#### Support me + get Rewards #### www.patreon.com/oliviotutorials www.youtube.com/@OlivioSarikas/membership #### Links from the Video #### Gradient Control: ua-cam.com/video/ZW1xuOB3OMk/v-deo.html Perfect Face: ua-cam.com/video/9UOdXLYwhYY/v-deo.html github.com/Gourieff/sd-webui-reactor github.com/Gourieff/comfyui-reactor-node visualstudio.microsoft.com/downloads/
In text 2 image , please try reactor + ip adapter face together, just use the same face input for both, u can can even use that reference image you used in your example in img2img. Use realistic vision 5.1 with cfg at 3.5 and use more basic prompts . You will be blown away by the quality and authenticness of your face. It will even do your full beard in all its glory with no extra prompting!
Something to check if you're having trouble with errors when running the extension - make sure you disable roop in the extensions tab before running reactor,
Ever tries the faceswap lab extension? I think it's even better. It can synthesize a face from a Hole batch pictures. I'm pretty sure you will love it.
Not a lot of people talk about face swap lab, when I next go to make a few more characters, I'm going to try and add some expressions to the check point
exactly, although with one reference picture they give the same (99.9% similar) result not very close to the original, but the batch can improve it by a lot
I'm stuck on the part where I can install Visual Studio Community 2022 but I don't see anything called Visual Studio Build tools 2022. Is it by another name instead?
After 1 hour of installing and troubleshooting, it finaly works! hahahaha, but after the first try, I got an NSFW image with the head of Mona Lisa hahahaha
I found and installed the Visual Studio community but am not seeing where the build tools are for VS edit: So for those of you confused on where to download the build tools it's not in the installer like he would have you believe, its in the webpage where you scroll down to find the build tools exe download
so i am getting reActor status running but i still don't have anything in the insightface\models folder. It download the inswapper_128.onnx file in the main folder (insightface) but it didnt download the models. How do download the models? Never mind, A1111 downloaded it when I used the extension during generation.
Been using reactor for a month, it runs like a charm. But once reactor is used, it basically slows down every other operations like simple txt2img, img2img. Only way i found is to restart a1111 to fix it.
hey, awesome video! My insightface folder doesnt appear, and you dont really explain how you fixed that issue. Automatic1111 works, and in the cmd window when I launch it it even says that Reactor is working, but when it opens I don't have the extra options when I scroll down.
yes, but you need more than just the model. It's like saying you use android, so why are there different phones. because different developers still bring different ways to use the same thing. for example i can't remember that roop ever tracked the age and gender of a person in the photo
@@OlivioSarikasif roop developer was still working he would have added it. the co developer is now working on facefusion and face fusion has age gender etc built in. but it sucks in detection as roop algorithm is better
I also had problems with getting reactor to work in Comfy so I reverted back to a1111. Swapping was I was doing in my project so I couldn't go ahead with learning Comfy. Hope that gets sorted.
unfortunately I don't get Visual studio build tools 2022. seems to be a new version without that feature. reactor doesn't work for me. Maybe that's the reason.
The current model in Reactor can't be used for commercial work. Does anyone know if there is an alternative model to use for commercial use. Potentially at a cost?
Olivio, mate, awesome video, can you please do a tutorial on the video 2 video on stable diffusion? Animatediff and others , which one is better please
he said both install methods did not create the insight face folder, so where did he get it and why did he jump to saying after it gets done installing...im lost with this guy
haven't really been successful installing reactor and having it work for me, it always gives me errors after trying to generate an image, iv updated visual studio community, updated the pip p, and even tried installing and restating but no luck, one of the lines in the errors is TypeError: 'NoneType' object is not subscriptable what can i do?
I am having problems with objects over the face. Reactor doesn't overlap objects over or in front of the faces, it just draws the face over the objects. How to achieve this?
@@MABtheGAMEI should clarify, mostly I am using Faceswaplab as a replacement for dreambooth. The face is rendered at the same resolution as the image. So, in img2img, you are correct. Still looks better than what I got from Roop GE(reactor).
