Midjourney v6 Tips and Tricks. Text. Chaos. Weird. Stylize. Part 2

hello my name is th and I like to experiment with all kinds of AI Technologies in this video we shall be taking another look at Mid Journey's latest version 6 and its features there is a link in the description to the part one video of this series so go watch that as well if you want to take a look at some of the really basic features in version 6 but let's start right where we left off one of the really big new things with ver six is that you can now get text to appear in your images previously it was really difficult to do that and you basically always ended up with gibberish that never matched what you wanted but now things have improved significantly with version six it is not yet 100% accurate but you can get some good results occasionally all you have to do is put the text you want in the image in quotation marks in your prompt previously quotation marks were simply just symbols for mid Journey but version six actually recognizes them and treats them differently just like Curly braces are used for permutations double colon followed by a number is used for multi- prompting and two dashes in front of certain words or letters are used for parameter syntax so now on to some of my observations that I have noticed so far with this new feature I think the bigger struggle to generate words in the image is with more obscure words or phrases for example in my latest video I wanted to write-- V6 and-- v6.0 but after several attempts it became clear that it gets close but never exactly right and the reason being that my text was not something that is commonly written but more recognized words do get better results longer sentences may also be a bit difficult to get right simply due to the amount of words that may get scrambled more easily so it's possible but be prepared for a lot of trial and error single words may actually work best at the moment because the letters tend to turn out in the right order most of the time sometimes using the-- style Raw May help if your text isn't showing up correctly and to get your image right you can direct it by telling it how the text has been created like typed written or other phrases like that you can also direct the medium like on a sign or billboard paper or post it your creativity is the limit but again there is a need for a lot of trial and error but there is a potential Now to create better logos or even entire new fonts and I do realize that there are plenty of AI tools out there already that can do this better and faster at this point but for Mid Journey this is a really big leap forward now let's look at some older features and what they do in version six first up is the chaos feature with-- chaos or-- c and a value from 0 to 100 the default value is usually zero but the higher you go with it the more variety you will see in the outputs so basically whenever you don't clearly know what you are looking for this will give you different and unexpected options quicker for this test I chose the prompt rabbit made a velvet in a Desert Oasis I use the exact same seed number for all the images I generated and I chose the chaos values 0 10 50 and 100 obviously there is other options between as well but I wanted to Showcase how it goes from nothing to extreme it's been a while since I have looked at chaos parameter like this but to me it feels like the zero and 10 images are really close and also 50 and 100 look kind of similar in some of the images it does seem to stray from The Prompt quite a bit with the higher chaos Valu and then the exact same thing but with raw style again I feel like the zero and 10 are really close as are 50 and 100 partly with some differences and I see the chaos 10 colorful rabbit continuing as a colorful tree in chaos 50 and kind of a colorful totem in chaos 100 but the straying from The Prompt is very strong with higher chaos values so my recommendation would be to stay in the lower values if you want to see things that you actually Pro Ed another really similar parameter in my opinion is the-- weird or --w as the name suggests it controls how unusual the images will look the accepted values go from 0 to 3,000 and the default is again zero meaning that this parameter is not normally used if you don't specifically mention it in your prompt so let's see how version 6 behaves with this par parameter this time I used the prompt chair carved from sandstone in a mystical foggy Valley I have again used the same seed number for better comparison and this time I selected the break points of 0 300 1,000 and 3,000 the images didn't come out as the prettiest but they are sufficient to show what weird does weird zero represents the default and it looks different from the other bunch I also see three sets of images repeat meting across weird 300 weird 1,000 and weird 3000 when this feature came out with version 5 I didn't know what the purpose of this was other than to create really ugly images and I believe that the same is the case with version six raw style looks actually much better for most of the images we can see some images again repeating across weird 300 weird 1,000 and weird 3000 I feel like they look somewhat like very subtle results so across both default style and raw style I don't think you even need to go very high with the values you can get similar results at 300 already and although the raw style doesn't actually look as weird or unexpected to me with version 6 I don't think I will be using this feature in the future if you are into this kind of experimentation you can also combine the two let me know in the comments below if you use chaos or weird and why I would really like to know if I am missing something here with the usefulness of these features there is another feature that has been around for ages and is somewhat similar to the features we have looked at the only difference is that it is sometimes actually useful it is the stylize feature with-- stylize or-- s going from0 to 1,000 the default for it is 100 so you actually use it in every prompt it is basically for controlling how heavily the mid Journey's default style is applied to your images let's look at it in action with version six I used the prompt birdhouse painted with colorful nebula motifs in a cherry blossom Grove and I chose the break points 0 100 250 and 1,000 the differences for this prompt look really subtle actually this came out in a photographic style so there may be different results with a more painted look the default look tends to vary from photographic to painted with version six but zero is indeed the ugliest selection of images there is only a tiny difference between 100 and 250 and 1,000 does have significant beautification added on top there so in contrast to weird and Chaos this parameter makes images more beautiful instead of ugly and now the raw style it looks very similar to the default style in the case of this prompt because raw style tends to be overall skewed towards photo realism but there does seem to be slight differences to default style styze zero still looks like the ugliest and styze 1000 has some beautification on it but I feel that it's less than with the default style along the way while testing these things I have got the feeling that the default style and raw styles are more connected now meaning that when you use the same seed number the default images and raw images are actually using the same seed space so as another experiment I overlaid the default style images on top of the raw style images with 50% opacity and this shows clearly that the compositions are really similar styze zero seems to have the biggest differences but there are grids that are nearly identical there are some more features available for version six like Das Dash tile or the image blending I'm may make a video about those features again in the future if I figure out how to best showcase the difference they have now in version 6 but I think tile works just like it is supposed to and it makes your images have seamless edges that connect to each other and image blending is supposed to work better now but I haven't been experimenting with that so much that I would have any useful tips for you beyond what I have already shared with you in a previous video I made when the feature was released anyway I thank you so much for watching consider subscribing to my channel if you like this kind of content and let's continue prompting

Channel: Thaeyne

Views: 12,810

Keywords: Midjourney, aesthetic, ai, aiart, d-id, exploration, imageprompting, midjourneyart, midjourneyv5, midjourneyv5.2, mubert, runway, runway ml, feature, prompt, prompts, midjourney prompts, midjourney styles, photo styles, ai photo, styles, prompting, prompt engineering, ai prompts, kaiber, midjourney tutorial, midjourney ai, midjourney ai tutorial, how to use midjourney ai, elevenlabs, midjourney v6, midjourney v6 features, midjourney tips, midjourney tips and tricks, midjourney tips and tricks v6

Id: O7dEhLYL2bc

Length: 9min 55sec (595 seconds)

Published: Mon Jan 01 2024