Perfect portraits in MidJourney AI with photos

hello there in this video we're going to look how to make a portrait based on the photo inside the mid Journey with help of AI so let's go ahead and look what's happening about a few weeks ago I posted on my Facebook social media and it says Hey guys uh if you want any character any of your photos submitted and I'll make characters so and you can see it was the response was 576 comments this includes some of my response on average I would say about 200 people send me photos and the best parts of this they send photos in all different formats sizes this is my way actual size let's see what size was this one 145 I mean there's a different sizes different formats different shapes uh blurring images they send black and white send some with the Instagram filter so it was so huge variety of different images that it was excellent to work and try to see how it will work so what I end up doing trying with minimum process if it's possible put in mid journey in just with the strings or text prompts create images so in I did create few images here and then create more and create more and create more and create more and create more and more this is not all of them but it's few of the images on difference because it's always great over I think now it's close to about 800 plus plus images photos basic so I create on different with a text string try to them look nice based on what people submitted and everything so requirements was this way send me your photo the mag shot preferably and who you want to be or what theme So based on this I used me journey to creative so let's go ahead and now jump inside the mid journey and I will show you how you first you can upload it and do it by yourself and next I will show you some problems you may occurred and how to overcome them first right here on the bottom of the Discord you'll see the plus symbol click on this plus symbol and you will have it upload for it all right I played upload files so from there you can go to collection and this is just one portion of the collection I have it and we can go select the smallest image um maybe you know let's select different image as well the one nice things you can select more than one image and you can select up to 10 images if you're on a paid version of Discord if not you can have it I think like three images but just even one at a time should work just fine so right here I'll select three different images when I select press enter and it will upload them so after upload complete you'll see them just playing in your Discord what we need to do at this point just click on one image don't right click left click on the image so it's open and preview you'll notice on the bottom says open original so we don't need open instead on middle of image we need to right click and going to copy image address on Apple it will be copy image if I'm right so we'll go copy address next we're going down and as usual we'll start with imagine the next we can pass parameters so and we can pass two parameter two images sorry we just copy paste the path we have it and we can put it another image if we want integrated but in this case I want to work just with one image so we'll put there and next approach it's what I call guided prompt okay what is meaning um and again that's my own personal terminology but I like it to explain me I want to guide my prompt so I want to start with a simplistic as possible and add more and more to this the reason is why that when I try to put all complex strings something does not show something missing and I don't know which part or how they react in this case and the best things with a portrait start with simplicity so for example here let's go just type Viking okay it's all what we need to do and right now the it will base it this is run on a version four so we'll go let's type settings and I'll show right here you can see it's run mid Journey version four it says best quality style medium regular upscale I'm running in public mode so you're more than welcome to access normally I almost always have a remix mode on it's allowed me to go and modify if I need it and right now I'm running in fast mode so we can leave it faster preview and generate it but relax mode work just fine for the portraits in fact if you're not on a hurry or whatever for a relaxed mode right now work almost as fast as the fast mode in many cases um you just cannot upscale it but again with version four it's what's happening okay so let's go look right here with almost 50 70 okay and this is done let's go look and you can see it's kind of okay start mostly I'm looking on a facial right now ignore elements that like no um stereotypical this you know horn kind of helmet which is historically inaccurate but it's what we like to see okay but facial is look okay so it's recognized next let's go ahead copy and usually at this moment a point I'm just copy with this short path name because it's much easy and we'll go again imagine and we have a viking already established here's a little trick if we specify um body if you want to create a full body or anything you just need to specify shoes and a head so let's go ahead um liking helmet so we'll just put it it's a Viking and you know what sometimes it does work if you put emotions it's kind of I think it's look fun so let's go for example we go with a screaming so now we have a screaming or biking wearing liking helmet okay let's go ahead see what we have at this point okay here's completed let's look and you can see we already have a top head see like before it was cut off because we put this helmet when AI recognize helmets that try to embed it what is better way just pull out and add top of the head this is kind of trick what I'm using in some photos when I'm on display top I maybe say a big hair a lot of special hat or something