Don't know who else to ask, can you please make a video on Comfy UI inpainting at full resolution similar to how it works in A111, where it masks and renders a smaller section of the image, not the entire thing. Thanks man
I've been using reactor occasionally for some weeks now and it's very good. Especiall the version specially made for nvidia gpus. Yesterday I rendered around 8k frames from a small clip with the face of mine and the outcome is astounding. (atleast if the original face is looking towards the camera, otherwise it can distort sometimes)
Hi can you show us how to set up Comfyui ReActor because I am unable to get it to work even though I do see a folder showing it is uploaded but I can not see it or figure out how to fix the issue.
I found the solution, so on the txt to IMG page, go all the way down, there's a menu and click on "reload UI". After this it worked. When I used the reload UI link from the "installed" tab, it doesn't work. Try this
Glad to hear that it worked for you. I reloaded the UI and still is not showing up. I saw files like w600k_r50.onnx missing from the original gitHub page link. I don't know why but for some reason the only file after the installation that was showing in my folder was inswapper_128.onnx nothing else. No buffalo_l folder and the rest of the modules were all missing. why? I have no idea. I had to search the web and find all the different modules and manually download them and put them in the proper place and still is not working. I am running the latest version of SD as well. @@mariomills
Thank you for video! One question: why you still using xformers with torch 2.0? “-Opt-sdp-attention” is faster and use less vram than xformers on RTX card.
thank you a lot Olivio, we are all doing good progress ... how can we got the same pass of what you have? I want to test on google colab, I did the test, works but I think I got very basic interfase, do you have videos how to create or work with /comfyui ?
PLEASE, help me to install Face Swap in PaperSpace. I don't have the machine to use A1111 locally therefore installing Visual Studio does not make any change in the PaperSpace environment.
I cant get it to look anything like the photo - used a high quality clear full on face pic I took, and no matter how I use it it just looks like a vague-ish version
@@martinwoodfilms Yes. No need for the MS Visual Studio install. Macs already have the right bits for this to work without that. Jst follow the install instructions, and works fine..
Any advice on how to troubleshoot the installation? I cant get reactor to launch in A111. When I start the server I get an error message to restart the server, so I restart it, and the same error occurs again. As a result reactor is never available.
FileNotFoundError: [Errno 2] No such file or directory: 'C:\\Users\\xx6\\Documents\\A1111 Web UI Autoinstaller\\stable-diffusion-webui\\models\\insightface\\inswapper_128.onnx'
I manually create Insightface folder in the '..webui\models' folder and copy inswapper_128.onnx file and it works. I am still missing subfolder named bufallo_l and dozen of .onnx files inside. Please someone put link on that folder to download manually
It's much more better not to install VS-Studio. In installation instruction to Reactor there is another way, much more simple and in this case you save your disk space.
I prefer to stay on FaceSwapLab because I completely bricked my A1111 (I had to install again the whole SD with A1111) when I tried to install Reactor one week ago. Maybe I did something wrong but I prefer to no taunt the devil. Is Reactor really better or is it the same ?
@@OlivioSarikas Tutorial is well made, nothing against. It's better to point out both and make sure that "everything should be fine". Someone who really want go to with details will able to find out that build tools with specific options is enought.
Seems like it ought to be possible to make a blender model that outputs openpose poses. Are you aware of anyone doing this? It outght to be a pretty easy geometry to rig.
HALP! I've tried in multiple ways to install this extension to A1, but nothing seems to work. I have it checked within A1 extensions interface. The CMD gives no errors. But I can't see the drop down menu item in the main interface. One thing I discovered just now is that within the Insightface folder you mention, I do not have all the models you talk about. It is empty, apart from the file "inswapper_128.onx". How do I get it to download and fill that folder with those models?
