Loading images in Midjourney ai | Tutorial | AI Art

[@aussie_boosh]

I think the key insight for me with image prompts was realizing that when you give it an image it is first breaking the image down in to a text prompt by reversing its standard process, and then feedback that in... which means you can use it quite successfully to replicate characters for example by adding some anchoring words your new text prompt like "lady with red hair".

[@revdancatt]

Yeah, the more you play with it and discover tricks the better you get at it. People think you can just throw words at it and get amazing results (and you can), but it's so much better when you get to have even more control over it.

[@samirshaikh862]

If I wanted to put my face to a character in an image created in midjourney what would you guys recommend? Prompt an image url with the desired look of my face next to descriptive prompt of the scene?

[@aussie_boosh]

@@samirshaikh862 pretty much... best way is to pass in a photo of you, with a rough high level description of yourself, along side with "portrait, digital matte painting" and then keep regenerating it until you get one that looks pretty close (you won't get perfect). Once you have that one, you can then pass that image in to another prompt with same description but "35mm photo" and then yet another with "colored pencil sketch"... once you have all three versions, you can then pass them in to any other prompt and as long as there is a lose description for it to latch on to like "thin man" or "tall woman" or so forth as part of the prompt it should put you in.

[@samirshaikh862]

@@aussie_boosh I'm gonna try and mix midjourney with deep fake first to see if that works faster:)!! But thanks a lot for your input man!!!! Take care!

[@mercuryman1250]

Thank you so much. Your channel helped me out really well with pictures I've drawn by hand and enhanced them really well, I thank you. That's why I am a subscriber to your channel

[@danielsweeney8164]

Well explained "the right tool for the right job". Liked and subscribed just for that as you are the first person to clearly explain that. Excellent!

[@revdancatt]

Hah, thanks, glad you liked it.

[@sambillingham7052]

Hey Dan. I'm having a lot of fun with Midjourney so far. Solid info. Keep up the great work with the videos.

[@revdancatt]

Thanks, and yeah, it's a lot of fun. Sometimes the kids gather round and we spend the afternoon jumping from one idea to another.

[@77dris]

Thank you! I had been trying to get images to look a certain way and it never worked in Midjourney. I just gave up thinking I was doing something wrong. Your video is the first to explain this all out in perfect detail. Great job and thank you.

[@revdancatt]

Ah, it's a pleasure, I hope you're having fun with it.

[@JaimeLannister13]

Thank you so much for pairing this down into something palatable.. much appreciated & overall educational (THANKS!!)

[@baraka99]

Thank you for explaining seed and image weight. Those were codes that I was failing to use. I've been using the other descriptors such as lighting, artists, image size, etc.

[@revdancatt]

Ah, glad I could help. Hope you're having fun making images.

[@amj2048]

I didn't even think of the image weight property, that help me a ton, thank you!

[@revdancatt]

Glad I could help.

[@Bensboyforreal]

the new test beta is insane! the anatomy has gotten 10x better. Im truly amazed

[@revdancatt]

Oh yeah, I saw the "Remaster" button pop up and was all, "hey, what does this do?", I was blown away by the result. Feels like I have less control over the final image though but I guess that will come with time.

[@KokohNicky]

Many, many thanks for making this video. It is very helpful. May your channel continue to grow.

[@revdancatt]

Thanks, that's very kind of you to say!

[@morgierwin6641]

You are glorious, this is exactly what I was trying to figure out! thank you!

[@revdancatt]

Yay, glad to been of help. Hope you're having fun exploring Midjourney, such a fun tool!

[@im_Dafox]

Good thing you pointed out the fact that MJ shouldn't be considered as something like StyleGAN and especially "why". It might not be obvious for newcomers when they see they can use images in their prompts. Clearly fell for it too when I first started 😅 Great guide for sure :]

[@revdancatt]

Yeah, I wasn't sure how it'd handle images when I first tried it. It could have gone either way depending on where they put their focus. Would be lovely which did it all, but you can understand why it doesn't.

