Recreate ANY Picture in Midjourney

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
have you ever seen a picture and wondered how you could recreate it in mid-journey well there's one thing you should try like here this screenshot from Star Wars John boyega's character Finn holding a lightsaber in the climactic battle at the end of the movie it's a pretty cool looking picture but how can we make this in mid-journey we are going to use the describe feature in your prompt box you're going to hit forward slash describe now as long as you have the picture saved on your computer you can select it and mid-journey will do its best to describe what it sees it will give you four prompts a man holding a Blue Star Wars lightsaber in the woods a Star Wars villain is aiming a blue saber in the snow the mysterious boy who's waiting for his lightsaber in the snow A man is holding a lightsaber in the dark etc etc and what it allows you to do is to just click on these numbers at the bottom that correspond to those prompts you can hit submit and mid Journey will run these prompts for you but let's take a look at the results these aren't exactly what we wanted I will admit that this particular prompt turned out pretty good these not so great great nor were these what we were looking for so what's the point of using the describe feature if it can't actually help you recreate what you want well there's one big thing you can try and that involves using the original picture as an image prompt you're going to hit the plus icon down here beside the prompt box upload a file you're going to upload the picture to Discord now you'll be able to click on it it expands this is where we're going to right click and hit copy image address now we can include this in our prompts we're going to go back up to our describe prompts click on them then hit Ctrl V or command V and paste the image address into the prompt now when we submit mid-journey is going to generate with our reference picture as the foundation and look what it comes up with like whoa these are rich tea cool pretty cool I don't even mind these but the prompt really isn't that good either then you have these the mysterious boy waiting for his lightsaber in the snow like these are really cool same with these like number two is a great picture there is actually one more thing you can do if you want to make your Generations even closer to the reference image you know what let's try it with all of them and see how it goes you're going to copy the image address then at the end of your prompt you're going to assign a weight to the reference image and you can do this by hitting dash dash IW and then a number between 0.5 and 2. now this number represents the relationship between the picture and the words in your prompt if we add a weight of two that means our reference picture is twice as important as any words we have in the prompt whereas if you made it 0.5 the words would be more important than the picture so if you want to recreate a picture more closely you're going to bump up the image weight whoa these are pretty amazing number three is great oh man look how powerful these are the hands aren't perfect but that'll be fixed in like a couple of months and to be honest they're not even that bad number one's a cool pick I really like number three and number four because it kind of looks like he's wearing Rey's outfit like isn't that what she had let me know that vest with those sleeves underneath cool picture and then you have these that's pretty cool like number two a lot and this works even in the most extreme cases like here a picture of a bow and arrow if you've ever tried to make a bow and arrow in mid-journey you'll know just how bad of a job it does it could just be a data set issue where mid-journey wasn't trained on bows and arrows but either way it can be a pretty frustrating problem to run into because you're like well mid-journey can do everything why can't it do this I got this picture off of Google by the way this is not made by mid-journey so we're going to run it through the describe feature and it comes up with these bow and arrow a man shooting a bow and arrow a young man practicing archery describe knows it's a bow and arrow but watch when I run these prompts it's not going to come up with anything good thank you look how bad these are number three is hilarious number one is like okay and they these are not the best actually these are surprisingly good like number one would people are going nuts for that same with number three that's actually quite impressive but take a look at these when we include the reference image in our prompt I think number two is absolutely fantastic granted the others didn't work out that great but mid-journey is a slot machine best of luck to you and for full disclosure I will say this didn't work out perfectly I didn't think these turned out that great these are in fact pretty awful and these are just the worst but if we took the one that worked out really well a man in Black is shooting an arrow and we add the image prompt and then bump up the weight of the image let's see what it comes up with oh these are pretty good like number one's awesome number two is okay then the rest you know not that great and keep in mind there are also some parameters you can add to these describe prompts like the stylized value and the chaos value if you're looking for just the perfect picture you might want to include a higher chaos value this will expand the variety of your grid meaning each of the pictures will look quite different from each other so instead of getting four half decent pictures you might get three weird ones and one really really good one and if you're going to be adding chaos to your prompts I highly suggest you bump up the stylized value as well this will give mid-journey more creative freedom and probably result in prettier pictures here it is with the chaos 8 stylized 400 look how good number three is like holy cow that is perfect and by God it is not easy to get bows and arrows in mid-journey leave a like on this video so we can share it with more people I hope you're doing well take care and I'll see you next time peace
Info
Channel: Future Tech Pilot
Views: 40,529
Rating: undefined out of 5
Keywords: midjourney ai, midjourney, ai art, midjourney tutorial, midjourney tutorial for beginners, midjourney tutorial prompts, midjourney prompt help, midjourney tricks, midjourney tips, midjourney v5 tutorial, midjourney image prompting
Id: 4FQgkyMGdVY
Channel Id: undefined
Length: 5min 36sec (336 seconds)
Published: Wed Jun 14 2023
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.