that is prompt AI to push up so we'll have a little bit more head and right here look we have nice perfect screaming okay and the next one let's go copy what we have here okay and oops we don't need this we need imagine of course first let's go paste this biking helmet and we can add a little bit more to this for example I can say um wearing Kilt we don't have a feed or anything so I'm just saying hey it's new to wear and kill so we'll see what AI come up with then we can say mountains bottle and background Viking ship so I'll give it now variety the interesting things about this that if we add like for example a bottle on the background to biking ships AI will give it as all those different versions somewhere we see a ship somewhere there and this is one thing what I tell avoid to do this at the beginning because if you give it too many variations too many variables EI maybe go and create one way or another because weights is not yet set and that way you have confusion creating so when you create with a guided cyst like what I would say small things you're guiding step by step AI to add complexity then you achieve better same like right here and I'm kind of already done but if otherwise I already will render once now we'll render second another time but maybe add like golden hour more for the lighting effect you know you can even add Rembrandt lighting as well and so on so you can add elements um I'll put it link for you so you can watch some video on the prompt what I was recommend how to do with a specific settings it's not require and going specific order however I prefer to do this way because allowed me to focus a little bit more on a um weights and on the wards itself okay so it's uh ready to render the couple things for me personally in the string when I put it right here when I put it wherein killed that is you know it maybe ignore it or do something weird stuff because when we're doing with the porches close-up portraits it's usually yeah one expand that big um we could create a little cheat we could create a body and actually import a second image I'll show you what I was mean by that okay so let's see what we have right here and you can notice we have now that we have it background set we have the water ocean some piece of ship or other things and we have the lighting so you can see how we add more elements so in case if you want full body but full body is not provided what you can do is going online and look for some of the reference images for me personally because I'm a photographer I have admire already collections of images what I've done I can reuse it but if you don't have it for example if I want to do like full body miking in this case I'll just go and Google search find something like this image um notice right here we have a face and this is a little bit more advanced maybe but problem is sometimes it will take face that already exists this is not what we wanted so you can go inside like Adobe Photoshop example open this image and you know what I'm going to do is just take this face and remove it or blur out I will remove the information from AI to reference in because I want to use the face that is already I have so we'll just go ahead and Escape take small blur tool okay uh where's my blur tool right here my blur two and we'll just blur this face out notice I want to keep it coloring in because I don't want there but I want blur so AI will freak out and says hey I don't know what face to use here so I'll use that one so like this and we'll go and save it now we can go back in Discord and we already know how to upload files so we'll go select and upload our files anyways my biking right here we'll just go and upload this if icon um be careful when you take any image from online because they may copyright it so just be sure you take non-corporated free images which is you can easily find okay right here we have and now what do we have we have our two in this string it's what I'm going to just take one of this whole string okay we'll go to imagine paste it and next I'm click on this image right click copy image address and remember what I says we have the first image going right here now between this image and the text strings we can put a second image so I'll go ahead press space we need a space between nothing between these two and reference second image so we have one image right there we have a second image and now we have its strings that we already have right there preview okay let's go ahead press enter and see what we come up with okay let's look what we have it and you can see well we have our arms chest same face so we started getting a full body to stand all what we needed it just adds something else to trick our AI to render this so if we're going to imagine and we can add wearing Kilt just before wearing kit let's remove the wear and kill we just say standing on the top of the Hill okay and next we need to put it maybe like uh skulls or something you can you can add like bonds whatever you want you know like a bottle just create kind of so we can this but it's Standing On Top of the Hills um above Village maybe actually I think it's a double L but we'll see come on they go if you're interested you'll notice I have a sometimes good in stuff coming out underline for me English not my first thing language I do a lot of mistakes a lot of errors but with the grammarly it's a free plug-in you don't need to pay any money for this to use it you can install in your web browser it does help you to kind of underline and fix it grammar spelling and other stuff so it's very good plus very good benefit okay let's go try this okay let's go check where we have it and look right here we have our standing on the top of the hill overview I mean in the village on the back it's a perfect okay we can again work from this we can add more elements