[@Justiceguitar]

midjourney will actually paint in any style I used a picture of my friend and had it make a leonardo da vinci painting with her in it and it looked exactly like her it took some time to do it but can be done

[@ВеликийИнквизиторЮтуба]

Hey Dan. Cool video!

[@revdancatt]

Cheers!

[@pentacleman1000]

Hmmm, I’m a newbie, so it took me a while to follow this successfully. First of all, I thought that the image had to be on the internet somewhere with its own URL. But you uploaded yours from the downloads folder on your HD. That’s good, because I wanted to use an image that I had just created in another AI art program. Originally, you had lost me at 2:09 when you said “sending them up”. I managed to get to the point with the eye, pencil and trash can icons, just like you showed. But did not understand what you had done to “send them up.” I eventually figured it out, just put the curser in the text box and hit return. Okay. Now I have the URL. So I entered the URL at the prompt with dash dash iw 5. I got images that were a lot different than the original, with only one identifiable feature carrying over. I tried adding the original text prompt after the URL. It looked like it ignored the sample image and just used my prompt. You can’t put the dash dash iw 5 after the URL and then follow it with a prompts, can you, it has to be at the very end. And did you just make up a seed number? I thought you had to DM the bot to get a seed number (another thing I have not managed to do successfully yet). What would have happened if I had entered the seed number from the other AI program that made my sample image? Would MidJourney understand that and be more accurate to my image? Pretty complicated.

[@michaelandremovies]

Yes tutorials should be clear and precise explaining every step.

[@LizzieMayne]

HI I had the same problem it's only when I hit return as you said that it worked, thank you

[@atindrabiswasDP]

Thanks for this tutorial . It was needed

[@revdancatt]

Glad it helped 😁

[@dudemandudebro4235]

Thank you! This was very helpful and well explained

[@scottrandall524]

I was hoping you were going to show an ai with a pen plotter vector output

[@revdancatt]

Funnily enough I just do get ChatGPT to output SVG to plot. It was terrible!!! Needs so much more work. We'll get there.

[@Adrischa]

Usefull info. Thanks. Have been playing around a lot with MidJournes. Great fun

[@revdancatt]

Yeah, it's such a good time, glad you're having fun with it.

[@geroscrittore537]

You are amazing! Really really thank you! Fast and clear.Perfect!

[@revdancatt]

Thanks! Always trying to get the important info in that first 30 quick overview. Terrible for growing a KZbin channel, but much better for the viewer, lol.

[@geroscrittore537]

@@revdancatt maybe a tutorial on how to use different camera angles to get different perspectives would be interesting!!!

[@game_cropped4452]

Great video my guy, im just getting started and this helps a lot.

[@revdancatt]

Happy to help! Hope you have good fun creating some cool images.

[@Maikaz]

Thanks so much, this really helped me!

[@yuhanghe4666]

thanks so much Dan, this helps a lot of how to refer a image, it is really great tutorial.

[@revdancatt]

Glad it helped! Got it made in the end.

[@StevesMakerspace]

Another great video. Thanks, Dan! I highly recommend people read through Midjourney's guide before using up all their free tries. And don't bother trying to do people.

[@revdancatt]

Thank you, I blame the dentist painkillers for that! Although that doesn't excuse all the other typos I make. Glad I didn't mess the one up in the video though!

[@xkloveshin]

thanks for your tips!it is really useful!

[@revdancatt]

Always happy to help.

[@timbailey7646]

really enjoying your journey with mid journey!

[@revdancatt]

Hey thanks Tim, it's a weirdly interesting thing that's going to cause both anger and delight for artists and illustrators.

[@timbailey7646]

@@revdancatt I did the trial and found it wonderful!

[@revdancatt]

Awesome, I'm using it pretty much every other day. It's allowed me to go back to older artwork I'd given up on as being too small and given it a new lease of life, so that's nice.

[@homemacai]

Awesome explanation thanks!

[@goth_ross]

While im not a huge proponent of this tech yet. ( TBD) I certainly love your whole vibe and beard of course my brother in arts!!!

