The --sref placing in Midjourney aids in keeping a unified style amongst photos. Your visuals will be consistent and unified if you make sure fresh images follow the same style as the reference by referencing it with --sref. For making series or sticking to a particular style, this is perfect.
I would use this method, to match your style try to get as many similar images as you can for it. If you want it to use the layout you made, Id use Leo ardo Ai's image guidance and edge to imagetl to replicate the layout.
Thanks for the vid, this new function is way better then the describe function. Atleast I found this function to deliver the images I really want when using older MJ creations then when having MJ describe them.
Question: what’s the difference between putting the image link at the beginning of the prompt compared to putting it at the end of the prompt before the -sref parameter?
Thanks for all your videos, fantastic. Quick question, is it possible to have an image that you want to apply a style to (example a picture of your cat of dog), then add a style reference image, to that image, and get a result that looks like your pet in the style that you have chosen? If yes, how to do you do this on the Midjourney website?
It's not perfect, but it gives interesting--if not accurate--results. It maintains the "style" but when using this sref on established character creations, the results are far from the original base image. Perhaps the upcoming cref will work better. Now what I mean by "far from the original image" is that the style is "enhanced" not "original". For example, the original image is soft "concept art" flavored by scifi/cyber. The resulting image has a hard "glow" that's way different than the "accented" neon reference in the prompt.
What are steps to use my own drawings and own brush strokes for reference I tried to upload to my Google drive and the copy that image link but this did not work and then I tired to upload to Google photos and I copyed image link but this did not work either please help
I have found the most reliable way to upload images for use in Midjourney is PostIMg.cc - this is how I get it work: th-cam.com/video/_WF3OQPYqR0/w-d-xo.htmlsi=MAglHjaG5E572z8e
Hi Wade, I have a question about using the sref parameter. I was hoping to upload a photo and use sref to generate variations of that image without providing any additional text prompts. However, it seems sref requires a prompt. Is there a workaround to generate images using only the uploaded picture as input, without a prompt that could potentially alter the image significantly? If a prompt is absolutely necessary, what would you recommend as a minimal prompt that preserves the original image as much as possible?
Midjourney will typically be a little creative with it. You can experiment with adding it as an image prompt also? Maybe even add ::5 directly after (no space) the url of the image prompt to max the image weight. It may not work but worth a try
Maybe! They may have a plan, I have no idea haha. I do know, Midjourney was head and shoulders better for some time but the other platforms a catching up fast. So they do have to make some moves!
I do think there's an element of the style.being converted to a text description and then added to the prompt in a way (purely a theory) as I often find elements from image prompts also turn out more like a description of an i.age used and not the image itself. Hopefully this improves in time too
You are consistently one of the main reasons I come to TH-cam - you keep me inspired and excited about Midjourney 🙏🏼
Thank you so much! I really appreciate you saying that and the kind gesture! Have a great weekend mate :)
The --sref placing in Midjourney aids in keeping a unified style amongst photos. Your visuals will be consistent and unified if you make sure fresh images follow the same style as the reference by referencing it with --sref. For making series or sticking to a particular style, this is perfect.
Excellent tutorial. Very concise, uncomplicated and easy to follow.
Thank you so much!
Another stellar, on-point tutorial, m8. Thanks again for taking the time to produce & upload.
No problem mate, thank you for the positive comment, I really appreciate it :)
As always, your content is gold !
Thank you so much!
This is priceless! Oh man I’ll be using this amazingness bigtime and so great for client brand creation
Yeah! Especially if you really dial in a style with 3 images and the strengths to go with them, I'd image it'd be really consistent.
What are tips in using my own brush strokes and the way I draw to stronger Influence the end result?
I would use this method, to match your style try to get as many similar images as you can for it. If you want it to use the layout you made, Id use Leo ardo Ai's image guidance and edge to imagetl to replicate the layout.
Thanks for the vid, this new function is way better then the describe function. Atleast I found this function to deliver the images I really want when using older MJ creations then when having MJ describe them.
Definitely, it's a big step forward for Midjourney, 100%
Thank you for breaking this down so quickly and simply 👍
Thank you! I appreciate it
2:04 Why when I click on the remix button, it automatic generates new images without show the promt box?
You need to turn on remix - type /settings then click the remix button so it's green - that should fix it!
@@WadeMcMaster thank you
Great video mate. Youre a good teacher - slow walkthrough, then working example, then reinforcement. 100% 🎉
Thank you so much!
Thanks for your content always enjoy your videos ❤
I appreciate it, thank you! I hope you have some fun with this one :)
Thank you for such a brilliant video. Learnt so much today.
Thank you!
Excellent explanation! Thank you very much for your content!
You're very welcome! Thank you :)
Question: what’s the difference between putting the image link at the beginning of the prompt compared to putting it at the end of the prompt before the -sref parameter?
I'm just going by the advice of the midjourney team on that one
When you put it in the beginning of the prompt it tries to copy the composition of the reference image rather than the style
Thanks super simple vid!
