@@latestmusic1575 in this link the script files will install latest version of rope pearl and rope development fork (Alucard24 fork) : www.patreon.com/posts/most-advanced-1-105123768
Hey Great tutorial, I've been using this tool for a while. You can also create embeddings by holding shift and selecting several images from the same subject, this greatly increases the quality of the swap.
This is really useful. Thanks. The only issue with DeepFakes/FaceSwaps is the face/head shape as it doesn't replace the entire head, and in many instances, it makes the subject look like but not so like the target face.
Hi there, I'm getting the following error when running rope.bat "RuntimeError: Found no NVIDIA driver on your system..." Do I need an nvidia video card for Rope to run?
i wish i had found this video before going through the install that i did! awesome work man! i had no idea what i was doing but got it installed and would have loved to have found your video first haha
If you don't use the 1-click installer, when you manually install Rope will it download the needed models like you showed in your video or do you manually have to install them from another site?
Wow wow wow. After reviewing several guides, this is by far the best one. Thanks for the thorough explanation of every bit and byte. Quick question: I noticed that the software is slightly cropping the images in the FACES pane and in the FIND FACES row. Particularly, it crops the hair. So, if the person that I want to swap is bald and the person that I want to swap-to has full hair, it keeps the baldness after the swapping (while I want to have the fill hair of the FACE model). Is there a workaround for it?
sadly i don't know any model that can change hair at the moment :/ they added some extra features you can try them. for example now our updated version on patreon has face expression changing working amazing
How does this compare to deepfacelab? I tried using deepfacelab and got stuck when the characters head is behind someone elses and the face glitches when the camera pans. The unobscured parts worked somewhat. Not entirely accurate but more training will fix that.
this is the most accurate one. but i can't tell it will work 100% in every scenario. Also on patreon we have a development fork installer as well with more features
it is here in the attachments : www.patreon.com/posts/most-advanced-1-105123768 Rope_V4. zip Rope_LandMarks_V4. zip Rope_Live_Stream_v4. zip - this one is most advanced version
I just learnt with pressing shift you can use multiple images of subject to make multi image using. It generates embeddings automatically thus may improve quality try it
I would like to understand better how to download the application you mentioned for generating videos and images. The explanation wasn't very clear to me, and I'm unsure if it's necessary to pay to access it. Could you please provide more details on how to install the application and if there are any associated costs? I appreciate your help in advance!
yes if you want installers it is at my patreon, monthly 5$, but you can use forever downloaded scripts. you can also use author original repo and install manually
@@harshdeep7015 64 gb ram , rtx 3090, 13900k, but i have installers for cloud you can use on massed compute if you dont have gpu. 31 cents per hour for A6000 = 48 GB GPU
in this link the script files will install latest version of rope pearl and rope development fork (Alucard24 fork) : www.patreon.com/posts/most-advanced-1-105123768
One problem with this one and for example faceswaplab (they use the same model inswapper_128) And as you mentioned they cannot correctly determine mouth changes. (as I remember roop didn't have such an issue) So it's almost useless for "talking" videos, because it just ignores some pronunciations and the mouth movements do not match the sound. Face parser is a strange fix that replaces the target mouth with the original mouth which immediately destroys the desired result. For not talking videos and static pictures, mouth is not a problem
Roop also had same issue. both using that old inswapper_128 but this guy made some improvements to the resolution. until someone else releases a better model than that 128 px this is what we got best
@@SECourses He uses some upscaling tricks but the results can often be worse and the 512 is close to unusuable quality. They also do not work well with the Strength value. They're just faster to process than the prior upscaling methods.
firstly thank you for putting out such amazing videos. I have been playing around with rope ai and it is fantastic. I do however have one particular issue. There is this one particular face that keeps turning up as a Black Box. Why is that? Is it like a security thing? I tried using a high res image but still the same and I can only assume it is to do with the face? Ps. this is just me fiddling around and whilst I don't care if I cant use that face, i am curious as to why does that black box show up?
@@SECourses I downloaded all the models, tried different ones and kinda also used another image, but same issue. Ahh it's alright I was just curious when that happens. I can only assume when the ai can't detect the face correctly, that's when I guess.
@@SECourses Ahh i just joined their discord.. apparently it is a known issue. The issue being with East Asian Faces. lol. Not really sure how to fix that one, but at least I know why it is happening. :)
here was no problem with the installation, and the program runs fine, but when I click [swap faces], no matter how long I wait, the faces don’t change. Could it be a performance issue? I have a 3050 graphics card, so I think it should be fine...
