This Face Swapper is MIND BLOWING! Roop Tutorial.

Рет қаралды 133,389

Жыл бұрын

This content is educational for academic research purposes only.
Roop guide for Stable diffusion where I show you how to change any face to a reference face.
Roop: github.com/s0md3v/sd-webui-roop
How to install Stable diffusion: • How to Install Stable ...
Stable diffusion for beginners: • How to use Stable Diff...
Prompt styles here:
/ sebs-hilis-79649068
Support me on Patreon to get access to unique perks! / sebastiankamph
Chat with me in our community discord: / discord
My Weekly AI Art Challenges • Let's AI Paint - Weekl...
My Stable diffusion workflow to Perfect Images • Revealing my Workflow ...
ControlNet tutorial and install guide • NEW ControlNet for Sta...
Famous Scenes Remade by ControlNet AI • Famous Scenes Remade b...
LIVE Pose in Stable Diffusion • LIVE Pose in Stable Di...
Control Lights in Stable Diffusion • Control Light in AI Im...
Ultimate Stable diffusion guide • Stable diffusion tutor...
Inpainting Tutorial - Stable Diffusion • Inpainting Tutorial - ...
The Rise of AI Art: A Creative Revolution • The Rise of AI Art - A...
7 Secrets to writing with ChatGPT (Don't tell your boss!) • 7 Secrets in ChatGPT (...
Ultimate Animation guide in Stable diffusion • Stable diffusion anima...
Dreambooth tutorial for Stable diffusion • Dreambooth tutorial fo...
5 tricks you're not using in Stable diffusion • Top 5 Stable diffusion...
Avoid these 7 mistakes in Stable diffusion • Don't make these 7 mis...
How to ChatGPT. ChatGPT explained in 1 minute • How to ChatGPT? Chat G...
This is Adobe Firefly. AI For Professionals • This Is Adobe Firefly....
Adobe Firefly Tutorial • Adobe Firefly Tutorial...
ChatGPT Playlist • ChatGPT

Пікірлер: 347

@sebastiankamph Жыл бұрын

Prompt styles here: www.patreon.com/posts/sebs-hilis-79649068

@LouisGedo Жыл бұрын

👋

@oaahmed7515 Жыл бұрын

Are you up to date sir?

@gameswownow7213 8 ай бұрын

they not free

@theguru143 Жыл бұрын

I've been waiting SO long for something like this. The trick with 2 simultaneous Roops running was awesome. Thank you

@AIMusicExperiment Жыл бұрын

Adding the second roop was such a simple but genius idea.

@ijayraj 6 ай бұрын

hey do you know how can i download styles i can't find it in description which he added at 2:59

@xaedowulf Жыл бұрын

Ive been waiting for this exact functionality to be able to replicate family photos with pictures of my wife and kiddos. Now we can find any pic we like and just swap ourselves into it. Very cool!

@ijayraj 6 ай бұрын

hey do you know how can i download styles i can't find it in description which he added at 2:59

@crazymoe8494 Жыл бұрын

Thanks so much for the multiple roop tip, I was playing around with one facce last night, now I will try it with many.

@sazarod Жыл бұрын

That double roop trick at the end is brilliant. Nice one! Thank you.

@sebastiankamph Жыл бұрын

Happy to help and thank you very much for your support, it is greatly appreciated! You absolut gem, you! 😘🌟

@vincedodge321 Жыл бұрын

multiple the roop extensions is something I have not thought about, and very useful. Hopefully the next version of roop will be having that included, like the way controlnet has it.

@johannvandebron986 Жыл бұрын

Nice! The Idea with the second Folder is great.

@arothmanmusic Жыл бұрын

Using Roop plus a textual inversion of the same person yields fantastic results.

@ronbere Жыл бұрын

add adetailer also and perfect results

@adrianmucci4860 Жыл бұрын

Thanks for sharing this information. Your really good in this man.

@frankschannel2642 Жыл бұрын

Awesome tip on creating TWO instances of roop inside A1111. THANKS!!!!

@KDawg5000 Жыл бұрын

That's a great tip for the 2nd roop. Thanks!

@sebastiankamph Жыл бұрын

Glad it was helpful!

@cybergarri Жыл бұрын

epic, spectacular result and consumes few resources, the bad thing is that I ended up getting fired at work every time I went through the computer took a picture with some of the coworkers, but it will be worth what we have laughed is priceless.

@dschroeder7354 Жыл бұрын

Thank You very much for the hard work you do for us,you do a great job!