because it's a guided way remember we add more to twink around but generally we achieve what we got from the portraits where it look like a person we add body in all these other elements um personal suggestions for you if you really want to do like these portraits and everything you want to have at your own Library maybe of the people what you can reference like second without face or anything just a reference the better way best things to do this it is go ahead and download it does Studio that studio is a free software from dots 3D okay and it's come with already a lot of models that you can use it for free male and female then inside it's like very simplistic if you wanted I have plenty of the videos just search on my YouTube channel for the dial Studio you'll find how to do everything but you can pre-create actually pauses of their um models render them and use it in different areas because all what you need to do is for AI to reference full body and it's work very very well and best parts has said before it is free to use okay so we'll look on some of this one what else we can do with this the one things that I also want to point when you work with porches and sometimes you run to this problem when you upload it normal photo you try to render and May Johnny come with this NSFW note safe for workplace I mean this is fully legit image problem is the NSF filter it is not yet optimized this still in progress they're still kind of adjusting but it is frustration when you try to work some portraits and certainly it just does not want process because thinking it's not work safe um by the way you know as instant things it's all relatives yeah at one time I used to own um strip dancing club and you know not safe workplace there was very edgy so it's all relative but seriously speaking you can see right here you know it's a normal image it's a totally work safe but you're getting this error and what I was trying sometimes you try to play around like for example okay let's say what's happening if I just blur out maybe it's color based or other things nope it still have this one so in this case what I found it's work best is just cropping like crap on the face on an image this way because for some reason this filter is does not work usually if I loaded this image and I know it's work safe and I know filter inside the mid Journey it's fail I try to submit this twice three times so it will flag out make a flag and other mean or developer team can look and says oh yeah it's a safe it's something wrong with our system so they can adjust this as well I'm going to the Chart groups and with the developers and I just posted with members just raise the flag and let them know hey guys your filter does not perform properly please adjust or this because um in some cases if I work this for the client and client well actually does in some cases send me this image and says hey I want to do this I cannot actually work for the client job I cannot perform this on that because the filter does not work correctly so what you do in this type of the cases you have the one option so you need to go inside some other application on your cell phone or photoshop and what I found it's coloring sometimes does not work this will work just like this just crop to the face okay that worked the best it's cut off elements and it will start working and processing then you can work with a specific image if you need Again full body shot use it to reference maybe or other photos with other person might do recommendation if using other photo blur outface or you can just even put it same color like skin color and coloring over the face because otherwise it will start merging those faces and not necessarily look good it's almost will be similar if we just have a face and just says SpongeBob and you can see it's come with some very weird faces so this point what I want to do is actually give you some tips on the strings text strings how I use it and what I use it and what's happening so let me go ahead and we'll just use this string right here okay our poor guy is our experimental guide so what is happening if I want to create a Grinch The Grange Crystal the Christmas okay install the Christmas okay okay let's correct our super fine thinking but what is happening you'll notice it will try to create but you'll notice it does not necessarily look like a grange okay so let's see what's missing and this is a way I'm actually again guiding my prompt so I'm going back I'm typing this but I've noticed it is missing green skin so let's go green skin that is what missing and it's neat have it fur or long hair so long fur fuzzy for so I'll just add those ones I'm usually going and see what does it missing from the portraits that I want to display okay so let's look right here look it's much better okay so we'll start having green shade we still have the original face and in some cases this is before it was look like a gnome or something else maybe but here we're getting more closer to the Grinch um you can go all the way green screen and we can readjust the values like right here and remember that says it's positioning um sometimes green who stole Christmas okay we can have it I don't think we needed this we can just remove the Christmas remove their this with a green screen we can emphasize a little bit more on this and that is well created so we can guide our prompt to add or remove elements but I think this is work very well okay so look at this we wow that has come up ah scary stuff okay so the next things for example we can play more let me show you the next step probably it will be lost on this portraits before we jump to another one so right here I have it greenish with green screen uh let's go like very long hair for a lot s of fur we can do this and now we want to add also background we created our main character yes