@@SECourses The program doesn’t freeze, and nothing shows up in the command prompt either. It's literally unresponsive. I have no idea what kind of error it is, so I can’t even search for a solution... 😥
Mac with cpu? Is gpu a must? It’s uses inswapper? Inswapper only support 128(very sad), what is the 256 and 512 in ui? That’s not inswapper I’ll know that. Enhancers and upscale?
Thank you very much for such a detailed video.I have a question to everyone, please tell me, I have a video promt where a person is standing and waving his hand, and there is another photo with a different background and another person, how can I make the photo turned into a video so that the background remains the same but the person from the photo completely repeats the movements and poses of the person from the video promt? what ai tool to use for this task?
@@SECourses I am using your google colab 1_click_deep_fake_for_free_by_SECourses.ipynb, but I found an error when trying the colab, the result is the face is shaking or swaying, please fix it
then I found this in your google colab : ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behavior is the source of the following dependency conflicts. pandas-stubs 2.1.4.231227 requires numpy>=1.26.0; python_version < "3.13", but you have numpy 1.24.3 which is incompatible. tensorflow-metadata 1.15.0 requires protobuf=3.20.3; python_version < "3.11", but you have protobuf 4.23.4 which is incompatible. tensorstore 0.1.64 requires ml-dtypes>=0.3.1, but you have ml-dtypes 0.2.0 which is incompatible. tf-keras 2.17.0 requires tensorflow=2.17, but you have tensorflow 2.14.0 which is incompatible. torchaudio 2.4.0+cu121 requires torch==2.4.0, but you have torch 2.0.1+cu118 which is incompatible. Successfully installed addict-2.4.0 albumentations-1.3.1 basicsr-1.4.2 customtkinter-5.2.0 darkdetect-0.8.0 facexlib-0.3.0 filterpy-1.4.5 gfpgan-1.3.8 google-auth-oauthlib-1.0. 0 insightface-0.7.3keras-2.14.0 lit-18.1.8 lmdb-1.5.1 ml-dtypes-0.2.0 numpy-1.24.3 onnx-1.14.0 onnxruntime-gpu-1.15.1 opencv-python-4.8 .0.74 opennsfw2-0.14.0 pillow-10.0.0 protobuf-4.23.4 qudida-0.0.4 tb-nightly-2.18.0a20240908 tensorboard-2.14.1 tensorflow-2.14.0 tensorflow-estimator-2.14.0 tk-0.1.0 tkinterdnd2-0.3.0 torch-2.0.1+cu118 torchvision-0.15.2+cu118 tqdm- 4.65.0 triton-2.0.0 wrapt-1.14.1 yapf-0.40.2 WARNING: The following packages were previously imported in this runtime: [PIL,numpy] You must restart the runtime in order to use newly installed version.
It would be cool a tutorial to combine this with LivePortrait to have a vid2vid reenactment, with the focus on maximizing quality, but also avoid the warping of anything that is not the face (which means the rotation in the driving video is not taken into account).
Can I ask, when I am recording the video to export, the VRAM seems full (red color) and its a lot slow. Should I clear the VRAM during recording. Still isn't clear on how to use this function?
I am using a 3060 12GB and when I turn on Restorer only and press the playback button, it slows down. Are there any possible improvements? It remains slow even if I set the thread to 1 and the swapper resolution to 128. Thanks.
How much vram your system uses before starting the app? 12 gb should work fairly well. Even 8gb works single thread. If you need speed on massed compute 20 threads works at best settings : m.th-cam.com/video/HLWLSszHwEc/w-d-xo.html
@@SECourses The task manager shows that the CPU is always at 100% while the GPU has room to spare. Is the CPU the cause of the slowdown? Of course, Python, Git, CUDA and FFmpeg have been tested. :'(
@@SECourses Intel i5-12400 is used. The VRM is 16 GB and only 50% is used when playing with restorer on. The VRAM in the top right-hand corner of the screen is also less than half. The video plays at normal speed for scenes where no faces are shown, but slows down when faces are shown and a face swap is performed...
@@mainevvv if you have used my installer and followed python tutorial i guess that is limit. if you are a gold patreon supporter i can connect your pc and check it out
Can anyone comment on the differences between ROPE, Roop, and Facefusion? Roop is great for photos. But not so great when things move in front of the face for video. Facefusion does better masking for video, but you can only convert a single pic at a time. I cant find any decent video showing the end results of a ROPE swapped video. Just a few frames less than 5 seconds of video.