@bentp4891 Жыл бұрын

Super trick with the double Roop. Thanks

@brianneal9204 10 ай бұрын

Anyone else have a problem when they use the command prompt and enter pip install insightface==0.7.3 ? I get this message: 'pip' is not recognized as an internal or external command, operable program or batch file.

@Mr.Every1 8 ай бұрын

you can fix . find the python folder looking for script folder . here u have the pip.exe

@amj2048 Жыл бұрын

Yeah I worked out the second Roop method too 🙂 I've haven't had time yet, but I want to test the same thing, but with using a symbolic link. That would only be for Windows though, not sure if other OSs have the same function. A symbolic link is where you tell Windows that the contents of a folder is actually in a different location. So for example if you did mklink /D c:\folder2 c:\folder1 then c:\folder2 is just an instant copy of c:\folder1 because they actually share the exact same location on the drive. When you would go into Explorer and look at the folder c:\folder2, you are actually looking at the c:\folder1 folder. The reason I want to test that is because you will be able to have a second copy of Roop, but when the first copy gets updated, the second copy instantly gets updated too, because the second copy is actually just pointing to the original Roop folder. This would be a great way to have multiple copies of Roop running, all of them being updated by just updating the original Roop folder and also because the same files are being referenced, you wouldn't be using any extra disk space. I'll let you know if this works 🙂

@wernerblahota6055 9 ай бұрын

Thank You very much for the hard work you do for us,you do a great job!👍👍👍👍👍

@cofiddle Жыл бұрын

im not gonna lie, that double roop trick is pretty cool. gotta give respect for living up to your word lol

@chariots8x230 Жыл бұрын

5:07 The example with the two characters is interesting. I want to create scenes with multiple different characters, so I need each character to have a custom face. I would love to see more videos on how to use AI to create scenes with multiple characters. Thank you for sharing 🙏🏻

@ijayraj 6 ай бұрын

hey do you know how can i download styles i can't find it in description which he added at 2:59

@RikkTheGaijin Жыл бұрын

I'm pretty sure in the future Roop will add the possibility to add additional faces without duplicating the folder, but for now that's a neat trick

@kemicalyemster 11 ай бұрын

The multiple face tech is genius!

@kliajesal4592 Жыл бұрын

This is a godsend for the images I'm trying to make with multiple specific people.

@RuneW0lf Жыл бұрын

I LOVE the multiple face trick!!!! you can also do this with img2img and inpainting the faces!!! (low denoise strength)

@sebastiankamph Жыл бұрын

Sweet! Yeah, I played around a little with that for the thumbail 🌟😊

@holhydro Жыл бұрын

This is a great addon! Thanks for recommendation. I always use A1111 for faceswap its a lot of fun, this is another tool that just make things a lot easier.

@fenrir20678 Жыл бұрын

Instead of adding a second roop you can also do this from img2img in steps. Like swap first face in txt2img, send to img2img and swap 2nd face from there. If you have more faces, just send the result to img2img again and repeat. :)

@jazmaan Жыл бұрын

That method still creates subtle unwanted changes to the other faces. Instead you can inpaint mask the face you want to change and then use Roop on it, leaving the unmasked faces undisturbed. I need to try this multi-roop method see if it leaves the other faces alone.

@Drokkstar_ Жыл бұрын

That's true, but this is a single pass straight from txt2img, which is way more convenient.

@ntvirusgaming1995 Жыл бұрын

thank you for your valuable tutorial. Keep up the good work

@sebastiankamph Жыл бұрын

Glad it was helpful!

@RSV9 Жыл бұрын

It works perfectly. I already had A1111 installed and I only needed to install Visual Studio and the extension. Thanks

@RSV9 Жыл бұрын

Is there any way to be able to choose the "weight" that the ROOP image can have in the final result ? For example, that the new face is identical to the one used in ROOP or that only has some resemblance.

@dreamzdziner8484 Жыл бұрын

Oh Man!! That's an insanely cool idea😀. Thank you

@cybermad64 Жыл бұрын

Late to the party here, but thanks a lot for that video tutorial ! This is soooooo much easier than doing a full lora training and the result are simply amazing.

@ijayraj 6 ай бұрын

hey do you know how can i download styles i can't find it in description which he added at 2:59

@SouthbayCreations Жыл бұрын

Amazing!! Thank you!