I sorted it by reinstalling insightface as per this guide: VIII. (For Windows users) If you still cannot build Insightface for some reasons or just don't want to install Visual Studio or VS C++ Build Tools - do the following: Close (stop) your SD WebUI Server if it's running Download and put prebuilt Insightface package into the stable-diffusion-webui (or SD.Next) root folder (where you have "webui-user.bat" file) From stable-diffusion-webui (or SD.Next) root folder run CMD and .\venv\Scripts\activate Then update your PIP: python -m pip install -U pip Then install Insightface: pip install insightface-0.7.3-cp310-cp310-win_amd64.whl Enjoy!
you can do that by changing the text prompt. look for online prompt lists for different hair styles for example. Also you could use controlnet. that does require a photo, but you could just take a photo of yourself and use that as a tracking in openpose without it actually taking the shape of your face
#### Support me + get Rewards ####
www.patreon.com/oliviotutorials
www.youtube.com/@OlivioSarikas/membership
#### Links from the Video ####
Gradient Control: ua-cam.com/video/ZW1xuOB3OMk/v-deo.html
Perfect Face: ua-cam.com/video/9UOdXLYwhYY/v-deo.html
github.com/Gourieff/sd-webui-reactor
github.com/Gourieff/comfyui-reactor-node
visualstudio.microsoft.com/downloads/
👋
In text 2 image , please try reactor + ip adapter face together, just use the same face input for both, u can can even use that reference image you used in your example in img2img. Use realistic vision 5.1 with cfg at 3.5 and use more basic prompts . You will be blown away by the quality and authenticness of your face. It will even do your full beard in all its glory with no extra prompting!
Now we know what Olivio's mugshot might look like if he ever got in trouble with the law...🤣
ai gangsta for life 😂
ReActor and Faceswaplab are the goats at the moment.
Something to check if you're having trouble with errors when running the extension - make sure you disable roop in the extensions tab before running reactor,
Getting this work had to go through lot of pain! i somehow finally made it work
same. took me 1 hour to figure out how to get it to run. bit of a mess to install
Did you get the issue of Reactor not found in comfyui as well ? and can you recall what was the solution if so ?
Been using this for a few weeks now and it's amazing how it can make your face from a single image. I've only tested with sdxl and it's flawless
I think earlier and better is to use a controlnet face swap with a comfy UI ...
ipAdaper face swap works even with videos.
It creates a 3D model of the source and target faces and then tries to match one with the other so the orientations are the same. It's pretty amazing.
Ever tries the faceswap lab extension? I think it's even better. It can synthesize a face from a Hole batch pictures. I'm pretty sure you will love it.
Lotsa work involved but good yes. Im sure its animation focused only, havent used in a while.
Not a lot of people talk about face swap lab, when I next go to make a few more characters, I'm going to try and add some expressions to the check point
exactly, although with one reference picture they give the same (99.9% similar) result not very close to the original, but the batch can improve it by a lot
cant find a tut for it! i just installed
There is on youtube. I guess in Italian. I keep saying Facelab is way better than any other. Even better than Lora for people.@@ParvathyKapoor
Faceswaplab is better to be honnest :)
I can't find Visual studio built tools 2022 in the installer
its on the website itself, scroll down to "all downloads"
I'm stuck on the part where I can install Visual Studio Community 2022 but I don't see anything called Visual Studio Build tools 2022. Is it by another name instead?
Click the Available tab and you'll see it listed. You only have Community installed so the Installer is showing only that.
hi Olivio.. I love you 😍 you always make very easy to understand tutorial you're the best
After 1 hour of installing and troubleshooting, it finaly works! hahahaha, but after the first try, I got an NSFW image with the head of Mona Lisa hahahaha
I found and installed the Visual Studio community but am not seeing where the build tools are for VS
edit: So for those of you confused on where to download the build tools it's not in the installer like he would have you believe, its in the webpage where you scroll down to find the build tools exe download
thank you. I confused that in my video
Thank You!!!!
Thank you for being the BEST!
so i am getting reActor status running but i still don't have anything in the insightface\models folder. It download the inswapper_128.onnx file in the main folder (insightface) but it didnt download the models. How do download the models?
Never mind, A1111 downloaded it when I used the extension during generation.
I didn’t like roop first time I tested... These methods are better, thank you again!
same as with roop, reactor just doesn't show up after installing
It is great - Thank you for the video.
Thank you too!
Been using reactor for a month, it runs like a charm.
But once reactor is used, it basically slows down every other operations like simple txt2img, img2img.
