You Can Now INSERT Faces in MidJourney
Vložit
- čas přidán 13. 08. 2023
- Here's a quick video to show you how to inject your own face into MidJourney!
Discover More From Me:
🛠️ Explore hundreds of AI Tools: futuretools.io/
📰 Weekly Newsletter: www.futuretools.io/newsletter
😊 Discord Community: futuretools.io/discord
🐤 Follow me on Twitter: / mreflow
🧵 Follow me on Threads: www.threads.net/@mr.eflow
🐺 My personal blog: mattwolfe.com/
Resources From Today's Video:
github.com/deepinsight/insigh...
Outro music generated by Mubert mubert.com/render
Sponsorship/Media Inquiries: tally.so/r/nrBVlp
#AINews #AITools #GenerativeArt - Věda a technologie
The ones you think don't look like you, actually do look like you
💯%
My thoughts exactly 😂😂🤣🤣😂
Yeah, spot on!!
Yessss!!!
Like a lot!
Matt judges the similarity to his own face much stricter than other people's faces. All of the faces of Matt are more than good enough for the desired thumbnails. The workflow is so much faster.
My thoughts exactly!
I was thinking the same thing. A lot of the results were an immediate pass for me, but I guess Matt wasn't feeling too adventurous about his haircut lol.
LoL 😂@@christophergreen9928
Agreed. I thought the images looked pretty close. On a scale of 1 - 10, with 10 being a perfect match, I'd give it a very solid 8 for most of them.
@@christophergreen9928high standards are awesome when other people go thru all the trouble FOR US ahead of time!
Dude, you are so inspiring!!! I truly love your videos, your newsletters, when I get a break from work I watch all your cool stuff. It's the best time to be alive for people like us loving AI and you not only prove that but also add so much value to this subject. I really really appreciate you and your work you do for us. Just wanted to say that.
The face replacements actually look great! You should definitely use this method, to save you time!
Our big brains sure like to change things up! Great to hear about resetting and returning to your roots. Love the personal touches! 🙏
Congratulations on 10k in Discord buddy! 😎👊 That's a great place to be growing. Day 1s are your A-1s!
THIS IS HOW.A TUTORIAL IS DONE!!! THANK YOU!! You walk us through it, you don't cut to other crap!! Well done. You're a very good instructor. You try to cover as much as possible and you try your best to not miss any details.
I’ve been using this for months. It’s great! Now that you’ve shared it, I’m nervous the service will slow down as you have a large audience. 😂
Where can I get these service? & Pricing?
Me too 😂
Hello sir I can't see the three dot at the upper right side of my image that allows me to use the inswapper app. Neither can I even rightclick on the image when I give the prompt through my server. However when I go to direct messages i can see the three dot when i create images and even right click on the image. Please can u help?
🤯🔥 I was waiting for this!! Thank you
great video, I have been using Insight for about 2 months. It's been a game changer. It's really cool to watch old movies, and grab a face of someone obscure, and swap their face, and then make a lora or embedding of them. It used to be 50 ifree mages a day, no I think it's 21 or so. Still not bad. also it doesn't matter of they have glasses. I did my Mom, the only good image of her when she was young was a portrait with cat glass from the early 60's ,works like a charm.
I've been using this one for months.
Dragging your swapped image into Photoshop and using the liquify filter allows you to perfect the head and face shape.
Oh man! I thought you were the one who told me about this a couple months ago. I actually used it to swap out faces with a bunch of collaborators when we did a writing relay on CZcams, and we put this on our thumbnails as a marvel superhero team. Thanks so much for sharing this with those who don't know about it!
there is a trick where if you have a row of people. You can swap out them in order. just seperate them with a comma.
Wow, thank you for this tutorial. I had fun with it all afternoon with excellent results. Your channel is fantastic.
Been using this for a little while. Best one for this option
"not much changed about it" 🤣🤣 I mean you are hansome but not as chiseled as the first image lol
If you apply multiple times the INSwapper it actually looks better. I've been doing it and it helps me get closer to my subject. I usually apply 3 times.
I didn't even think about running it several times to refine it even more! Geez, how simple. Thanks!!
Yep, I discovered this trick as well and can confirm it does indeed help.
You mean we should swap upto 3 times on same image?
@@RobertLoPintohey bro please explain me too what you are saying I am not able to get
@@nadexvaibhav57892 Apply the swap 3 times in a row instead of just once.
✨PROTIP✨ If you include the parameter --iw 2.0 at the end of your MJ prompt when including your own picture as an image reference (aka image-prompt). It can help make the resulting images resemble you twice as much as it possibly can. By default this weight number is at 1.0.... so increasing it to the max of 2.0 does help.
Do you have a course? Paid or free.
I have done hundreds of face swaps and never thought of that. Thx, man!