@sigmata0 Жыл бұрын

Amazing. Thank you 🙂

@danwpc 10 ай бұрын

Sebastian, thanks for sharing your work. All the installation instructions I've see thus far are for Windows users. I use a Mac M1. I'm assuming installing Visual Studio and running "pip install insightface==0.7.3" does not apply to Mac users (I've tried it, just in case!). Automatic1111 shows that I have roop installed, but it doesn't show up in my txt2txt or img2img windows. What am I missing?

@alucodev 7 ай бұрын

Roop isn't showing up for me on the web-ui, under controlnet. I've installed it and restarted as shown in the video, multiple times, but nothing has changed. I've made sure all the Visual Studio requirements are installed (python and C++) and the file inswapper_128.onnx is in the right location (\stable-diffusion-webui\models oop). Anyone know what could be going wrong?

@LeoAi-Artwork Жыл бұрын

Great instructor, his lectures have helped me create great works on my channel, I hope he gets more tutorials for everyone to learn. Thank Sebastian Kamph!

@Ultimum Жыл бұрын

I find it very handy! Now I can use the faces on my models and my AI generated art :)

@redbiglone Жыл бұрын

Hey man this was great, keep it up

@sebastiankamph Жыл бұрын

Thank you kindly! 😊🌟

@ksiobrga Жыл бұрын

Your videos about stable diffusion is so fckn GOOD man! THANKS

@sebastiankamph Жыл бұрын

Happy to hear that, thank you! 😊🌟

@nahiyanalamgir7056 Жыл бұрын

Thanks for explaining what you did and why you chose the presented settings instead of just asking us to blindly follow your steps.

@sebastiankamph Жыл бұрын

Glad it was helpful! 😊🌟

@SFzip Жыл бұрын

dual roop is simply brilliant

@DmitriySokhach Жыл бұрын

@sebastiankamph does it always trigger roop's face swap or randomly skips it and just generates normal image?

@Mocorn Жыл бұрын

Seb, when comparing faces like this it would be very helpful if you maximize the windows for those of us watching from the phone. Just a friendly tip.

@canadajasminestudio Жыл бұрын

确实非常好用！要是插件中把视频换脸的功能也加上就好了！ Amazing! 👍

@izakhatton588 11 ай бұрын

Any clue how to fix the error where the roop model isn't being detected in the roop directory?

@Music_Light_Show 9 ай бұрын

Great video, do you have a guide for just face swapping from one image to another?

@TheAiConqueror Жыл бұрын

Cool Wallpaper my Friend! 😁👍

@sebastiankamph Жыл бұрын

Sweet isn't it? 😁 Hope you're well!

@Cantetinza17 Жыл бұрын

Oh, that's cool. I want to try that.

@PuckStar 10 ай бұрын

how to sharpen or improve the new face? most of my face swaps end up that the image is all ok but the face is low res or something.

@JohnDoe-cb9ep Жыл бұрын

Oh, this trick is amazingly mighty

@philippeheritier9364 Жыл бұрын

Nice trick at the end thanks

@sebastiankamph Жыл бұрын

Glad you liked it! 😊🌟

@mfarshidfar Жыл бұрын

Thanks for another great video! I installed it but It dosn't show it on my Webui. Do you happne to know what might me the problem? I'm using mac with M1 chip. Thanks

@cyril1111 Жыл бұрын

ah nice, I knew this would be an helpful reddit post! Even tho the community downvoted... Ps: I also saw the comment to double the install on Nerdy yesterday ;)

@chrisbraeuer9476 Жыл бұрын

Its always the little things... Thanks for your videos.

@sebastiankamph Жыл бұрын

You bet, thanks for watching!

@heikohesse4666 Жыл бұрын

super cool 👍

@tokatoofficial Жыл бұрын

are there any ways that I can use roop in my google colab setup? since I wont have visualstudio in local

@mrBrownstoneist Жыл бұрын

the fast way is using img2img with denoising 0, you can do to any existing photo

@Jumpyfoot Жыл бұрын

I don't suppose you could explain this process in a little bit greater detail? How are you able to swap the faces using an Old_face.jpg and a new_face.jpg, just as an example?

@Chrono.. Жыл бұрын

@@Jumpyfoot In the img2img tab put the old_face, in the roop the new_face, and then set the denoising strength to a very low value (or 0) so that the original image is not changed and generate the image. Roop does the process only after the image is generated as it seems.

@mrBrownstoneist Жыл бұрын

@@Jumpyfoot put any photo eg. (old_face.jpg) in img2img tab, set denoise to 0 (means no change at all) and then put your new_face,jpg into roop extension.