[@revdancatt]

Haha, thanks. I suspect it's something we'll have to learn to live with. Artist weren't fans of photography when it got popular, seems this is similar.

[@goth_ross]

Its a mixed bag. ( full of cats ). =)

[@revdancatt]

Yup, all the best bags are full of cats!

[@theartoftk1]

this finally helped me understand this. Could you speak to how to weight multiple images?

[@revdancatt]

Hey hey, you can't weigh images separately from each other. The image weight command `--iw`affects all the images equally. Which is a shame. I guess you could pull in one image a couple of time (so you feed it in once, and then feed it in again with the result of the first time and repeat as many times as you want), then pull in the second image. But until we get finer controls you just have to experiment and see what happens.

[@theartoftk1]

@@revdancatt yes someone on the discord suggested copy pasting the url multiple times and that does seem to work! appreciate the video!

[@gabogaloga]

Very nice video, sir. Keep it up! Thank you for this helpful content. 😁

[@revdancatt]

use '--iw' to adjust the image weight Full user manual here: midjourney.gitbook.io/docs/user-manual

[@infographie]

Excellent

[@gylesferrao6277]

Great video!

[@cheeseballzzz1601]

Thanks Dan! Super helpful video!

[@revdancatt]

Thanks, glad you found it helpful.

[@EddyRice]

Do you know how to take a video ofa person moving and replace seeing the human form with seeing a background image like a forrest or a waerfall etc.? I have cyberlink power director 19.

[@ScholarsonScreen]

Thank you for the video. very helpful.

[@heartshinemusic]

Thanks Dan!

[@revdancatt]

Happy to help

[@alissonagroo5061]

so if I take a picture of myself and want to do an illustration with angels, can I do it?

[@SysterYster]

If I add an URL, nothing happens but me getting an error message telling me it's an invalid link, or invalid command. :(

[@mundelator]

This is very helpful, thanks !

[@revdancatt]

Hey, anytime!

[@JeremyFrandsen]

Do you have any StyleGAN services you recommend?

[@eles8633]

thank you!! super helpful!

[@revdancatt]

Hope it helped, and hope you're having fun with Midjourney!

[@TheFoel]

Cannot use --version 4 with only a single image prompt. Please add another image prompt, or a text prompt. show me this error in desktop

[@ateliercreosote4589]

Hi Dan, you are a very good teacher thank you so much for sharing your knowledges. Last week i saw the work of a designer on Midjourney and i wonder how he did to be able to control effects in midjourney only on the faces of the people in the photography without retouching anything in Photoshop. Do you have an idea how the magic is done?

[@revdancatt]

That sounds amazing, no idea how the magic is done, but I'm going to have to dig in and figure it out. I know some people get good results by asking for variations over and over, only accepting the results when it's closer to what they want, and asking for variations on that. Allows people to slowly drill down randomly but precisely. Perhaps it's that.

[@romenpheonixx7005]

GOOOOD LORD!!! THANK YOU!!! YES IM YELLING!!!😂😂😂🤣🤣🤣

[@revdancatt]

Hahah, IKR, seems like it should be such an easy thing, but hard to find out anything about it. Hope you're having fun on your Midjourney adventures.

[@reduXstereo]

Experimenting with A.I art is like commissioning an arrogant artist who does the opposite of what you commissioned him or her for but the artist at least gets the subject matter right.

[@bejinariucatalin5380]

Hi Dan! I upload my pic in Mid Journey to get a linkedin profile pic but unfortunately mid journey returned me 4 different pics and not similar with the uploaded pic. Do you know what could be the issue? Thanks!

[@teddyoldenbourg6361]

What happens if the copy button doesnt appear?

[@geertreivers]

Works well!! DANKEEE

[@revdancatt]

Pleasure!

[@michaelandremovies]

Why does it always return this? Invalid link! Could not fetch image prompt! Received http status code 403

[@groovyraffraff5300]

Fantastic content

[@revdancatt]

Thank you!