ROPE has the best face tracking features. in my patreon post i included experimental development fork as well which has even more features. by the way entrance is 21 seconds that i changed with rope
hi! just subscribed your channed as member thouhg, used this program only for image changing, but the result is not really good enough , just wondered, is there another to get the better result, only for better image, not video. thank you! in advacne
Hi. Thank you for sub. Now we have 2 installer files. The landmark version has new experimental features. Have you tried it? Also to get better results your used faces need to be similar looking person. And can you tell me more info why bad image? I mean which part.
@@SECourses thank you for really quick response. firstly just used the file in the video, no idea of land mark, (bad image : compared to the face of the high-definition original image, the printed image is not sophisticated and the color is very awkward)
@@SECourses now testing new version with same setting~ as of now, GPEN1024 is making much better result than any other option!!! for making iamge , GPEN1024 is best , thank you for help
@@behrampatel4872 this can run perfectly run locally. in this video i run this locally on my pc. cloud tutorial here : th-cam.com/video/HLWLSszHwEc/w-d-xo.html
@@SECourses Thank you for the clarity. From now on I will assume there will always be 2 videos showing both workflows. Local and remote. Cheers and thanks for such detailed videos.
@@SECourses bro its paid But I thought it was free, so what if I make it myself? i known that its possible but it has more many files i can't be done it.so what i do bro can i buy this to you?
this has more features. we have a development fork installer on Patreon and it has way more features. sadly it wasnt ready when i published video so it wasn't shown. also this is uber fast with 20 threads on Massed Compute - 31 cents per hour
@@harsh_108 Massed Compute is a cloud platform like RunPod. you can watch this tutorial first 10 minute to see : th-cam.com/video/LeHfgq_lAXU/w-d-xo.html
I'm an English teacher. I used to jave an accent like your current accent😅 but you're the teacher here in youtube. Thanks for this amazing tutorial man. Really appreciated!
what is better out there? i just deleted FaceFusion, this tool here is lightyears ahead, on my 4090 the videos with faceswap almost runs in realtime, lol, incredible
Why make easy when you can make it complicated… when will someone make a single installation file and have a single single icon for the application? Should we ask to AI to do it because humans are so dumb to do it???
I'm going to tell you guys the truth about this very misleading video. This tool is no where near this good. Not even 20% as good as they try to make it out. It is easier than Roop and stuff, but does not produce better results. They, literally, cherry picked the example in the video just to make it look this good. It will not work remotely close to this well typically. Imagine making a competent video only to outright lie in your presentation of the results, claims of value, and title ruining an otherwise informative video.
@@SECourses hey but Q is for kaggle simple policy its not wrong if we import our own personal any AI tool notebook so like that way I think we can upload ts code too there?? simply
🔗 Rope Pearl Live and Alucard24 Installers Scripts ⤵
▶ www.patreon.com/posts/most-advanced-1-105123768
🔗 Requirements Step by Step Tutorial ⤵
▶ th-cam.com/video/-NjNy7afOQ0/w-d-xo.html
🔗 Main Windows Tutorial ⤵
▶ th-cam.com/video/RdWKOUlenaY/w-d-xo.html
🔗 Cloud Massed Compute Tutorial (Mac users can follow this tutorial) ⤵
▶ th-cam.com/video/HLWLSszHwEc/w-d-xo.html
🔗 Official Rope Pearl Live GitHub Repository ⤵
▶ github.com/argenspin/Rope-Live
🔗 Official Rope Alucard GitHub Repository ⤵
▶ github.com/Alucard24/Rope
🔗 SECourses Discord Channel to Get Full Support ⤵
▶ discord.com/servers/software-engineering-courses-secourses-772774097734074388
🔗 Our GitHub Repository ⤵
▶ github.com/FurkanGozukara/Stable-Diffusion
🔗 Our Reddit ⤵
▶ www.reddit.com/r/SECourses/
Bruh the link doesn't open bruh
@@latestmusic1575 in this link the script files will install latest version of rope pearl and rope development fork (Alucard24 fork) : www.patreon.com/posts/most-advanced-1-105123768
@mormacfey you can make your own ones and share lol
Hey Great tutorial, I've been using this tool for a while. You can also create embeddings by holding shift and selecting several images from the same subject, this greatly increases the quality of the swap.
Wow I didn't know this. I should cover this in next tutorial hopefully
This is really useful. Thanks. The only issue with DeepFakes/FaceSwaps is the face/head shape as it doesn't replace the entire head, and in many instances, it makes the subject look like but not so like the target face.
True. Because open source none of the models trained like that.
@@SECourseshow do you train. Face masks?
