FaceFusion 2.5+ Complete Install for Face Swapping and Lip Syncing on Linux, Mac and Windows

[@shadyendeavor]

This installation version no longer works. I have a completely new updated installation video for version 3.0. Please go check that one out. kzbin.info/www/bejne/pYrPkqiMnreSr5Y

[@Anna_Anna-cg9wr]

The best step-by-step installation instructions, there are no analogues. Thank you very much for your work.

[@ELChamuco-ug7tf]

Thanks. This looks much much easier than the install process used around the time that you made your first install video months ago.

[@shadyendeavor]

Heh, I hope so. And you're welcome. I like to think that I've been making better videos over time and even this one, to me, is already dated a bit. I will be coming out with a newer one very soon, as the process has already changed a bit because consistency doesn't seem to be a thing with the install process of FF. But a completely new version of FF will be release very soon and I will be making it for that one specifically so that it lasts longer...hopefully.

[@ffwrude]

Very comprehensible. Well spoken. And a comment for the algorithm for you

[@tye829]

There's already a slight change on the website since this -- also using a mac, when I click MacOS in step 1, it shows only "GIT" "Conda" then "FFmpeg"... Going to try to figure this out but if anyone knows how this impacts installation it would be much appreciated!

[@shadyendeavor]

The Python and PIP parts of the install would have zero adverse effects either way. As to why they are no longer in the install docs, I'm not sure, but I'm waiting for a response on that. So you're welcome to try it without doing those parts, however it will still work if you do. I'll let you know if I find anything else out. You will still need Homebrew installed.

[@tye829]

EDIT/UPDATE: I quickly found that I basically just follow this video but starting with "GIT" instead of the first ones, and it is now working. HOWEVER, I should say that I already had Homebrew installed and had tried this previously, so I got a bunch of "already installed" messages while I was doing this and not sure how this affected the process. But it worked for me.

[@shadyendeavor]

@@tye829 python is required, so I'm not sure why that first part is taken from the install documents. But because of what you said, you most likely had that installed already too. Along with PIP. Either way, I'm glad you got it installed.

[@shadyendeavor]

Ok, so it turns out that Python and Pip were rolled into one of the other parts automatically, so nothing to worry about. Basically, as I said before, if you run those 2 commands, it will have no adverse effects. But skipping those is now the "official" way now.

[@danielcormier5733]

Once you have done a faceswap, where does the output video reside???

[@shadyendeavor]

You should have an "output path" field. In there, you should see where your files go. If you do not, I recommend you have that filled in for future videos. That's the easiest way to do it and you can have it saved to a specific place permanently by added it to the ini file in the FaceFusion directory. If it is empty, then your file would have been saved to the temporary folder, which is different for the system you are on and not simple to explain how to find it. So the easiest thing to do, if you don't want to change the ini is to hit the little download button in the top right of the "output" area and that will download your file to your normal download directory. Just be sure you download it before you start another swap.

[@frankpitale]

i am running facefusion on pinokio on a enw mac book with m2. i have tried to faceswap on a simple elon musk video of 9 seconds 3.5 MB size. it's running since 45 minutes and still nothing. what NASA computer do we need to have a results in minutes?

[@shadyendeavor]

I'm doing it all on an M1, so yours should be faster. Make sure you are set to CoreML and not CPU. Change your thread count to 1.5-2x of what your shared memory is. Make sure you aren't running Frame Enhancer. For just a face swap, you should only have face_swapper selected along with face_enhancer if you want it to not look so pixelated. That's the basics, which is in my first "FaceFusion Basics" video. If you want to learn more on best practices and other features like lip syncing, I've got videos on all of it.

[@freakguitargod]

Ive followed everything exactly to the part 3: the Conda init --all - shows the error - 'conda' is not recognized as an internal or external command, operable program or batch file."" what does that mean.. ? thanks

[@shadyendeavor]

@@freakguitargod this video is now outdated for the new version 3.0. I'm finishing up a new one that should be out tomorrow.

[@freakguitargod]

@shadyendeavor ahh. I see. No wonder things look different. Thank you for responding so quickly frind. BTW Is there a way to get version 2.5 using this method? Or a way to downgrade to previous versions? I've always used Pinikio but seems to be buggy at times. Thanks

[@shadyendeavor]

@@freakguitargod I'm not sure how, you will have to ask on the Discord. But as far as I know, only to 2.6.1.

[@freakguitargod]

@shadyendeavor I meant the latest version, if it was 2.6.1 then that is what i meant . Reason: I created an Instagram model and some of her pictures are explicit and i use fusion for better and clearer faces and fusion has a NSFW filter on it and now using 3.0.0 I cannot fix photos with only a bikini or crop top shirt because of that filter. I bypassed the NSFW filter on previous versions is the main reason I ask this. Do you know a way to bypass that filter? Thanks again friend.

[@TheFiguresDontLie]

Thank you for this, can you help me? On Windows, I am being told it can't find the NativeDesktop workload when it installs and opens Visual Studio. I went on, thinking this would just be like an interface QOL issue, but my computer is telling me 'conda' is not recognized as an internal or external command, operable program or batch file. When I ask where conda, it doesn't see it. I def installed anaconda tho...