@hatuey6326 Жыл бұрын

great tuto thanks but Does using the same image twice :1 in roop 1 and the same in roop 2 improve the resemblance ?

@arsov9885 Жыл бұрын

Many thanks

@sebastiankamph Жыл бұрын

Happy to help!

@phizc Жыл бұрын

I haven't tested yet, but I think you could make a symbolic link instead of copying the folder. That would save disk space.

@Yeeeeeehaw 10 ай бұрын

Great video

@sebastiankamph 10 ай бұрын

Thank you, tell a friend 😊

@JDRos 11 ай бұрын

What to do if the face becomes clearer than the body?

@BrunoFernandes_84 Жыл бұрын

Great videos, i learn a lot, but same times i m getting this error message when i try to generate some images, specially when im working with cntrol net: NansException: A tensor with all NaNs was produced in Unet. This could be either because there's not enough precision to represent the picture, or because your video card does not support half type. Try setting the "Upcast cross attention layer to float32" option in Settings > Stable Diffusion or using the --no-half commandline argument to fix this. Use --disable-nan-check commandline argument to disable this check. Time taken: 15.46sTorch active/reserved: 3464/3822 MiB, Sys VRAM: 5878/6144 MiB (95.67%) Any idea on how to fix it? Appreciate

@robbdeeze Жыл бұрын

Can this be installed on MacBook and used with a Google Collab folder stable diffusion? Thanks in advance.

@tukanhamen Жыл бұрын

I assume I wont be able to run this on my cheap Dell laptop and will need a beefy gpu with alot of vram right?

@jrobertsz71 Жыл бұрын

It installed easy and worked. The double roop was a good idea. Came back to it today and both roops are gone from my webui under the controlnet menu. Can't figure out what happened.

@user-gq2bq3zf1f 8 ай бұрын

Thank you always. I succeeded in changing my face through roop, is there a way to change my outfit and hairstyle naturally?

@MaisnerProductions Жыл бұрын

very cool

@fingerprint8479 10 ай бұрын

Sebastian, thanks for the tutorial. I followed the instructions, everything seems to have been installed ok but roop option does not apears under Controlnet. Any clue why it is not working for me? Thanks

@VanzSolo 10 ай бұрын

same.

@Aisaaax 10 ай бұрын

Does it copy the expression from the original or from the face reference? Can't really tell with your exanples. Also is it better to always feed a straight neutral expression portrait to roop or a reference that matches angle/expression?

@Ray2kay6HG Жыл бұрын

can this be used with warpfusion?

@tonywhite4476 Жыл бұрын

Roop is also available in the img2img tab but I can't figure out how to use it there.

@PSYCHOPATHiO Жыл бұрын

dude i love you, I wish i can give u a 100 likes lol. amazing stuff

@sebastiankamph Жыл бұрын

Wow, thank you! 😊🌟

@PSYCHOPATHiO Жыл бұрын

@@sebastiankamph I've been trying to do that for many years as I was facinated by that but I was limited to doing face cutouts in photoshop, with this I'm just creating amazing stuff.

@Luimar 11 ай бұрын

Amazing video! Do you know a way to install Roop in ComfyUI?

@zigma5706 Жыл бұрын

If you want to do img2img of real images, set denoise close to 0 and all is well! this roop thing works spectacularly vs effort required

@bos9824 6 ай бұрын

this was it!!! ty

@allunitedproduction 11 ай бұрын

hi, i used std mac (v 1.5.1) , ModuleNotFoundError: No module named 'insightface' and AttributeError: 'GeneralConditioner' object has no attribute 'process_tokens, who can help me, all my LORA models are no longer displayed and extension and extension ROOP does not work

@loadedvibes Жыл бұрын

Can it be used for live programs ?

@michallv Жыл бұрын

And here I was fixing the extensions code when all I had to do was copy the folder :D

@michaellightii1112 Жыл бұрын

i keep getting the 'pip' is not recognized as an internal or external command. any suggestions?

@user-qi7vx7nr5i Жыл бұрын

Is it ok to make yet another roop extension folder (roop3) for 3 added faces to a picture?

@ichibanb Жыл бұрын

I did it and it worked as long as I didn't try to upscale the face images any. 8 gigs of RAM is apparently not enough. Lots of out of memory errors.

@lucianodaluz5414 10 ай бұрын

Where did you install the insightface?

@juko8794 Жыл бұрын

How did u get the stable diffusion screen at 1:47??

