and maybe also the sdvae quick Chooser here so for that we're going to click on settings then on the left side you want to find the user interface and when you have found that you want to scroll down to the quick setting list in here you can see I have multiple choices already set for me so what you need here is sdvae and also clip stop at last layers now you don't have to type the full thing because when you click inside of that field you get a long list and you can simply choose it from that list when you've chosen both of them you want to save your set things and then reload the UI so let's have a look at some examples that I prepared for you with the prompt and the negative prompt so here this is the same example as we have seen on the page for the model and again it is very beautiful it has a very nice quality of course this is upscale two times in image to image to give a better quality this is not the 512 by 768 resolution this is the 1024 by 1536 resolution as are all of the other images as suggested by the page I'm using in the negative prompt both bad dream and unrealistic dream with a weight of 1.3 and then I'm using the exact same prompt as on the website I think the only thing that I added here is highly detailed skin with a weight of 1.2 to give more detail to the face over here now here we have a close-up of the face looks really beautiful one thing this model is very good with photorealistic images is the hair where it just looks very nice you see every single hair strand but one thing that also stands out here is that often in the upscaling this has a little bit of a problem to get the image completely sharp and crisp I will show you a trick on how to get much better quality at the end of the video here we have the chainmail she is wearing the detail is beautiful the quality of the reflection is very nice and I'm surprised to see here that we have a reflection of a city in this chain mail so that is really amazing and makes it look very realistic here we have another example it's just a slightly change of the prompt in this case you can also see that I'm using the Laura funic stress version 2 which I have also linked in the video information below and I have the add detail detail tweaker where I have a video about about this here this is a very useful tool to get better details in your images and make them even more amazing again here we have a close up and as you can see the face is beautiful the hair is amazing and here we have the chain mail that she's wearing in this case and this is more decorated more ornamented because of the phonic stress Laura and I really like that kind of dreamy fantasy style here we have another example this one I have created with the one button prompt extension check out my video here this extension is really beautiful I'm using it a lot since I've installed it because this suggests such amazing prompts and actually most of the images I've created for this example are done by the one button prompt here we have another close up now the quality of the face is beautiful but again it isn't completely crisp and sharp and the same goes for the clothing where it looks very good when you have a smaller view of the image but if you go in and do a little bit of pixel peaking you see that it's kind of blurry and it looks like just a bigger version of the low resolution image so let's talk about how to improve this and here I have another of the one button prompt prompts this is a very beautiful image and in this case I've not touched a prompt at all now when we look at the close-up you can see that the details are pretty much pixelated the face is pretty nice this is because of the face restore but now I change the prompt and what I would advise you with this model to do is to write at least full sharp with double round brackets around it so it has a high impact on your image and then maybe also highly detailed and to make it even stronger in this case I even use the Laura at detail which is the detail tweaker Laura that I talked about before so when we now look at the image you can see that we have a lot more detail in the dress everything is a lot sharper and just looks more beautiful and here's a direct comparison where you can see how much this has actually added to the quality of the image so for this model is actually very effective to play around with different terms of adding more details and adding higher resolution to the output of the image of course it's also important to notice that I zoom into the image here pretty hard you wouldn't usually have the image zoomed in that far this is why you see the pixel so strongly but when we zoom out to the size you would usually see that or even smaller you can see that the quality is just it's just very crisp very nice I like the shine of the silk dress everything about that is really amazing so here we have some additional images from different Artistic Styles that are created with this model and as you can see all of them are pretty amazing and really awesome to look at let me know in the comments which model I should make a video about next thanks for watching and see you soon bye oh you're still here so uh This is the End screen there's other stuff you can watch like this or that's really cool and yeah I hope I see you soon leave a like if you haven't yet and well um yeah