[@shadyendeavor]

@@TheFiguresDontLie your best option is to check out the FaceFusion Discord (link in the description) and ask for help. A few people there that are extremely knowledgeable and helpful.

[@TheFiguresDontLie]

@@shadyendeavor Excellent, thanks.

[@SMDrecordzz]

In the first step i only see GIT Conda & ffmpeg , i dont see python & pipp

[@shadyendeavor]

Correct. It changed to where python and pip are automatically installed with one of the other steps. You can skip those. However, if you ran them already, it will have no adverse effect.

[@biantian3334]

I have tried all the steps and followed the same as you showed in the video but it show this "CondaError: Run 'conda init' before 'conda activate' " What should I do Please Help.

[@shadyendeavor]

Close terminal and open it back up and continue from there. For some reason it requires a restart of Terminal sometimes. Hopefully that works. If not, come ask the question on the FaceFusion Discord. Linked in the description.

[@hebercloward1695]

I was curious if it is possible to get the 'frame colorizer' to go in reverse? Like make a new looking video look like it was done on old b/w film?

[@shadyendeavor]

No, sorry. That's a tool that's been available for 15+ years now with most video editing software.

[@moviebuster22]

It shows an error , No module found cv2

[@shadyendeavor]

I'm sorry, but I have no idea what that would be. As I say in the video, if you followed the instructions exactly and it doesn't work, that means there is something wrong with your system and your best option is to check out the FaceFusion Discord for help. Link is in the description.

[@ayushnayak4082]

Brother how to use in iPhone

[@shadyendeavor]

You can't install it on your phone. The only way you could actually run it on your phone is if you use it on a cloud service like RunDiffusion or ThinkDiffusion. I do have a video for using FaceFusion on RunDiffusion too. kzbin.info/www/bejne/oquvg3aJjNZqd9k

[@shadyendeavor]

The instructions have changed. Now having said that, if you follow the instructions in the video, it will still work and accomplish the same thing. The only change (to date) is that the very first 2 steps that were to install Python and PIP are no longer needed as their installation was rolled into another step automatically. So you can skip the first 2 steps of installing Python and PIP. But it will work the same if you don't skip it.

[@wellnessvibe]

Thank you so much for the update. Appreciate it.

[@mariomauriesgarcia8831]

and now with V3?

[@shadyendeavor]

@@mariomauriesgarcia8831 I'll have a new video in a day or 2 that is completely updated for version 3.0. Yes, this one doesn't work anymore.

[@orrowyi4219]

This site can’t be reached127.0.0.1 refused to connect.

[@shadyendeavor]

heh, I believe you figured out the problem on Discord ;)

[@orrowyi4219]

@@shadyendeavor haha thanks

[@Ntain86]

did it step by step in terminal but became "CondaError: Run 'conda init' before 'conda activate'"

[@shadyendeavor]

Did you run this? conda init --all If so, did you completely restart your terminal? If still not working, check out the Discord for more help. Link in the description.

[@Ntain86]

@@shadyendeavor I missed the restart terminal step. now it works. thank you

[@shadyendeavor]

@@Ntain86 you're very welcome! Have fun!

[@Fshdva]

Thanks, It was super clear. I just follow everything and at the step 5 this is the message : python install.py zsh: command not found: python Do you know how can I do ? thanks

[@Fshdva]

/usr/local/Cellar/python@3.12/3.12.4/Frameworks/Python.framework/Versions/3.12/Resources/Python.app/Contents/MacOS/Python: can't open file '/Users/(hide)/install.py': [Errno 2] No such file or directory

[@shadyendeavor]

@@Fshdva I could be wrong, but I know sometimes I get this issue and you literally just need to close your terminal and then just come back in and do the install. Just make sure you CD again and start your conda environment again too. Then try the python install.py If that doesn't work, you're going to need to ask for help on the Discord channel. The link is in the description.

[@Fshdva]

@@shadyendeavor thank you so much it works ! but now I have an other error 😅 I put it here juste in case you can again save me : ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. gradio 3.50.2 requires numpy~=1.0, but you have numpy 2.0.1 which is incompatible. 🙏

[@shadyendeavor]

@@Fshdva it should be fine. Does the install not continue after that and say that it successfully installed a different version of numpy? If so, you are ready to run facefusion. If not, back to Discord option.

[@Fshdva]

@@shadyendeavor It works for the photos but not videos .. thanks again

[@shubh-oc3lg]

Getting Error when i select CUDA : return self._sess.run(output_names, input_feed, run_options) onnxruntime.capi.onnxruntime_pybind11_state.Fail: [ONNXRuntimeError] : 1 : FAIL : Non-zero status code returned while running QuickGelu node. N ame:'QuickGelu' Status Message: CUDA error cudaErrorNoKernelImageForDevice:no kernel image is available for execution on the device

[@shadyendeavor]

you're going to need to ask for help on the FaceFusion Discord. The link is in the description.

[@martinkaiser5263]

when try install pip i get command not found

[@NinjaKittyBonks]

Within the description, there are LINKS to Discord and others to seek help. Depending on many factors, install may fail and not least of all is the OS you are using.