@ariftagunawan Жыл бұрын

Can you use it with inpainting?

@DreamframerPhotography 11 ай бұрын

Thank you so much for th video! However, I see this error when I start my webui, which also causes the roop extension not to show up. I'm not sure if this can be fixed. *** raise HTTPError(req.full_url, code, msg, hdrs, fp) *** urllib.error.HTTPError: HTTP Error 401: Unauthorized

@mohamedmahdy8306 11 ай бұрын

did you find any fix for this?

@user-sk7qy3ct1d 10 ай бұрын

Roop is not shown in my UI

@86lngd25 Жыл бұрын

Does anyone know if this extension can be installed in Easy Diffusion local? any help figuring that out would be awesome thanks

@sebastiankamph Жыл бұрын

I doubt it

@sathien9158 11 ай бұрын

can this works on img2img?

@rencelgutierrez4977 6 ай бұрын

Hi, When I run pip install insightface==0.7.3 it gives me "Requirement already satisfied: insightface==0.7.3 . . . . . . ." what does it means? I didn't see the installation process

@user-ue5cb1gx7g Жыл бұрын

It seems like roop is always applied at the end and details added with lora are removed. Is it possible to make roop render before the loras?

@sebastiankamph Жыл бұрын

I don't know of a way. Roop has been pretty limited to photo-style and is quite lowres still.

@Nova-ul3vv Жыл бұрын

Its a great tool maybe you can guide US through upscaling process and show us how to preserve likeness as face restore with 1 tends have a negative impact on that

@nicolasheinemusic 7 ай бұрын

anyone having trouble with Error message " ModuleNotFoundError: No module named 'ifnude' "? Installed Visual Studio, installed "insightface", but it simply won't show and keeps spitting out the same error. Also did a complete re-installation, which didnt help...

@eladshitrit Жыл бұрын

Hey, trying to install, no matter what i do, when pasting the: pip install insightface==0.7.3, i get this reaction: 'pip' is not recognized as an internal or external command, operable program or batch file. Any solution for that?

@bornTwoFight Жыл бұрын

I am not an expert but had the same problem. Found this solution and it worked for me: python -m pip install insightface==0.7.3

@sideburnchef4957 Жыл бұрын

Why is his Available - Load From adress different than mine? I can't find Roop in the list either.

@JackyTheCoww Жыл бұрын

How do I do it the other way round? For example, I generate a photo of 1 person and I want to put their face onto another existing photo? Generated Photo Face => Existing photo. Does anyone know how to do this?

@raininheart9967 Ай бұрын

Genius

@ThisOrThat13 10 ай бұрын

Does the 0,1,2,3,4 work left to right for the characters? As I specifically wanted Emma Watson on the left and Daniel Radcliffe on the right. Watson would be 0 and Radcliffe would be 1?

@TonyStark-ik9em Жыл бұрын

The faces are only inserted as 128x128px - so the use sadly is a bit limited. This is working so good though that i hope someone will be able to motivate the original dev to at least train a 512 model for it.

@UrfanFahada Жыл бұрын

1. change to GFPGAN 2. choose upscaler 4x_ultrasharp 3. upscale scale 3x

@MrSongib Жыл бұрын

@@UrfanFahada ill try this. ty hmm I got artifact in my testing, idk maybe there some problem with my gfpgan.

@UrfanFahada Жыл бұрын

@@MrSongib try with inpainting too, denoise strenght 0

@korontos9038 11 ай бұрын

the og dev sold out and started whoring for midjourney by providing them a 512 model what a cuck

@korontos9038 11 ай бұрын

@@UrfanFahada codeformer is better tho

@Philson Жыл бұрын

Duuuuude. This is so useful!

@PureFxAI Жыл бұрын

good tick 😄😄

@BenaresMusic 10 ай бұрын

Your face looks always funny 🤭

@erin6692 Жыл бұрын

I've been checking your channel for this for a couple days. lol

@sebastiankamph Жыл бұрын

It's been Midsummer's celebrations in Sweden! Hope the wait was worth it 😊🌟

@erin6692 Жыл бұрын

@@sebastiankamph Oh no. I hope I didn’t come off as complaining! Lol. I know people have lives. I just only understand SD stuff when you explain it. Lol. And I am American and had no idea what Midsummer was so I went and read about it. Thanks for sharing!

@sebastiankamph Жыл бұрын

@@erin6692 I didn't take it that way at all. Really glad you're enjoying the videos and coming back for more 😘