@@AGVenge i train via kohya or onetrainer here 1 big tutorial : th-cam.com/video/0t5l6CP9eBg/w-d-xo.html
Hi there, I'm getting the following error when running rope.bat "RuntimeError: Found no NVIDIA driver on your system..."
Do I need an nvidia video card for Rope to run?
yes unfortunately it is necessary. your best shot is using massed compute - 31 cents per hour
i wish i had found this video before going through the install that i did! awesome work man! i had no idea what i was doing but got it installed and would have loved to have found your video first haha
awesome. also on patreon we have now best forks as well alucard and live web carm
I was trying to search this today. and this made my understanding and knowledge much better. Great
you are welcome thanks for comment
This man is a legend. As a Patreon member, worth every penny and much much more, insane value.
Thank you so much for the amazing comment
If you don't use the 1-click installer, when you manually install Rope will it download the needed models like you showed in your video or do you manually have to install them from another site?
You have to manually find and download them
damn, this roop pearl is scary good and detailled
yep much better than others
Hello! i can't pay patreon just to get the link. Any other way to get the downloader Roop for windows? thanks!
Hi I accept crypto as well you can message me from discord : discord.com/servers/software-engineering-courses-secourses-772774097734074388
Wow wow wow. After reviewing several guides, this is by far the best one. Thanks for the thorough explanation of every bit and byte. Quick question:
I noticed that the software is slightly cropping the images in the FACES pane and in the FIND FACES row. Particularly, it crops the hair. So, if the person that I want to swap is bald and the person that I want to swap-to has full hair, it keeps the baldness after the swapping (while I want to have the fill hair of the FACE model). Is there a workaround for it?
sadly i don't know any model that can change hair at the moment :/ they added some extra features you can try them. for example now our updated version on patreon has face expression changing working amazing
@@SECourses WTF!? It took you 6 minutes to answer! Brilliant. Thanks for the answer :)
@@mitzmootz2974 you are welcome. i usually answer asap if i am on computer :D
Does this work with live camera? Or only saved videos
newest version works with live camera here : th-cam.com/video/whDt36YwEKQ/w-d-xo.html
How does this compare to deepfacelab? I tried using deepfacelab and got stuck when the characters head is behind someone elses and the face glitches when the camera pans. The unobscured parts worked somewhat. Not entirely accurate but more training will fix that.
this is the most accurate one. but i can't tell it will work 100% in every scenario. Also on patreon we have a development fork installer as well with more features
cant find the self installer
it is here in the attachments : www.patreon.com/posts/most-advanced-1-105123768
Rope_V4. zip
Rope_LandMarks_V4. zip
Rope_Live_Stream_v4. zip - this one is most advanced version
Can you also use multiple source face reference images together (different angles of the source person etc.) to improve the target quality? Thanks.
I think not. But not sure. You can render in multiple runs though. They have discord maybe you can ask there
I just learnt with pressing shift you can use multiple images of subject to make multi image using. It generates embeddings automatically thus may improve quality try it
I would like to understand better how to download the application you mentioned for generating videos and images. The explanation wasn't very clear to me, and I'm unsure if it's necessary to pay to access it. Could you please provide more details on how to install the application and if there are any associated costs? I appreciate your help in advance!
yes if you want installers it is at my patreon, monthly 5$, but you can use forever downloaded scripts. you can also use author original repo and install manually
have you seen a solution that also replaces and matches the target persona head / face shape entirely ?
nope i haven't seen any
It would be cool if we could use generative AI to perform this task
@@ysy69 well no open source yet. lets see if it get published
@@SECourses my understanding is that the top deepfake solutions are open source.
i dont know any open source that changes head
does it work on Mac?
nope. this requires nvidia gpu. but you can use on massed compute
It took me a while to realize you faceswapped Brad Pitt with Leonardo Di Caprio. :D
haha yes :D
So this model has a 512 pixels swapping model right? This is remarkable since things you can do with comfy ui for example only has resolution of 128
Still 128px model but the guy does some tiling algorithm to improve which really works.
Looks like UI is super simple
Yes ui working perfect
None of the faces load from the faces folder
check cmd you have error. also i just published a tutorial that shows how to fix all errors : th-cam.com/video/DrhUHnYfwC0/w-d-xo.html
Thanks a lot for this tutorial. What would you consider to be the best free faceswap tool for images today?
i think rope is the best
Possible in mac ?
sadly i dont know but i think it should support. i have cloud tutorial too : th-cam.com/video/HLWLSszHwEc/w-d-xo.html
how can we use it on mac please tell?
i dont have mac sadly. you can use on cloud : th-cam.com/video/HLWLSszHwEc/w-d-xo.html
Please, what is your system specs??
it works on CPU as well but i have RTX 3060 and RTX 3090
@@SECourses which specs is used in this video?? Ram cpu GPU??