@@shorerocks Anytime! :-)
So fun! Love the pics it created. Going to try it out for some thumbs on one of my channels. Thanks for showing us how to use it! :)
(0:14) That second image of you backpacking through the Kingdom of Arnor is SPOT ON!
How hilarious, I’m literally trying to do this as your video comes out
Looking forward to more videos like this Matt! And hahaha agreeing with everyone else that you know your face too well to notice but the Inswap images really actually do look like you.
Matt, I really appreciate your work and videos! I suggested your channel to teachers willing to know how to properly use AI in education. Best wishes from Romania! Adrian
Matt I love your videos. You are the best AI youtube channel by far.
Awesome I look forward to watching the new content!
This is how I’m creating all my AI videos for a few months now :) I love it!
thank you for this update Matt! 🙏
I love the idea of this. It needs a lot of work, but they'll get there eventually. It's nice to know it's being developed.
You can also generate an image with multiple people, though 2 or 3 works well, not huge groups. Then you can swap in several saved ids. So for example, I could generate me and 2 friends in front of the Eiffel Tower, then swap in our saved ids putting them in order from left to right, separated by commas. Thanks for another useful video insight Matt.
@leebebbington3413
Absolutely, you're spot on! With the Basic and Pro plans, you can indeed morph multiple faces in a single image, which is fantastic for group photos or scenarios like the one you mentioned with friends in front of the Eiffel Tower. However, just to ensure full transparency for everyone here, the Free Plan focuses on the most prominent face in the image. So for group photos or multiple face morphs, one would need to upgrade to our paid plans. Thanks for highlighting this and glad you found Matt's video insightful! 📸🗼🤓
The speed and accuracy of face swapping with mid-journey has definitely improved. Matt's meticulousness in judging the similarity to his own face showcases his dedication to perfecting the process.
This should help with cconsistent character images for graphic novels!
Oh love all ur updates ❤always look forward to ur videos
Wow... this works and it easy to do. Thank you Matt for sharing and being an awesome human being
Im super proud of you my man! don't let this channel overwhelm you, take things one at a time and don't skimp on family time. We'll be here, waiting, so take your time buddy!
Have I missed your process doing your thumbnails??? That would be cool to watch for sure. Love from Brazil and tks for all the great you’re putting out there 🙌
Thanks for the tutorials! Super helpful😊
THANK YOU MY GOD BEEN LOOKING FOR THIS KINDA OF THING FOR SO LING
You are so awesome!!! Thank you for all you do and sharing so much!!! I learn a ton from you and share your vids like crazy!!!
Thank you Matt for bringing the good stuff! 😎
Fire Man! Love The Content! 🔥🔥🔥
Yes the Insight Face Swap works well for photos, it doesn't work for drawings yet though. ALSO it has trouble with extreme mouth positions. Generally it works very well.
Great video, thank you Matt👍
Haa LUVED seeing our AI RingMaster as a Ken bro!!
Great videos Matt! Thank you!
Love this video plz continue to make more like these
Thank you buddy, much love ❤
These are the best videos!! Can you make one about your video editing workflow?
Interesting stuff. Not bad. Going to experiment thanks to your review. Thanks man.
You can also inswap an image multiple times. So if you inswap the face that has already been swapped you will get closer
Good stuff!
Doing a /blend of the Mid Journey image and your source image before using inswapper could improve the results.
Appreciate you!🙏🏾
You as Ken is killin' 😂
BTW all of the are really close!
oh wow I actually knew about this several weeks before Matt!! lol. In all seriousness it does work pretty well unless you have a really stylized image.
It's a great tool, a little work on getting the eye colour (they always look dark) right will go along way. I use photoshop, create 2 layers (make image same size), original as bottom layer and with a little work you can use the eraser tool to get some colour in the eyes. I do subscribe to it and worth every penny, has alot of uses.
its like everyday a new expectation. i love this knowledge
Looks really interesting.. I might have to steal some hours and explore the animation option they mention. 👍
I've been using it for a few months, it works well.
Liked this one definitely Matt🤘
You can use the prompt --iw 2 to indicate the weighting of the source image (2 being max)
Nice tutorial, thank you!
Thanks!
Been using this for a month! Thought I learned it from you? Oh geez. I should have filled you in. Ok. Next time.
Your amazing Mr. WOLF 🐺
good editing!
Thanks for sharing, we had no idea you could go premium!
Hi Matt thanks for your videos. Love them. Lots of help. Question.... can you make a video explaining how to get good quality caricatures style from a selfish picture using midjourney? Thank you
Matt, you need to select faces that are closer to the shape of your face. Try using a painterly style, a pencil sketch, or an illustration.
Thanks man, that was helpful.❤❤
Best AI content creator , hands down.