[@HasnainChess]

Informative, and nice beard ❤

[@revdancatt]

Hah, I try my best on both counts.

[@seduttoridaincubo1722]

Something wrong: I always receive a code 404 error message, what's the matter? I'm sure I do all you say. Moreover, if I click on an uploaded imagine it does not open!

[@KiwiHawk-downunder-nz]

Is there a way to add two reference images? please

[@revdancatt]

Yup, you just paste two different urls into the same prompt and it'll grab both of them. Interesting for mixing things together.

[@bySterling]

Wow I didn’t know you could create this insanity via your phone! Curious shouldn’t your laptop discord show what your phone from the same Account?

[@revdancatt]

Insanity is right. Magic of video editing, the phone and desktop videos were taken at slightly different times and I'd done a couple of tests, so they're showing similar things but differently.

[@bpdude2277]

can I add several links for each element I want MJ to draw or only an image link that includes all the contents and features? I found it a little bit confusing please reply to me

[@revdancatt]

You can drop in as many links to images as you want. But Midjourney isn't very good at extracting and understanding the different elements of the images, so you may not end up with the results you are expecting or want.

[@bpdude2277]

@@revdancatt i got it, thank you!

[@raviswami9852]

I've followed this to the letter....but, it just isn't working for me, perhaps in MJ Version 3. No matter what I do, change the order of the image URL, the prompt, the seed value (which *does* do what it's supposed to do), the "--iw" weight value makes no significant difference to the resulting image as illustrated in your video or in the MJ user-manual. Any suggestions as to what I might be doing wrong ?

[@raviswami9852]

I've used it with some degree of success to match a pose (of a person for example) within a reference image but it's very much a roll of the dice.

[@revdancatt]

Yeah, very much hitting the re-roll button. Sometimes I start building up more information into the prompt I feed in. But it feels like there's a tipping point where if you do it too much the results start to get worse (or locked into a direction you didn't want to go in)

[@scuprial8842]

im new to this and Im wondering if I can use a picture of myself and then make it draw me as something else

[@revdancatt]

Midjounrey won't be very good for that, or rather it'll be entertainingly bad at it. I'm keeping an eye out for something that does a good job of that though.

[@Shakespeare1612]

Is there a software that specializes in color replacement? Maybe you'll say "photo shop." But I would love something more sophisticated that can dis regard the lighting, shadows, dirt, and corrosion and almost literally re paint something. Then I could give it a picture of my car a ask, "How would this look with a cobalt blue paint job?"

[@ImJustSayingYKnow]

Checkout Piximperfect's tutorial on High and Low Frequency separations in Photoshop

[@spiritualrealm555]

Hi, I dont have a subscription yet. Does that mean I cant use image prompt? I have been facing this problem for dew days. lot of people saying that i dont need to be on a subscription to do an image prompt and i just tried it with the process that you showed above. Its just not working for me. I keep getting error message :( :(

[@revdancatt]

You shouldn't need a subscription, the free version should work fine. What is the prompt you are trying to use?

[@calvinallen3424]

Is this for an iphone? I am on an android and I can't get it to copy the url for the image

[@revdancatt]

This is for the iPhone. Android has so many different ways of doing things that it's harder to show the _one way it works_.

[@samirshaikh862]

If I wanted to put my face to a character in an image created in midjourney what would you guys recommend? Prompt an image url with the desired look of my face next to descriptive prompt of the scene?

[@revdancatt]

I think that would be hard to do in Midjourney, it works best with faces it has lots of source images for, and there's no way to add your own face to the dataset. It's very likely to mangle your face with a single photo. What you're probably looking for are "Deep Fake" AI systems, which Midjourney isn't.

[@samirshaikh862]

@@revdancatt Thank you so much! I totally forgot about deepfake! What I'll probably do is use midjourney for the general scene and character (using a recognized actor for the character) then add deep fake to change the face and see how that works. If this works it would optimize production so much! This new tech is amazin!!! Thanks again for the reply!