@@harshdeep7015 64 gb ram , rtx 3090, 13900k, but i have installers for cloud you can use on massed compute if you dont have gpu. 31 cents per hour for A6000 = 48 GB GPU
@@SECourses how is acer predator Neo 16 i9 14th gen Rtx 4070 for these tasks?
@@SECourses I am planning to buy this month
great video, thank u
thank you so much. on patreon we have installers for alucard and live web cam versions as well - 1 click install
Does it work with AMD users?
i think not. either cpu or nvidia gpu but i can't tell for sure. i don't have AMD. but i have installers for runpod and massed compute works perfect
can i get updated link for script files ?
pls
in this link the script files will install latest version of rope pearl and rope development fork (Alucard24 fork) : www.patreon.com/posts/most-advanced-1-105123768
One problem with this one and for example faceswaplab (they use the same model inswapper_128) And as you mentioned they cannot correctly determine mouth changes. (as I remember roop didn't have such an issue) So it's almost useless for "talking" videos, because it just ignores some pronunciations and the mouth movements do not match the sound. Face parser is a strange fix that replaces the target mouth with the original mouth which immediately destroys the desired result. For not talking videos and static pictures, mouth is not a problem
Roop also had same issue. both using that old inswapper_128 but this guy made some improvements to the resolution. until someone else releases a better model than that 128 px this is what we got best
@@SECourses He uses some upscaling tricks but the results can often be worse and the 512 is close to unusuable quality. They also do not work well with the Strength value. They're just faster to process than the prior upscaling methods.
@@Zerod-rn3ye there are also some new developments being made into rope. i think it will be best 1 click app atm
@@SECourses 1 click wise? Yes, I agree.
@@Zerod-rn3ye yep
Curious on how this can be use to make face morphing as the Michael Jackson - Black Or White (1991) - First full photorealistic face morphing?
this can't do it. you need full fine tuning / training.
When I press record and play it doesn't do anything, any help? Thanks for video
that would likely to be caused by missing FFmpeg. can you verify that? please watch this tutorial : th-cam.com/video/-NjNy7afOQ0/w-d-xo.html
@@SECourses You are right, its working now but no audio haha, any idea? Thanks for your helpful video and response
@@mavericks2k811 well i would ask audio issue on their discord channel :D
this works whitout CUDA?
it works with cpu but i dont know others like amd
@@SECourses yeah it works! I was able to install it with certain parameters and it is working perfectly on the CPU. Thanks!
@@MarinaOnePiece awesome you are welcome
Can I output to sequenced images instead of an MP4?
i think not. but with ffmpeg you can extract frames so so easy : ffmpeg -i input_video.mp4 -vf "select='gte(n,0)'" -vsync 0 output_frames_%04d.png
firstly thank you for putting out such amazing videos.
I have been playing around with rope ai and it is fantastic. I do however have one particular issue. There is this one particular face that keeps turning up as a Black Box. Why is that? Is it like a security thing? I tried using a high res image but still the same and I can only assume it is to do with the face?
Ps. this is just me fiddling around and whilst I don't care if I cant use that face, i am curious as to why does that black box show up?
could be related to face model that can't accurately process that face i think. and thank you for comment
@@SECourses I downloaded all the models, tried different ones and kinda also used another image, but same issue. Ahh it's alright I was just curious when that happens. I can only assume when the ai can't detect the face correctly, that's when I guess.
@@SECourses Ahh i just joined their discord.. apparently it is a known issue. The issue being with East Asian Faces. lol. Not really sure how to fix that one, but at least I know why it is happening. :)
@@floyd1411 interesting thanks for the extra info
here was no problem with the installation, and the program runs fine, but when I click [swap faces], no matter how long I wait, the faces don’t change. Could it be a performance issue? I have a 3050 graphics card, so I think it should be fine...
can you tell me what do you see on CMD? initial freeze for few minutes is totally expected
@@SECourses The program doesn’t freeze, and nothing shows up in the command prompt either. It's literally unresponsive. I have no idea what kind of error it is, so I can’t even search for a solution... 😥
After reinstalling, it works fine now. Sorry about that
@@힝구-m7i great
@@힝구-m7i if you become my gold supporter i can connect your pc and hopefully fix it
Mac with cpu? Is gpu a must? It’s uses inswapper? Inswapper only support 128(very sad), what is the 256 and 512 in ui? That’s not inswapper I’ll know that. Enhancers and upscale?
i think it should work but i didnt test. this 512 is uses that model but does some tiling algorithm to make it like 512 and it really helps.