@matt wolfe, have you ever tried to use "/describe" to get prompts from midjourney for your uploaded image? with this you should get results, that are looking more like the image you want to merge. If you then add additional prompts of new scenery and use the swopface tool in the end you may improve your final result. Let me know if this works. ;)
Yeah definitely insight face swap has been amazing
Tutorials and tool breakdowns all the way! 🤩
very infromative, thanks for sharing
Your vids are great. I’m dying to turn my face into a simple cartoon in the style of Calvin and Hobbes.
I Made a video about this too , well done Matt 🎉
imagine this tech 6 months from now, amazing
In less than a year, look at how much AI has advanced. Absolutely incredible
We need feeding updates and tutorials...
Ai art
Ai music
Ai voice over
Ai video
Ai 3d models
Great stuff matt ;)
I'm curious if you took one of the first 2 and told it to in swap again if it would keep getting closer to you. Just multiple iterations to keep getting closer.
You're right about the face shape though, so it might not do much.
Thank you
I'd like to share this opinion with you. At the end of your video, you mentioned that you want to return to the type of videos you were producing on your channel. I wanted to let you know that, for me, I don't see a change per se. That is, to me, your channel signifies evolution. I believe that, as far as I'm concerned, the path and the variety of videos and topics you address on your channel doesn't warrant any changes. In other words, if you feel that your channel has changed, it's more because it has evolved. And that has happened because you're approaching it intelligently. Your decisions have led you to present this kind of content, which from my perspective is refreshing, always with your unique viewpoint that I value. It makes me feel like you're guiding me in this world of AI. Thank you so much for your contributions and congratulations.
Amazing Video 🤩🤩😻😻
Hit the fkn goldmine on this page. Thank you Good Sir!
You’re gonna make me activate my midjourney subscription again!
What are you doing here bro 🤣🤣
if you have a good pc you can use stable diffusion freely
midjourney subscription is wort it? how much is it?
@@korontos9038is stable diffusion better than midjourney for face swap and all bro
@@maxpet90 30 a month, or 10, but 10 doesnt get you much
This is good
This was such a helpful video, thank you! I'm with you the whole way... except when I upgrade my image, I don't actually get the right click option (or the 3 dots) made available for the image.
I can see them above in the InsightFaceSwamp bot messages, so I know I added it to my server, and I'm in the app... is it the starter plan I'm on?
What could I be doing wrong?
Good Video
I feel like you're a little late to the party on this one Matt, this came out a few months ago! I found the best method is to use an image prompt and prompt to generate a photo and then using insight face - just like you showed in one of your methods. Was definitely the best output for me
Matt, thanks for your consistently awesome content. I'm searching for the best way to face swap in SD - on a Mac M1. I downloaded Roop successfully (SD confirms that I have it installed) via the extensions tab, but it doesn't actually show up in my txt2img or img2img generators. The only tutorials I've found on installation (quite a few) only give instructions for Windows - which involves also downloading Microsoft's Visual Studio. Am I to conclude Roop is not really available for Mac? No one actually says that. If that's true, do you know of a way to face swap that is compatible with my setup? Thx.
Gotta love some hat brows 😂!
Another idea - you can upload a pic and have MJ give you 4 descriptions of that pic with /describe. And also use that as part of a prompt. Got that from another YT video.
8:40 lol 😂
You should tag Mr Musk 😂
yesssss
If you think about it they may not look like you but it's going off of the facial shape that's also defined by the shape of the body. If you were actually in whatever military or whatever then you might be surprised that those actually do look like you in those circumstances. Also they actually do look like you anyway even though they didn't change the hair and some things lol.
hello Matt. so is there a way to add your seeded characters to like a pic of 4 characters in a group
I love the Nyan cat! Also, the video was very helpful as usual.
You are amazing💕💕💕💕💕💕
Insightface is quite a name in the open source face recognition community. So I think (without having a look at their work in detail) they are tackling the problem from a different angle compared to like dream booth or other stuff. It's hard getting the model do create your face image cause you are unknown from training. You would have to manually figure out the semantic metrics the model is using to recreate a face looking similar to yours, which is a huge load of work and often requires changing of weights or huge back propagation efforts. This seems more like a non-learning approach. It's figuring out your face by face detection and the same with the face in the image and then merging these faces (like e.g. face morphing) until your identity is higher then the original identity, while preserving as much background information as possible. This leads to eyebrows on a hat. There is no semantic information for this, the semantics haven't been touched. Therefore you also cannot change semantics of the image without losing your identity face.
But compared to the training process this is much faster. So I really like it. I'll have a look at their stuff later, thanks!
weird, i thought it was you who told me about this months ago 😁
Welp the secret is out 😂 I’ve been blowing my friends minds with this for a while now 😂
Those are you lmao 🤣 they were so good I’m like Wht heck is he saying 😂
Awesome tutorial as always ...
Question!
I see that there are a limit of 50 credits on the basic plan... does anyone know if this is a daily limit or 50 free for lifetime credits?