Only way i found is to restart a1111 to fix it.
ur my master seriouly i learned lot with ur videos love u from india
I test this yesterday, I had good results. I tested too the facelab and was pretty good too. Is nice because of hard angles it can get.
hey, awesome video! My insightface folder doesnt appear, and you dont really explain how you fixed that issue. Automatic1111 works, and in the cmd window when I launch it it even says that Reactor is working, but when it opens I don't have the extra options when I scroll down.
The minimum and maximum effect are reversed. 1 is maximum and 0 is minimum even though it says the opposite in ReActor.
fun! thanks Olivio
Finally I found compared video
The results look the exact same as roop to me and from the other comments they use the same model underneath so they are in fact the exact same.
yes, but you need more than just the model. It's like saying you use android, so why are there different phones. because different developers still bring different ways to use the same thing. for example i can't remember that roop ever tracked the age and gender of a person in the photo
@@OlivioSarikasif roop developer was still working he would have added it. the co developer is now working on facefusion and face fusion has age gender etc built in. but it sucks in detection as roop algorithm is better
@@OlivioSarikas it does, but in the background.
I also had problems with getting reactor to work in Comfy so I reverted back to a1111. Swapping was I was doing in my project so I couldn't go ahead with learning Comfy. Hope that gets sorted.
unfortunately I don't get Visual studio build tools 2022. seems to be a new version without that feature. reactor doesn't work for me. Maybe that's the reason.
I think FaceSwapLab may be better than reactor but not sure yet... the model building feature of faceswaplab seems to create better results
Works like a charm.... ✔
The current model in Reactor can't be used for commercial work. Does anyone know if there is an alternative model to use for commercial use. Potentially at a cost?
Olivio, mate, awesome video, can you please do a tutorial on the video 2 video on stable diffusion? Animatediff and others , which one is better please
he said both install methods did not create the insight face folder, so where did he get it and why did he jump to saying after it gets done installing...im lost with this guy
you make a new folder and call it "insightface"
haven't really been successful installing reactor and having it work for me, it always gives me errors after trying to generate an image, iv updated visual studio community, updated the pip p, and even tried installing and restating but no luck, one of the lines in the errors is TypeError: 'NoneType' object is not subscriptable what can i do?
I face the same issue
I am having problems with objects over the face. Reactor doesn't overlap objects over or in front of the faces, it just draws the face over the objects. How to achieve this?
They all use the same model (insightface 128), so how are they better?
Faceswaplab has built in upscaling.
@@GunnyNinja that upscaling is applied to whole image, not only the face, so the face remains low resolution as compare to whole body/image
@@MABtheGAME Not in my use.
@@MABtheGAMEI should clarify, mostly I am using Faceswaplab as a replacement for dreambooth. The face is rendered at the same resolution as the image. So, in img2img, you are correct. Still looks better than what I got from Roop GE(reactor).
visual studio link doesn't work, and also if installed the two options are not able to be found to be installed.
Don't know who else to ask, can you please make a video on Comfy UI inpainting at full resolution similar to how it works in A111, where it masks and renders a smaller section of the image, not the entire thing. Thanks man
So its still Roop? Using the 128x128 lowres insightface model. Ill pass.
I've been using reactor occasionally for some weeks now and it's very good. Especiall the version specially made for nvidia gpus. Yesterday I rendered around 8k frames from a small clip with the face of mine and the outcome is astounding. (atleast if the original face is looking towards the camera, otherwise it can distort sometimes)
how do you get high quality on reactor? I've just installed it and the resolution of the output is lower than the original image which is 1024*1024
1:58 Any idea why I don't have an A1111 folder? I have a stable-diffusion-webui folder. Is that the same thing?
yes it's the same thing
Sad to hear Roop is being discontinued!
I've got it "working" but the results are awful... I must be doing something wrong
Thank you. I tried to install this several weeks ago without success. Manually adding the [insightface] folder fixed it for me. =)
awesome, glad i could help
Could you share a link for the insightface folder?
it’s exactly same model as roop with same results
but roop isn no longer being updated. it was discontinued 2 months ago!
amazing
Hi can you show us how to set up Comfyui ReActor because I am unable to get it to work even though I do see a folder showing it is uploaded but I can not see it or figure out how to fix the issue.