Do you think they gtx 1060 3gb would be able to handle this? Thank you.
Should be able to but could be slow if uses shared vram. Also opening face restore may make it slow. Give it a try
thank's for the update greate tutorial
you are welcome thanks for the comment
Respect 🔥🔥🔥
thank you so much for comment
can we run it on runpod?
maybe with a good template not that I know. but i just published massed compute many times better : th-cam.com/video/HLWLSszHwEc/w-d-xo.html
Thank you very much for such a detailed video.I have a question to everyone, please tell me, I have a video promt where a person is standing and waving his hand, and there is another photo with a different background and another person, how can I make the photo turned into a video so that the background remains the same but the person from the photo completely repeats the movements and poses of the person from the video promt? what ai tool to use for this task?
@@СтаниславБлизнюк-ф9у sadly there is not a great model for this yet. Best we have is LivePortrait which animates head but not body.
does it work on macbook pro?
Nope only NVIDIA GPUs. but we have massed compute 1 click installer and works awesome there. only 31 cents per hour
how can i select faces folder and find faces it's not shown even with refresh and why it's so slow ? please someone help
Check out your cmd logs. It shows there
@alinor5802 Have you solved it? Having same issue
why after I press the change face button, the face part of the photo is covered by a black object
inaccurate models. i updated my model downloaders. did you install latest versions from the patreon?
@@SECourses I am using your google colab 1_click_deep_fake_for_free_by_SECourses.ipynb, but I found an error when trying the colab, the result is the face is shaking or swaying, please fix it
then I found this in your google colab : ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behavior is the source of the following dependency conflicts.
pandas-stubs 2.1.4.231227 requires numpy>=1.26.0; python_version < "3.13", but you have numpy 1.24.3 which is incompatible.
tensorflow-metadata 1.15.0 requires protobuf=3.20.3; python_version < "3.11", but you have protobuf 4.23.4 which is incompatible.
tensorstore 0.1.64 requires ml-dtypes>=0.3.1, but you have ml-dtypes 0.2.0 which is incompatible.
tf-keras 2.17.0 requires tensorflow=2.17, but you have tensorflow 2.14.0 which is incompatible.
torchaudio 2.4.0+cu121 requires torch==2.4.0, but you have torch 2.0.1+cu118 which is incompatible.
Successfully installed addict-2.4.0 albumentations-1.3.1 basicsr-1.4.2 customtkinter-5.2.0 darkdetect-0.8.0 facexlib-0.3.0 filterpy-1.4.5 gfpgan-1.3.8 google-auth-oauthlib-1.0. 0 insightface-0.7.3keras-2.14.0 lit-18.1.8 lmdb-1.5.1 ml-dtypes-0.2.0 numpy-1.24.3 onnx-1.14.0 onnxruntime-gpu-1.15.1 opencv-python-4.8 .0.74 opennsfw2-0.14.0 pillow-10.0.0 protobuf-4.23.4 qudida-0.0.4 tb-nightly-2.18.0a20240908 tensorboard-2.14.1 tensorflow-2.14.0 tensorflow-estimator-2.14.0 tk-0.1.0 tkinterdnd2-0.3.0 torch-2.0.1+cu118 torchvision-0.15.2+cu118 tqdm- 4.65.0 triton-2.0.0 wrapt-1.14.1 yapf-0.40.2 WARNING: The following packages were previously imported in this runtime: [PIL,numpy] You must restart the runtime in order to use newly installed version.
it cant be fixed. if face changed that means that is what it can do best
not important as long as face changed
Bro old Collab method is not working giving modules import error
I will check it. But this app much better use on massed compute
I just tested old colab it works no issues.
@@SECourses bro give me the link now
@@ElevateWellnessYt colab.research.google.com/github/FurkanGozukara/Stable-Diffusion/blob/main/ColabNotebooks/1_click_deep_fake_for_free_by_SECourses.ipynb
@@SECourses tysm I am already subscribed ☺️☺️
bro can it make video perfectly , i means eating and other movements , sometimes i see when the person eating her face has broken
give it a try. we have now the newest version on patreon Rope NEXT. so many new features and improvements it got
@@SECourses ok but can i use this in cpu because i don't have gpu
@@MahammudHassanLimon yes newest version has cpu on the interface
It would be cool a tutorial to combine this with LivePortrait to have a vid2vid reenactment, with the focus on maximizing quality, but also avoid the warping of anything that is not the face (which means the rotation in the driving video is not taken into account).
