Channel about exploring and experimenting with all kinds of AI technologies. My videos are mostly about Midjourney, prompting and trying to discover things that differ from the default style. I am not an artist.
Thank you, thank you and thank you 😀 Do you have any idea where the Midjourney styles come from and what they are used for in generating images in Midjourney when no style is specified?
If you do more of these, can you move the text up so that when we pause the video, the text isn't covered by RU-vid's controls? I thought maybe put it at the top, but you'd have to place them low enough so they aren't blocked by the title, which shows up in full screen. THANK YOU! I'm also thankful you put them exactly 5 seconds apart so we can move between them using the left and right arrows.
@@thaeyne thanks I have gen 3 and I have played around with it a bit, but my image to Vid generations have been a little blah. I am going to go back and try again!
Thanks for the update I find better results in text, using the vary region option after the image is generated. I remove all the text and for the region I just write something like text "whaterever" and it seems to work a bit better than trying to get the text to display from the first prompt. Their text generaton is still kinda wanky.
@@thaeyne well what ya know, thanks for that info. Text in general on MJ is really lagging behind other platforms, minor for me, but I know a lot of users use it more than I do. Thanks again.
Thanks for the update info, Thaeyne. I have a question about seeds. They haven't really been talked about much in v 6.0, and other than your examples, I haven't seen them used in any of the Midjourney groups I belong to or the RU-vid channels I subscribe to. Do you have any insight into seeds vs sref? Since an sref can be any number up to 4 billion, and a seed number can be any number, it's hard for me to see what kind of effect using a seed has on the image. 🥰
I use seeds for consistency when comparing things. There are actually also 4 billion seeds. Each image you actually generate has a seed number assigned automatically. That is why if you run the same prompt multiple times, you will end up with a different grid each time. I have just assigned my own number as the seed most of the time. Because then I will not get any of the other 4 billion numbers instead. And it's easier to set it before running the prompt rather than look it up later. I want my viewers to be able to generate the exact same image if they want to do so. If I provide the seed number too, they have a high probability of doing so.
@@thaeyne Oh, interesting. Thank you for taking the time to reply. I need to start paying more attention to seeds...is there a way to find the seed number that gets assigned to each image? I joined your Patreon, and I thank you for the prompt files. Exactly what I needed.
Yes, you can find it out. On the website go to an image and under the copy menu there is an option for seed. On Discord you need to react to the image you want to look up the seed of with the envelope emoji. That will make the bot send you a private message with the details of that particular image, seed number is among the info.
Hi, excuse my bad english ... i'm French and old ... i buy the pdf 500 codes 2 hours ago, i received it and open and only some prompts appear .......... my connection is ok, Adobe Acrobat too, can you tell me where is the problem ? Thanks . Friendly . Serge
Hi, excuse my bad english ... i'm French and old ... i buy the pdf 500 codes 2 hours ago, i received it and open and only a few prompts appear .......... my connection is ok, Adobe Acrobat too, can you tell me where is the problem ? Thanks . Friendly . Serge
Hello, I like your videos a lot and write the styles down. Now I made a subscription on midjourney. I tried some sref like you do, but nothing... not the styles you made they are totally simple-no styles. can you give me a advice?
I so look forward to your videos and I was blown away when I saw we'd be getting five HUNDRED sref codes demo'd....I think I speak for many when I say that we appreciate your hard work and presentation style!!
There are 4294967295 style codes. The same amount as there is different seeds in Midjourney. I have not yet come across any two completely identical ones. Some do look similar, but if you start prompting them with more words than just a . symbol, then they start to show their differences.
Those are just made up names for extra flair. I went over each grid of images with ChatGPT and prompted for possible names for a style that would describe a grid in the best way. These are the ones I thought described the styles the best.
There are several ways to do that. You have the --p that gives you your personal style based on your image rankings. But if you want to create your own style it is always possible to mix and match some different style codes together. In that way, you will probably get a really unique recipe, especially if you factor in that you can set weights for the different style codes as well with double semicolon.
I don't know what the point of releasing hundreds of codes most people will never use It would have made more sense to pick the 100 most important artistic elements and let people mix and match them Anything not in that 100 you just add yourself in the prompt
I like the current presentation. I'd rather pick the ones I like which would be different for other people. The only improvement I could see is if this was in a scrollable list of images and sref numbers on a website.
Thank you very much, and I look forward to seeing your next videos! Your content is not only informative, but also presented in a clear and captivating manner.
This video has inspired me to try out some new Midjourney styles, this is such a great resource for Midjourney style codes! Thanks for putting this together. I love how you included your favorite styles in the captions. It's helpful to see some examples of what each code can create.
Thanks, didn't realize I was a computer though. But it would sure feel nice to be one right now for personal reasons I don't want to disclose at this point.
Thanks again for another great video. I really enjoyed this in-depth video. I was wondering, though why all the -serf original examples were in landscape, the four prompted images were in portrait. Any particular reason for not creating the prompted images in the same aspect ratio as the original?
I guess it's because I have always liked the 9:16 aspect ratio the most when generating images. And I have found it handy to put four of them side-by-side and have room for the prompt text. But, for the quicker looks I decided to go with the 16:9 ratio, because it fits better to a RU-vid video and I wanted to just generate one image per style. So I guess just personal preference and reasons for video editing.
Great video! I've been searching for more monochrome styles. Also, you should try a concept prompt along with your real ones. For example: Factory, Victorian, Park are real things. Whimsical Surrealism, Abstract Joy, Suspicious Behavior are conceptual. I find that giving Midjourney concepts as prompts brings out hidden possibilities from styles.
They are, but if you go beyond the surface, they actually don't behave in the same way. At lest that's my theory. I just released a video where I looked at some similar monochromatic styles but when prompting the same longer prompt with each of the style codes produced really different results.
@@thaeyne You're probably right, but some styles are very similar. What I noticed is that depending on the prompts, the similarities are more or less obvious.
That is totally strange, I still can't see any held comments, so it's not even letting me moderate them. One would think that they would do it with some more serious topics. But they are probably using AI for something like this and one thing is certainly clear in my opinion - AI is not perfect or foolproof at the moment when it comes to anything. And if they are doing it on purpose, then that is even weirder and it sure makes one wonder why.
Another theory I have is that I have marked my videos as altered content. I don't want to deceive anyone and I want to make it known that I use a lot of AI generated stuff in my videos. So maybe that has grouped my videos or even my channel to more intensive word filtering that I have not been aware of. There is a lot of anti-AI sentiment going around so anyone openly using AI would be a target to some nasty comments. I don't know how these particular product names in combination have ended up on the to-be-deleted-without-trace list though.