Reactor just doesn't appear under control net even after checking for updates on the xtension and apply and restart UI ☹☹☹☹
same issue
I have tried many solutions. Nothing has worked out for me @@mariomills
I found the solution, so on the txt to IMG page, go all the way down, there's a menu and click on "reload UI". After this it worked. When I used the reload UI link from the "installed" tab, it doesn't work. Try this
Glad to hear that it worked for you. I reloaded the UI and still is not showing up. I saw files like w600k_r50.onnx missing from the original gitHub page link. I don't know why but for some reason the only file after the installation that was showing in my folder was inswapper_128.onnx nothing else. No buffalo_l folder and the rest of the modules were all missing. why? I have no idea. I had to search the web and find all the different modules and manually download them and put them in the proper place and still is not working. I am running the latest version of SD as well. @@mariomills
my reactor Running v0.7.0-b7 on Device but no "enable box" & not single box but 2box
ouh lala!!
How can I use A1111 and ComfyUI in the same portable sd folder?
Hello, I have a question. How can I use it functionally?
😊very cool
thank you
hi olivio: another great vid. how about one on using SD extensions Nextview and Reactor for face swapping video?
thanks, rob
Thank you for video! One question: why you still using xformers with torch 2.0? “-Opt-sdp-attention” is faster and use less vram than xformers on RTX card.
why wouldn't you show us how you fixed insightface....
I'm starting and can't make it work... It's not changing the face at all :(
To much hazzle for me. Is there no clicky clicky solution out there or a pro that can put this together in a working program?
What should I do if, when generating images, a long fringe covering the eyes results in the forehead appearing darkened?
Awesome video as usual! Can you give youre "prompts" please?
thank you a lot Olivio, we are all doing good progress ... how can we got the same pass of what you have? I want to test on google colab, I did the test, works but I think I got very basic interfase, do you have videos how to create or work with /comfyui ?
Love the rap!
Thank you 😍
hi i did everything right but i do not have the "enable"button in ReActor...do you have any ideas why?
What if there is an object in front of the face?
2:45 You literally say it was not installed but give no pointers on how to install.
Can this deal with glasses and beards and different expressions like Roop? And how you you rate it versus the Roop extension repo? Thanks!
Can you make on video on version 6.1? This seems quite different I can't follow instructions with 6.1
PLEASE, help me to install Face Swap in PaperSpace. I don't have the machine to use A1111 locally therefore installing Visual Studio does not make any change in the PaperSpace environment.
i cant get a beard to show up in the swap it shows up as like a 5 o'clock shadow. how do you get the beard to come in more?
I am trying to get reactor to show up, but it may i do not have insightface. How do i install Insightface?
I cant get it to look anything like the photo - used a high quality clear full on face pic I took, and no matter how I use it it just looks like a vague-ish version
Any method for a Macbook install of this? (ie is MS Visual Studio a pre-requisite, and will the Mac version of that work?)
Did you get working on Mac?
@@martinwoodfilms Yes. No need for the MS Visual Studio install. Macs already have the right bits for this to work without that. Jst follow the install instructions, and works fine..
Any advice on how to troubleshoot the installation? I cant get reactor to launch in A111. When I start the server I get an error message to restart the server, so I restart it, and the same error occurs again. As a result reactor is never available.
My insightface folder is empty, from where I can get it correctly installed?
I have same problem
FileNotFoundError: [Errno 2] No such file or directory: 'C:\\Users\\xx6\\Documents\\A1111 Web UI Autoinstaller\\stable-diffusion-webui\\models\\insightface\\inswapper_128.onnx'
I manually create Insightface folder in the '..webui\models' folder and copy inswapper_128.onnx file and it works. I am still missing subfolder named bufallo_l and dozen of .onnx files inside. Please someone put link on that folder to download manually
no answer ? 😞
No answers yet ;-;
Facefusion or Reactor? Which is better?
It didnt install the inswapper_128.onnx model for me where do I get the models?