I plan to make video to video tutorial for live portrait. It has huge potential
Can I ask, when I am recording the video to export, the VRAM seems full (red color) and its a lot slow. Should I clear the VRAM during recording. Still isn't clear on how to use this function?
How much vram you have? Before start recording yes I would clear vram
can i try on runpod?
If you can find a good desktop ui template there you can use. I am working on Massed compute tutorial. Cheaper better than runpod
@@SECourses yes pleasee... does Massed compute also work for comfyui?
@@frustasistumbleguys4900 yes it works. We already have installer.
@@SECourses yes, but no 3090 or 4090 options there
@@frustasistumbleguys4900 yes use A6000 with SECourses coupon. it is 31 cents per hour. dont select A6000 ALT, use A6000. apply coupon
which gpu to get for this? mostly for video, 3060 12gb or 3070 8gb or 4060 8gb? thanks
3060 12gb. Vram is the king. Ye 4060 would be faster but if not enough vram it will become 20 times slower
@@SECoursesAn RTX 4060 Ti (16gb) would be better ?
@@misaelmurrietaramirez8316 yep 100%
@@SECourses thanks
@@misaelmurrietaramirez8316 you are welcome
I press record, I press play and nothing happens.
Please test demo videos and let me know if they work
It doesn't support some input types like webm
@@SECourses I tested it on a demo video and got the same thing, it says in the console "Set saved video folder first!" but recording doesn't happen
@@Crypto_Goose ok it tells you the error. which folder did you set as save folder? recording wont start until you set it
@@SECourses haha rly) thx its work
@@Crypto_Goose great :)
I am using a 3060 12GB and when I turn on Restorer only and press the playback button, it slows down.
Are there any possible improvements?
It remains slow even if I set the thread to 1 and the swapper resolution to 128.
Thanks.
How much vram your system uses before starting the app? 12 gb should work fairly well. Even 8gb works single thread. If you need speed on massed compute 20 threads works at best settings : m.th-cam.com/video/HLWLSszHwEc/w-d-xo.html
@@SECourses The task manager shows that the CPU is always at 100% while the GPU has room to spare.
Is the CPU the cause of the slowdown?
Of course, Python, Git, CUDA and FFmpeg have been tested. :'(
@@mainevvv good question. what is your cpu model? it is highly unlikely but maybe
@@SECourses Intel i5-12400 is used.
The VRM is 16 GB and only 50% is used when playing with restorer on.
The VRAM in the top right-hand corner of the screen is also less than half.
The video plays at normal speed for scenes where no faces are shown, but slows down when faces are shown and a face swap is performed...
@@mainevvv if you have used my installer and followed python tutorial i guess that is limit. if you are a gold patreon supporter i can connect your pc and check it out
Can anyone comment on the differences between ROPE, Roop, and Facefusion?
Roop is great for photos. But not so great when things move in front of the face for video.
Facefusion does better masking for video, but you can only convert a single pic at a time.
I cant find any decent video showing the end results of a ROPE swapped video. Just a few frames less than 5 seconds of video.
ROPE has the best face tracking features. in my patreon post i included experimental development fork as well which has even more features. by the way entrance is 21 seconds that i changed with rope
hi! just subscribed your channed as member thouhg, used this program only for image changing, but the result is not really good enough , just wondered, is there another to get the better result, only for better image, not video. thank you! in advacne
Hi. Thank you for sub. Now we have 2 installer files. The landmark version has new experimental features. Have you tried it? Also to get better results your used faces need to be similar looking person. And can you tell me more info why bad image? I mean which part.
@@SECourses thank you for really quick response. firstly just used the file in the video, no idea of land mark, (bad image : compared to the face of the high-definition original image, the printed image is not sophisticated and the color is very awkward)
@@캐리커쳐애니메이션카 please show me settings of yourself in discord
@@SECourses i'm really a computer freak.
i don't even know discord, do you have url?
@@SECourses now testing new version with same setting~ as of now, GPEN1024 is making much better result than any other option!!! for making iamge , GPEN1024 is best , thank you for help
I remember reading a paper on roop and its creation 6 months ago, interesting to see how much it's grown.
yep it is next level now
@@SECourses can this be run locally or do we need a cloud . I have 24GB of Vram.
Thanks
@@behrampatel4872 this can run perfectly run locally. in this video i run this locally on my pc. cloud tutorial here : th-cam.com/video/HLWLSszHwEc/w-d-xo.html
@@SECourses Thank you for the clarity. From now on I will assume there will always be 2 videos showing both workflows. Local and remote.