for the background we want Christmas tree and lights in the background so we kind of specifying what we want to have it in a background okay now we created when you're done creating your characters and you can tweak around you can go back and forth for example if this does not work work too scared you won't create kind of more like this you can work with that one as well but this one is a little bit I don't know what the green is look like toothless greens anyway but yeah you can create this and then next we'll start working on the environment that our character is located so for example we can add like what I said before Christmas tree Christmas lights and all this element okay let's preview and right here look on our background it's perfectly putted stuff and a background of course if you don't want smiling Grinch you can go in and just as very angry Mr Grinch or playing around this you can also apply emotions which actually mid-journey and I will translate very good in this case still is rendering let's go to another ones and you know you don't necessarily always need click on a plus you can take sometimes image drag to the web and you can see right here uploaded if we click on this you'll have it same upload the single image on the multiple and right here we have it black and white portraits which is kind of okay but what I found under some prompts work a little bit better so I'm going to copy this one and I'm going to says imagine portrait and let's go most beautiful Disney princess why does the sum prompt work better than another ones related to the what training model what model they was using how the train was done okay let's just render kill its rendering here's our Grinch look on it so this is perfect okay perfect we have a green color skin we don't want be too off from the portraits because you create a portraits for the people living in December I think this is a profit resemblance a little bit angry a little bit Grinch so this is kind of winner almost an hour at least for me you may work more but for me these two porches the almost winners of the Grinch Christmas green kind of related to the person portraits that we just use it okay so as a render because if we use this black and white this is probably sometimes it will come out black and white but you always can specify so if we're going can we put it in Imagine portraits we can have it um Vivid colors okay and remember other things we can set as a background for example a witch house or Enchanted Forest okay you can add all these elements golden hour then the portraits actually will become colorized with our Disney princess look for this portrait so it will be nice it will have its colors in gen followers so we can all this modification to add to the portrait and here we have it we have perfect golden hours right here we add some coloring nice positioning and you can see how easy with the just Porters to start adding some information remember you can always use it secondary photo or secondary image as a reference if you unlike full body shot like Bell or something with a dress you can put right here and it will help to blend remember remove some of these elements from the AI like the face or other things so AI does not have an information to render on or actual reference to so if it's just a blur big blur it will take from first photo and put it on a second so this way you can literally guide it the AI prompt to create image you want it because you can work with more different prompts try experiments some work very good some images will inspire you for example right here I see the neon light and it's inspired to me as a cyberpunk so if we're going and we'll says imagine take this put it inside and we say cyberpunk CT Narrow Street neon light so notice this I don't necessarily specify the character in this case because I know it will take photo and a character and character already developed for me okay I can put it sunglasses maybe something but I want to also describe the environment in this case and let's see okay right here you can see with our prompt we have a perfect Cyber City person we don't even specify because it will take as much as possible of the person that already exists I don't need modified I can put it in and one more tip because I have a lot of them interesting ones you are also can um specify okay let me erase this stuff there you go you can specify the ratio of image will will effect as well how it is will perform like example right here you can see we have the expect rate show let's go ahead put it and this should give it us portrait mode which will change in some ways how the image will look remember to use them if you need it like if you have the mermaid you probably want landscape and that way we can pull out tail and other stuff but in a city narrow cyberpunk I want kind of narrow so I want a two by three so this is the nice things about mid Journey version 4 that we have it also those aspect rate shows okay so let's preview right here you can see we heard perfect see how it's work I think a little bit better for us narrow but again you can play around with different style overall hopefully these tips was helping for you have some fun time to greet it and again watching for those videos have a fun time create and remember subscribe and put notifications and spread this video have a great time

Channel: Vladimir Chopine [GeekatPlay]

Views: 112,879

Keywords: Geekatplay Studio, Vladimir Chopine, Digital art, How to, Learning Photoshop, How to do in Photoshop, How to do compositing, AI art, MidJourney, Stable Diffusion, Dreambooth, Dall-e, Free resources, Free learning, Digital art for begginers, Free tutorials, artificial intelligence, Digital Photography, Photography, Portrait ai, prefect portrait

Id: Gp61v9X23KA

Length: 28min 54sec (1734 seconds)

Published: Thu Dec 01 2022