[@tomi71]

When I upload an image from computer (desktop) to Discord Midjourney Bot. It will upload that image there with 3 symbols: eye, pencil and bin (trash). There is not an option to get the url (original image). I wonder what´s wrong?

[@tomi71]

Oh. Found it. Just needed to press Enter to really upload it there. Kind of a confusing as it already appeared there. Just kind of stupid or I am stupid. Yes. There is a strong possibility to that. I leave this here in case someone else has the samekind of issue as it is not apparent. Just watched that part of the video and noticed those three symbols and then he did something. I tried to eyeball what the hell is it, that turns it into a full image without those symbols.... yeah. Press Enter. Goddammit! lol

[@revdancatt]

Haha, yeah, sorry about that. I guess I could have entered a message with it and sent it to make it more obvious.

[@ChristerWesterlund]

Hi, is it possible to add some other text prompts also to the link for an image, or how do you combine image with text prompts …?

[@gwcstudio]

I feel so dumb. The examples I saw all used square brackets around the references, so I did, and kept getting 404s from the bot.

[@revdancatt]

Ah, not dumb at all, you go to the examples and follow them exactly you expect them to work. Not everyone knows what the [square brackets] are all about.

[@MushroomFleet]

excellent

[@revdancatt]

Thank you!

[@MushroomFleet]

@@revdancatt i love the effort you put into your productions! thanks again for the detailed explanations!

[@hudgaming_7022]

Hey man, great video! I do have a issue though, when I paste the url of a photo, I can’t use a aspect ratio of 9:16. It pulls up an error. Is there a way around that? Thank you!

[@revdancatt]

Hey, what's the error it gives you? If you're on your phone is it autocorrecting the two minus symbols to a single dash? Turning '--ar 9:16' into '-ar 9:16'? I've seen that happen a lot.

[@sshood317]

/settings

[@sshood317]

@@revdancatt the long - is the same

[@Noiretranquility]

What is the symbol for - - ? minus doesn’t work

[@revdancatt]

Sorry this is a delay. It's two minus symbols together -- BUT, if you're on your phone sometimes it autocorrects two minuses together -- into one long dash - That may be the problem.

[@Noiretranquility]

@@revdancatt thanks i found it , it was the auto punctuation setting on apple devices , cheers

[@susmoypaul5140]

How many images can be create in trial period?

[@revdancatt]

I think around 20 a month, which is enough to get you going.

[@ARTOFFICIVL]

i cant get this to work to save my life :(

[@o1ecypher]

how many pictures are you allowed to make

[@revdancatt]

With the free trial you can make about 25-30 to get a feel for it. On the $10/month level that goes up to around 200 (which also gives you the DM chatbot), I'm on the $30 level and haven't really hit into the limit yet.

[@podervermelho1]

I am quite sad with my results using the AI. I thought it'd be possible to create artworks of my own characters but the AI don't understand that my uploads are humans and when write "male character" along with the image for example, it creates a male character that has almost no similarity with the character that I've uploaded. And it is always just a face portrait even when I upload a full body image. As an artist I'm still way ahead of the ai when it comes to creating NON-Abstract humans in specific poses like "fighting". Honestly I am not sure if I am happy or sad about it, if anything, I'm sad for losing $30 on a sub that I'll never use 🤣

[@michaelandremovies]

keep getting 404

[@KungFuChess]

Great piece of tech but let's face it, this commoditises artwork to the value of a paper plate.

[@revdancatt]

I don't think so, depends on the artist and what they do with it. Art is already commodified, but not all of it.

[@ashmarbarbour]

omg

[@revdancatt]

😁

[@MrMarco7ify]

Dall-e is so much better

[@revdancatt]

Sure, Dall-e is great. But they do different things, Dall-e is much better at more phoorealistic images, Midjourney seems to be geared more towards artists.

[@dejangaletic7884]

Web junk

[@revdancatt]

Thank you, I will pass your feedback to the team.

[@shelldredd]

Nice video, thanks for the content, i love it.

[@revdancatt]

Thank you!

[@davidramio]

This was really helpful! Thanks

[@revdancatt]

Yay 🤘