Cheers and thanks for such detailed videos.
@@behrampatel4872 yes i plan to make that way. do you like this approach?
Error with installer, asking for $$$ to sign up.
yes installers are behind patreon. you can also manually install from official repo
Waiting for kaggle if possible ❤❤❤
Kaggle technically not possible because this doesn't run on web ui
What is kdggle and why is it better?
@@AGVenge kaggle is a free gpu service : th-cam.com/video/JF2P7BIUpIU/w-d-xo.html
brother i need your help
what help you need?
@@SECourses bro its paid But I thought it was free, so what if I make it myself? i known that its possible but it has more many files i can't be done it.so what i do bro can i buy this to you?
@@MahammudHassanLimon follow this link and subscribe 5$ : www.patreon.com/posts/most-advanced-1-105123768
@@SECourses what is the benefits of it ? I want to contact with you , i already messaged you if you see it then please reply .its very important
@@MahammudHassanLimonyou get installer files
Create notebook for kaggle please.
it is impossible. this works with desktop gui
@@SECourses ok 👍
hocam herşey paramı ya ücretsiz paylassanız ne olur sankı. yurt dısındakiler bile ücretsiz paylasıyor
hocam haklısınız ama şu anda ana gelirim bu ve youtube izlenme sayıları çok az
Oh Türk aksanlı İngilizce. nerede görsem tanırım :)
Doğru tahmin :)
Another good tool but not for us AMD users...😕
Sadly I don't have AMD so don't know would it work or not.
brother please reply my question
yes newest version rope next has cpu on the interface
@@SECourses sure ? ok i try but what is requirement for this files that i will perfectly install
@@MahammudHassanLimon requirements here : th-cam.com/video/-NjNy7afOQ0/w-d-xo.html
Why would i use this over FaceFusion? Sell me a reason
this has more features. we have a development fork installer on Patreon and it has way more features. sadly it wasnt ready when i published video so it wasn't shown. also this is uber fast with 20 threads on Massed Compute - 31 cents per hour
@@SECoursesthats cool. Does facefusion have an option where we can choose the strength of the original face used?
@@123playwright i dont know facefusion in details but Rope has it. this tutorial is about Rope
Sir please make this in colab link 🖇️🖇️🔗🔗
Because many of them including me doesn't have that powerful systems
Please 🙏🏻🙏🏻
sadly this wont have a Colab because it works with a GUI only. but hopefully I will make Massed Compute
@@SECourses SIR what is Massed Compute 🤔
@@harsh_108 Massed Compute is a cloud platform like RunPod. you can watch this tutorial first 10 minute to see : th-cam.com/video/LeHfgq_lAXU/w-d-xo.html
I BET YOU'RE FROM TURKEY 😅
@@BahadrAtes yes I am :)
I'm an English teacher. I used to jave an accent like your current accent😅 but you're the teacher here in youtube. Thanks for this amazing tutorial man. Really appreciated!
@@BahadrAtes thanks a lot. I am trying to improve my speaking to make it more understandable :D
@@SECourses your fluecny is wonderful. I suggest you to sing english songs. that way can help you out :D
@@BahadrAtes thanks a lot :)
Actually is not at all impressed by this. Good video though
:D i think it is pretty good
what is better out there? i just deleted FaceFusion, this tool here is lightyears ahead, on my 4090 the videos with faceswap almost runs in realtime, lol, incredible
@@nedo68 yep this one amazing best one so far
Why make easy when you can make it complicated… when will someone make a single installation file and have a single single icon for the application? Should we ask to AI to do it because humans are so dumb to do it???
i have single installer file and it works : www.patreon.com/posts/86307255
I'm going to tell you guys the truth about this very misleading video. This tool is no where near this good. Not even 20% as good as they try to make it out. It is easier than Roop and stuff, but does not produce better results. They, literally, cherry picked the example in the video just to make it look this good. It will not work remotely close to this well typically. Imagine making a competent video only to outright lie in your presentation of the results, claims of value, and title ruining an otherwise informative video.
well this is the easiest one you can get. i don't know any other app better than this one with such easiness
@@SECourses That part I can agree on.
give its code for kaggle free yusers who not able to pay a panny o plzz
After few days of my latest public Kaggle tutorial Kaggle blocked that way :(
@@SECourses oh great so technically you are making trouble not helping to people like that way :P
@@SECourses hey but Q is for kaggle simple policy its not wrong if we import our own personal any AI tool notebook so like that way I think we can upload ts code too there?? simply
@@SECourses ???
@@AaliDGr8 yes you can use as you wish as long as nothing illegal or NSFW