[@shadyendeavor]

The only way that would happen is if you didn't follow the directions explicitly or you got an error on a previous step. Discord is going to be your best option for support. Link in the description.

[@authenticjason5394]

How about for windows

[@shadyendeavor]

I'm not sure what you're asking. The title clearly states "...on Linux, Mac and Windows".

[@tonywhite4476]

Links?

[@NinjaKittyBonks]

They are in the description, above the chapters 🐱

[@shadyendeavor]

Which? I don't think any are missing from the description.

[@NinjaKittyBonks]

Interesting to see you are running a Mac, as I have long heard that these are often the preferable OS for those who's creativity lies in the arts. Yours really are excellent tutorials, so you maybe FF, Pinokio, etc or one of their parent companies should be considering supporting your work as an official guide(s) for not just installs, but concise "how to's". I see a Python folder in a number of my games and appears to be a server program. There is a least one PC racing game "RaceCity", but was a MMO, so do you have any tutorials on how to create a built-in server on my PC, so I could install and play that game? Tried to adapt files from NFS World, because that game's installer creates a server, but no luck. . PS> You do realize that since I run Win XP, we are now mortal enemies... right?😾

[@shadyendeavor]

Yes, I am on a Mac, and for the most part, Mac has been and is the preferred artistic creativity. The main reason has been (especially in the past) that everything just worked and you never had to deal with command line like you really had to in the past with DOS. And the computers are made by the same company making the software, so you know all of the hardware is going to work right out of the box. Not having to worry about any of that and just being able to create was the main reason for going this route. The main problem now is that doing most AI locally on your own computer is extremely computationally intense and are optimized for Nvidia graphics cards, which Macs do not have. I will never have a PC as my main computer, but I will admit, that as of now, Macs suck for AI computations. Hopefully that will change in June with the next Apple World Wide Developer Conference, where they are expected to have big announcements for AI. FaceFusion and Pinokio are both independent developers, so neither have a parent company. Python is a computer language that runs on top of the operating system's language. So it's not native, you need to install the language to the computer so that anything written with it will be understood. No idea on that game scenario/question. But I know I would have zero idea how to do something like that. And the fact that you run Win XP, a 23 year old operating system doesn't really make you much of a mortal enemy 😛

[@cbwinslow]

Liked and subscribed

[@shadyendeavor]

And commented! 😉 Thanks so much!

[@xiaomeng007]

(base) C:\Users\lzqgg\miniconda3>python run.py python.exe: can't open file 'C:\\Users\\lzqgg\\miniconda3\ un.py': [Errno 2] No such file or directory

[@shadyendeavor]

that's beyond my expertise, sorry. Please check out the FaceFusion discord. They will surely be able to help you there. The link is in the description.

[@wilcurran3377]

Pinokio makes it super simple.... when it works. The majority of installs fail.

[@shadyendeavor]

Yep, Pinokio is definitely another option. Which is why I have an install video for that route too. 😉

[@Skettalee]

Why does everyone out there just make videos of how to install these ai tools. Especially for Face Fusion and 3.0 I cant find any video telling you how to use lip sync or the way to batch process or run jobs on it and the live portrait feature making facial expressions change through time on a video like with keyframes or something. Dont you know how to use the tool? Especially for the Lip Syncing feature which you had in the title yet you dont say nothing about using it.

[@shadyendeavor]

I honestly can't tell if this comment is serious or not. I literally have a video specifically on lip syncing. I have one for queued jobs too. And I have an install video for 3.0 coming out sometime tomorrow. I have an entire playlist dedicated to videos of FaceFusion that I have made myself. So feel free to check that out. kzbin.info/aero/PLMwq9QXPN0iDReOut5vKekXT4V6wGTET8

[@Skettalee]

@@shadyendeavor Great Ill go look for it. Wait qeued jobs i thought that was just a new feature in FaceFusion 3. The job feature which I didn't see in any versions before that one. Ill go look through your videos, its mostly the batch processing that I cant find at all anywhere. Which i guess is actually called queued jobs or no?

[@shadyendeavor]

@@Skettalee Yes, Queued Jobs is the new 3.0 feature. Batch isn't something that the developer has or plans on adding to FaceFusion as he's stated many times on the Discord.

[@Skettalee]

@@shadyendeavor yeah I know I heard other people say that but that it’s completely stupid to me. So what is que jobs that Mr cause to me that’s the same thing as batch

[@shadyendeavor]

@@Skettalee It's a game of semantics. Batch implies that you are able to just grab a bunch of an item and put it with something and will just automatically do all of those. So, simply put, it would be that you have one video target and 10 source images and you want to just have all 10 images create a new individual video for each of them. Or the opposite where you have 10 targets and just want to automatically have all of them swapped using the same source. With the new jobs, instead, you have to create a "step" for each swap manually and add it to the job. Though it's more work, it's much more specific on what you can choose. With the batch (if you're using the term correctly), you wouldn't be able to choose which face is being swapped every single time if you have multiple people in a target. It lacks control. Either way, watch my Queued Jobs video to see what exactly it is and does.