Thank you Mandy! I definitely.hoped you'd find it easy to follow :) thanks for the support!
Thanks for all your videos, fantastic. Quick question, is it possible to have an image that you want to apply a style to (example a picture of your cat of dog), then add a style reference image, to that image, and get a result that looks like your pet in the style that you have chosen? If yes, how to do you do this on the Midjourney website?
Using this feature as the tutorial goes you can do that - however I don't believe its available on the alpha website yet sorry
It's not perfect, but it gives interesting--if not accurate--results. It maintains the "style" but when using this sref on established character creations, the results are far from the original base image. Perhaps the upcoming cref will work better. Now what I mean by "far from the original image" is that the style is "enhanced" not "original". For example, the original image is soft "concept art" flavored by scifi/cyber. The resulting image has a hard "glow" that's way different than the "accented" neon reference in the prompt.
Yeah, it's definitely a step forward . I do look forward to the consistent characters feature they're working on, that'll be interesting
What are steps to use my own drawings and own brush strokes for reference
I tried to upload to my Google drive and the copy that image link but this did not work and then I tired to upload to Google photos and I copyed image link but this did not work either please help
I have found the most reliable way to upload images for use in Midjourney is PostIMg.cc - this is how I get it work: th-cam.com/video/_WF3OQPYqR0/w-d-xo.htmlsi=MAglHjaG5E572z8e
Hi Wade, I have a question about using the sref parameter. I was hoping to upload a photo and use sref to generate variations of that image without providing any additional text prompts. However, it seems sref requires a prompt. Is there a workaround to generate images using only the uploaded picture as input, without a prompt that could potentially alter the image significantly? If a prompt is absolutely necessary, what would you recommend as a minimal prompt that preserves the original image as much as possible?
I just tried image::0 so a single word, with a zero word weight ( hence the ::0 ) and it seemed work with sref
@@WadeMcMaster tried but changed the word "image" with others like horse, dog, etc and seems to still be using the word even with the zero weight :(
Oh I see. I'm not too sure sorry mate. I've never had to Come up with this kind of prompt before.
This is incredible!
I can't wait to really throw some stuff at this :)
Thank you! Can I share these prompts with other people, or are they're gonna have to upload the images themselves and take the links out of them?
I don't know how long those URL's last, but I imagine you'd be able to share them.
@@WadeMcMastergot it, thanks!:)
How do you edit the promp when hitting the remix bottom?
I don't see that option
type /settings - then click remix and that should activate :)
loved it straight to the damn point 💯💥💫💣
Thank you!
That's really cool 🤩
Yeah man! Lots of fun :)
how you edit the promt after generating???
you pick the "refresh buttom"?
i cant understand
Type /settings and turn on 'remix' mode
@@WadeMcMaster that still doesn't work for me. it just keeps refreshing the 4 images
how come remix doesn't allow me to edit like it does in this video? it just re-uploads the Image
Hey mate, go to /settings and turn on Remix - that should do it
how to make the same design as in the reference image?
Midjourney will typically be a little creative with it. You can experiment with adding it as an image prompt also? Maybe even add ::5 directly after (no space) the url of the image prompt to max the image weight. It may not work but worth a try
Love this!
It's pretty cool! Lots of fun :)
Does mid journey provide any APIs? Like ChatGPT or Dalle ?
Not at the moment
@@WadeMcMaster lol, don’t think they will fail or be behind when OpenAI provides such via subscriptions?
Maybe! They may have a plan, I have no idea haha. I do know, Midjourney was head and shoulders better for some time but the other platforms a catching up fast. So they do have to make some moves!
I would be interested in Seed vs Style differences.
So you mean trying --SREF or different style phrases with the same seed? That would be a good idea :)
How come when I hit remix, midjourney create 4 new images for me, what am I doing wrong? I can't edit my prompt like you.
nvrmind, I fixed it under /settings. I had to turn it on
Was about to say this - I'm glad you got it sorted mate :)
@@WadeMcMaster glad that you were about to help. I'm still a newbie, and thanks for your tutorial
That’s so cool!!! 🎉
Yeah mate! Have a play, it's good fun :)
My drag and drop image disappears once I type, /imagine
Are you hitting enter to send those images into the chat? Also if you still have issues try posting.cc and copy the direct link
Thank you sir.. I did copy link and is so, so great .. appreciate ur help and awesome vids.. so cool !!
Thank you!
really good info
Thank you mate :)
Excelente.
Thank you :)
finally!
Yeah this feature will be a lot fun :)
👍👍👍
🤘
👋
👊
Not entirely impressed yet. Some styles simply don't convert well as a reference.
I do think there's an element of the style.being converted to a text description and then added to the prompt in a way (purely a theory) as I often find elements from image prompts also turn out more like a description of an i.age used and not the image itself. Hopefully this improves in time too
great job, well done, Thanks a lot
Thank you!