I get an error each time I generate an image with reactor:
ReActor - STATUS - Working: source face index [0], target face index [0] | 0/1 [00:00
Good video. Do you have an equivalent video of ComfyUI gradient @ 9:10 but for A1111 please?
Can we select faces individually if we use image with multiple faces.
It's much more better not to install VS-Studio. In installation instruction to Reactor there is another way, much more simple and in this case you save your disk space.
i keep geting error: AttributeError: 'NoneType' object has no attribute 'get' in both SD and SDXL, can anyone help me on this?
any pure extension methods without installing distribution files on local drive?
Is it higher sampling steps can effect the quality of face swap on img2img
I prefer to stay on FaceSwapLab because I completely bricked my A1111 (I had to install again the whole SD with A1111) when I tried to install Reactor one week ago. Maybe I did something wrong but I prefer to no taunt the devil.
Is Reactor really better or is it the same ?
mine is generating a random person and then pasting my source face onto that. anyone know what im doing wrong? pls
Why the heck would I need visual studio, I would understand build tools, but VS?
You dont. Only build tools.
@@OlivioSarikas Tutorial is well made, nothing against. It's better to point out both and make sure that "everything should be fine". Someone who really want go to with details will able to find out that build tools with specific options is enought.
@@jacekfr3252 sorry, me reply was meant for a different comment. I deleted it
How do i get Visual Studio Tools 2022 there is no such option?
11:18 "Chinese letters here"
Me, a native: Hmm yes, truely some Chinese letters of all time XD
was this video taken down and then re-instated?
can I combine reactor with expresion loras ??
Seems like it ought to be possible to make a blender model that outputs openpose poses. Are you aware of anyone doing this? It outght to be a pretty easy geometry to rig.
can we use this on google collaboration
I followed the steps, even though it shows in extensions, it's not available as a tab anywhere. No idea how to find it
I had the same problem. I had forgotten to select the C++ tools when installing Virtual Studio
HALP!
I've tried in multiple ways to install this extension to A1, but nothing seems to work.
I have it checked within A1 extensions interface. The CMD gives no errors. But I can't see the drop down menu item in the main interface.
One thing I discovered just now is that within the Insightface folder you mention, I do not have all the models you talk about. It is empty, apart from the file "inswapper_128.onx". How do I get it to download and fill that folder with those models?
I sorted it by reinstalling insightface as per this guide:
VIII. (For Windows users) If you still cannot build Insightface for some reasons or just don't want to install Visual Studio or VS C++ Build Tools - do the following:
Close (stop) your SD WebUI Server if it's running
Download and put prebuilt Insightface package into the stable-diffusion-webui (or SD.Next) root folder (where you have "webui-user.bat" file)
From stable-diffusion-webui (or SD.Next) root folder run CMD and .\venv\Scripts\activate
Then update your PIP: python -m pip install -U pip
Then install Insightface: pip install insightface-0.7.3-cp310-cp310-win_amd64.whl
Enjoy!
Hello, I'm wondering is there any techniques to alter face without a source? I mean can we modify hair, eyes? Particular attributes only?
you can do that by changing the text prompt. look for online prompt lists for different hair styles for example. Also you could use controlnet. that does require a photo, but you could just take a photo of yourself and use that as a tracking in openpose without it actually taking the shape of your face
Reactor simply wont work on my comfyui Have tried llike 20 different solutions and its getting to the point of just giving up on comfy ui
have you tried instantID instead? ua-cam.com/video/wMLiGhogOPE/v-deo.html
@@OlivioSarikas I got some error messages on that as well trying out various workflows but might give it a go on a clean install :) thanks for the tip
@@svenhaheim join the discord and ask there with a screenshot of the error messages
@@OlivioSarikas I will if this latest thing I'm trying fails thanks :)
ReActor - ERROR - Age Detection: No face with index = 0 was found
ReActor creates a pretty ugly pixelized rectangle around the face in the final image - just look at the neck and shoulder at 8:10
probably a matter of updating the process
any solution for MAC users?
Any users of A1111 on a M2 Book managed to get this to work?