MidJourney V6 Changes EVERYTHING

mid Journey version 6 is out and there are some major changes that all mid Journey users should know the first and most important change is just prompting style how you are able to prompt in version five is not how you're going to prompt in version six so there are a lot of changes and that's what we're going to mainly focus in today's episode by the end of the video you should be a mid Journey version 6 Pro prompter also the other big change here is the ability to add text this is an image that I just made here in ner and you can see it is amazing so let's take a closer look in today's episode all right so let's take a closer look at what's happening here with mid Journey the Alpha version of version 6 is out now what they do mention with the new V6 based model they have much more accurate prompt following as well as longer prompts and this is something that's very important they do mention that before I want to say after 15 words prompts started to die down now it has changed you can even have 250 300 words word length is not that important in mid Journey right now it's more the amount of rendering we're going to see a little bit about that in the future they do mention that hey look there are the ability for minor text drawing ability you must write your text in quotations and to kind of get the best solution here it's better to either use style raw or very low values in the stylize attribute uh they also do mention that there are certain things that you can't do in mid Journey V6 like panum VAR um described right now in the new version 6 version um but there are some other features and arguments that are already supported like aspect ratio like to stylize vary and so much more the other things that they do mention is they have both improved the upscaler with both subtle and creative and we're going to take a closer look at that all right so the first thing I want to do is test out this text and like I mentioned later on we're going to see how to properly prompt but here we can see a photo of the text Merry Christmas written with festival color marker on a sticky note so the first thing that you notice is my styze is going to be at the default 100 which is low right that's what they expect and overall we got some amazing amazing results three out of the four were correct spelling and that's pretty good right opposed to back then when you wouldn't get any anything close so after that what I wanted to do is kind of test out different styles as well now I started to add Style they do mention the higher your stylized value goes the less likely you're going to get the correct prompt here I use value 500 and we can see none of these are actually correct um this one's actually really nice but it's not in a sticky note like I asked before then I ended up using styze value 1,000 and very similar right everything here was pretty much a mistake the text was spelled correctly here on this marker but it was on the marker and not on a note and now if we go to my original prompt I did choose image one to upscale and within the image one I did try upscale subtle and upscale creative I do want to say I enjoy the creative one it did kind of give me some color changes here this one's the subtle not too many changes but we do see the upscale increase uh the overall resolution of the image here we can see the creative and what I really do enjoy about the creative I feel like the colors pop up a little bit more on the main text here now let's continue about coming prompting Masters here in mid Journey we can see that prompting with V6 is significantly significantly different than V5 you will need to relearn how to prompt they also mentioned to avoid junk like award-winning photo realistic 4K AK and this is pretty interesting because I do believe most prompts back then would use something like 4K AK RTX be explicit about what you want um version six is able to really understand and read out the prom promp if you want something more photograph uh photographic less opinionated and more literal it's very important to use the style raw uh for lower values of styli defa 100 may have better prompt understanding while higher values up to 1,000 may have better Aesthetics and this is something that in my opinion has always been the case they do mention that V6 is slower and more expensive versus V5 but obviously things are going to optimize as they continue to improve this model they do also mention that relax mode is supported it just takes a lot longer they do also mention that the community standard this model can generate much more realistic imagery than anything they have released before and they've also turned up their moderation system so definitely be careful now they have created an amazing mid Journey prompting style course and I'm going to take a closer look at the things that I find the most important uh first they do mention that the most important change might be to strive to write in simple sentences of English Eng that have good spelling and punctuations try not to rely too much on pronouns it helps to just repeat the subjects and we'll talk a little bit about that in a bit and they mentioned don't worry about the number of words so much as the prompt length in version six is completely completely different in version five only the first 15 to 20 words had strong influence on the picture um before they ran out of memory in version six it has completely changed you can safe they mean they're safe ly to say that they can have over 350 to even 500 Words it's not too much about the word cap but what's more important is how much rendering work you're asking mid journey to do uh and that's something we're going to learn so if you're just creating for example one subject that one subject you can have over 300 500 words now if you're creating one subject and start creating subject two and subject 3 and subject 4 the more subjects that you start to add the more rendering version uh V6 has to add and that's when they start to loose out on memory so usually if you have low amount of subjects you're going to be able to have more kind of creative and descriptive Solutions on that one subject if you want more multiple subjects those multiple subjects you're going to end up losing is on the amount of detail each subject can have um because you start to lose out on memory they also give us some example of words that will paint something on the canvas for example in directions above below beside against they also tell us words not to use extra Ultra Super Hyper insanely 4K 6K 8K rman vray unreal those are junk so get them out of your Vis version 6 prompting right away another thing that mid Journey wants to show us is that before the earlier the the text was the more influence it had in the canva but now mid Journey version 6 is really looking at kind of what's important in text uh so for example here they have this super long prompt where the subject is a ger a young goodlooking German man so you can see that and then there's kind of a lot of filler and garbage and at the end they put he is wearing red glasses and no matter after how long that prompt was these images all provided this young German man wearing red glasses so it understands what are important words and it doesn't matter necess necessarily where it's at in the text there's going to be some conflicting uh thoughts later on or or reports later on that sometimes if you might not be getting the best results it does somewhat help if you bring it up closer to the beginning of the text um but here's a perfect example that that's not always the case here is a different example of how prompt is completely different in version 6 this is definitely not your typical prompt style that you would do in version five um but pretty much is a sparring program very similar to the reality of the Matrix um with the same rules and gravities and they do mention at the end that one thing is a delicious intact large pepperoni pizza is on the floor and we can see on all the images pretty much all the images we can see the pepperoni pizza here pepperoni pizza here pepperoni pizza here and the pepperoni pizza here and very much similarly following that Matrix style uh so prompting is completely different uh so something we're going to have to get used to all right so next we're going to see that M Journey does understand language or grammar much better than version five and here are some interesting examples of prompting and look how completely different are than version five for example look at this last one and I'll put the image right now once again Herman couldn't find his keys and just like last night he stood in the middle of the room and cried about it so I do want to say three out of the four images really kind of give us Herman just sitting in standing in the room and looking pretty sad they do mention that you cannot describe a sequence or series of actions they mentioned that the details will be flatten out into a single scene or SnapChat the best results will be if you kind of create more of a a comic uh of a comic style so the first example that wouldn't work was a man went to purchase a hat he got in his car drove home and gave the Hat as a gift to his wife we can see they don't and all the images are a bit different one of them we got a hat we can see the other one dro driving with the man in the Hat and then we can see another one with the man driving with his wife and they all have hats so we don't know exactly what's happening here and we don't get what we wanted from the promt but what works really well is kind of making a comic book page has three pains the first paint depicts the moon the second paint depicts a three and the third paint depicts a bedroom and here is another one and those did actually give me some great results I mean all of these look amazing we can see the moon the tree the bed the moon the tree the bed here was another one where it kind of talks about a comic book page has three panels uh so overall things are a little bit better when you kind of depict them as a comic instead they do mention that you generally won't be able to say what isn't happening or isn't present for example you can't say a sandwich with no bread you're going to end up getting a bread completely off note I realized that a sandwich with no bread is pretty much a salad so I ended up drawing a salad in the bread and I got my sandwich veggie sandwich I guess it actually looks pretty good they do mention that sometimes you can use words like no um will not has not but it all depends here for example is a boy in the parking lot holding car keys but there are no Cars nearby and we can see we tried it sometimes they're going to have Cars nearby so they do mention that while this is not always going to work even if one image is correct that's a big change compared to V5 they do mention that now you can talk to Mid Journey a bit like it is chat GPT obviously you can't kind of say hey put more flowers in her hair or kind of direct an image but what you can do is examples like this where this would be your traditional version file style where uh you would draw a raven like this this could be your new chat GPT style which is more like writing a sentence and talking to a machine opposed to just creating some form of prompt next they talk about how to discuss multiple subjects in version six and they say the best out is first do kind of break create the simple scene then kind of do call back details then create setting details of the scene and finally add the vibe or the Aesthetics so first let's build the basic scene so they say uh okay prompt is three friends sitting on the Park Bench here you're working we're going to be working with multiple subjects and this is a good prompt and we can see we get the right results they say a better result is by adding some form of descriptiveness to it uh for example here they say three different friends sitting on a park bench by adding the word different we're seeing different races different colors they tell us that even a better prompt is by saying three different best friends sitting close together on the Park Bench Now by adding the word best friends and sitting close together you're going to get more of that friendly tone and true friendship emotion to some extent now that we built our original scene we're going to start calling back our subjects and the best way to call back our subjects is by always using the subject friend which we ended up using on our main prompt it's no point in calling someone Jennifer in the middle and also make sure to always use simple sentences don't use breakout sentences like this so for example here we put our three different best friends sitting close together on a park bench I do put that the friend in the middle is a cheerful blonde Caucasian woman where wearing jeans and a green tank top so we now get that subject in the middle is going to be consistent to our details and we do see that the friends around are still different to some extent so now I want to update the friend on the left originally we updated the friend in the middle so the friend in the left again we're using that same subject is a Hispanic male with black hair and a short beard wearing glasses a solid Rich shirt and black pants one thing that you are noticing though is now I'm asking journey to render a little bit more and more but we did get the results right now we have the Caucasian female in the middle and the kind of Hispanic male on the left and we're still getting different results for the one on the right now it's time to add details of the setting backdrop background and context they do mention a bad example of this is just adding there are life Oaks and pigeons a better example is getting a little more clearer there are some pigeons on the sidewalk in the background are some life Oaks the best is to get extremely specific there are two pigeons on the sidewalk in front of the bench in the background the empty Park contains some old life oak trees they do mention to be prepared to roll specifics back or out of the prompt if you start seeing things become incoherent and that's exactly what I saw right when I started to add that new prompt where we're adding the two pigeons in the sidewalk and the empty Park contains some old Life trees we started to get I I didn't get exactly the prompt I got I didn't see pigeons in all of them and sometimes I didn't see pigeons at all uh and that's the main reason where it doesn't and that's the perfect example where it doesn't matter how long the text is what's more important is the amount of subjects and the amount of rendering you're asking mid journey to do I believe what happening here is for my friend in the left I'm adding too much description here where I'm taking too much memory just on that one character I would guess if I was to remove a lot of this description here I would be able to add even more subjects into this canva they do mention the fourth thing is to add your aesthetic or Vibe a not great way to do this is by save photo u a better is to add color photography the best is to name a specific inspiration like a photographer's name or publication if it's not a photographic it's best to say things like digital art but the best use is to say digital art by the style of so and so so hopefully right now we are getting way better at this prompting style I do believe it's a lot to take in so make sure to kind of just try some on the out as we are watching this video if you guys are enjoying the episode so far make sure to hit the thumbs up and the Subscribe button I truly truly appreciate it I do believe my next video is going to be creating comics with version 6 as I do believe there's a lot of potential there so now if we continue here with the prompting guy they do mention that you can place things where you want on the canvas using language to control composition uh and they mention it's more reliable to describe a generic image in a short phrase than fill in the details starting with the focal point of the canva and that's pretty much what we did in the last example where we started with a simple of three friends in a bench and we worked our way on to when we started to discuss each friend individually here is a prompt that they mentioned there are three baskets full of fruits on the kitchen tables the basket in the middle has one thing the basket on the left has another and the basket on the right has another thing and they also tell you in the background there is a blank to your wall with circular windows so you're telling exactly mid Journey what you want like we saw earlier you could add text to your image and they do mention when you're adding text to the image use phrases like sis printed on entitled inscribed with labeled as marked with branded with so make sure that it knows that it's trying to draw some form of text um they also mentioned that you can also print text on things such as Peach bubble Post-it notes book covers posters like we saw here a hello world written with a marker on a sticky note also to have text or letters appear alone it sometimes helped to add the phrase typography design to the prompt you can explore using phrases like isolated on the white background if you like the Cana to be otherwise blank another great information that they tell us is what you should do if your image looks blurry or janky the first thing is like we saw earlier on when your prompt starts to get Stu re most reliably remove details right that's going to be the most important you're having too much details so mid journey is R is kind of losing its memory and it's not going to be working as intended they do mention sometimes if you're having like a blurry face or blurry hands or blurry feet or blurry something add key details to that so if you have blurry hands you're going to say hands doing something so hands on hip hands holding a coffee mug hands waving so over overall this is going to help increase uh the overall improvements of that jankiness and making it clear they do mention if you have GPU minuts Chase conery so just keep upscaling and upscaling and that will give you the better results next is what should I do if stuff doesn't show up on the canva and very similar remove other details to make room you're losing out memory if you are if you don't have everything in your canva so you have to remove details to save on the memory second thing sometimes it works to move the missing bit closer to the front of the prompt that's not necessarily always the case um but it does kind of have a little bit of weight on the front end it does say sometimes it also makes uh it's important to just try to phrase it a different way and by lowering your styze remember if your stylize is over a 100 you're going to start to see a loss inherency of your prompt the next thing we can look at is you can add a frame or border around an image by just describing it here the image has a thin gold border or frame around it uh and we can see all images actually came out pretty strong with that now if you are a mid Journey user obviously you are on Discord definitely take um advantage of the prompt chat and the prompt craft channels as there are a lot of tutorials there one that really caught my attention here with version six is now you are able to kind of do this split face I feel like back then it was pretty hard to do it but here a face with two halves the left half of the face is a girl the right half of the face is a clown and it looks amazing this is thanks to Mid Journey user clarinet uh and we can see more examples like this so version six is probably my favorite so far and before I was using Dolly 3 a lot I think I'm going to move back to Mid Journey with this version 6 update so I hope you guys enjoyed today's episode I know was a little bit longer but hopefully by this time you are now uh kind of a better prompting Master here in in mid Journey with version 6 so make sure to hit the thumbs up make sure to hit the Subscribe button take care have a good day and see you next time

Channel: Jose Najarro AI

Views: 4,153

Keywords: ai art, midjourney v6, midjourney ai, ai tools, midjourney tutorial, midjourney ai tutorial, midjourney news, midjourney v6 features, ai news, ai art generator, open ai, ai generated art, midjourney version 6, how to use midjourney, midjourney 6, dall e 3, how to use midjourney ai, midjourney v6 new features, midjourney update, midjourney v6 how to use, midjourney v6 prompts, midjourney v6 release, midjourney v6 vs dall e 3, midjourney ai v6, midjourney ai prompts

Id: JBz4I7GjL_w

Length: 19min 58sec (1198 seconds)

Published: Sat Dec 23 2023