I've been trying to run these models on my mb m3 and kept having version and incompatibility issues since ever. Your video made my day... magic.. thanks millions..
@@TheoreticallyMedia it's literally magic. you can even select the intended faces and swap only the ones you want. it's so hard to code this..(at least for me :) ) on M3 chip 16 threads swaps on average 20 frames/second which is pretty fast with no significant GPU temp rise. on M3 Pro/Max it would be probably much faster. amazing stuff, thanks again, you are my hero :)
This was just so mindblowing. The crazy and so fun things I have done so far with this are just off the wall--and I only watched for THE FIRST TIME about 3 hours from this post. Get in now because this will be mainstream in less than a year I predict where it just becomes a simple app that we all use.
Very cool! My biggest reservation about downloading executables is possible malware. It's easy enough for a hacker to test their products against anti-malware software to check if it triggers - and if it doesn't then you'll never know it has something bad in it. My advice to people is to only download executables on a computer that is strictly for AI on which they never do any financial transactions or enter a password for something that is not AI related. Better safe than sorry. Cheers!
That is 100% valid. It’s a little less likely on the Mac side, but totally not immune. I go all the way back to the Limewire days, and remember people clicking on an .exe to listen to the new Metallica song!
Not gonna lie...have been thinking about doing the same! I mean, we aren't at one-click for a whole movie quite yet-- so, it would take some massive effort to pull off-- but...I mean, maybe worth it for the reaction. I did have an old pal who would hack the subtitle tracks on foreign films and change the names of all the characters to whoever he was watching them with-- I thought that was pretty hilarious!
oh man that's a great idea. about a dozen friends of mine use my plex server. now I wanna put my face on Tom cruise in the newest mission impossible for a quick minute and see if anyone notices.
Awesome tutorial thank you. I had some pains getting it to work at first but after changing some security settings on my PC it now works as in the video. I can't believe how fast it is to process deepfakes with a RTX 4090. In the time of typing this comment it was able to do a 1 min faceswap on a 1920x1080p MP4 video, wow! The only downside of this method is that the quality of the results leaves a lot to be desired most of the time - lots of weird artifacts and it can't seem to handle teeth well for example.
Tim, great vid as usual. Your output is getting better and better with each iteration. I always know that you have done a deep dive into any new tech that you stumbled across instead of just pumping out a quick click bait thumbnail. You are my go to channel when i want to get i want to get on the pulse of new imaging tech..bravo sir bravo!
Oh, thank you so much! I'm in that mode of "make each video better than the last"-- so, I'm really happy to hear that you noticed! As far as the content-- yeah, that's been my bullpen from the start. I never wanted to make "Here's how I made $3000/week with ChatGPT!" type stuff...most of those channels are dead now anyhow. And, although there are those "crunch" weeks, where I'm just trying to showcase all the stuff that came out, one thing I ALWAYS want to do is add a tutorial piece in there. Mostly because I want to play with the toys too!!
A very thorough breakdown, Tim. I really appreciate your putting through its paces for our benefit. SuperChad! Laughed out loud with that appellation. Looked great though. Cheers!
Tim, I'm most interested in text to voice, which I know Eleven Labs has nearly perfected. But, an equally awesome free version running on my PC (and not hosted on a website) would be better.
I'll keep an eye out and try to get something for you for sure! I've recently started dabbling more into AI Voice-- and yeah, I tend to agree that Elevenlabs (currently) has that market cornered!
Looks awesome but I am still struggling with running it. I followed the steps but I did not get the "launch normal mode" after the install. After installing it said to go back to Dashboard and Launch application...but it doesn't seem to work. Not sure where the dashboard is but currently, it just shows FACEFUSION 2.3.0 icon on the left and on the far right, it has a circle rotating. It says it's "RUNNING" but all I can do now is click on the files folder, delete, or install. I'm confused
This is a top content. I always wanted to try deepface lab but it requires specific knowledge and a super powerful PC. Keep us updated, please. I hope in the coming year there will be an easy way to do high-quality deep fake videos similar to deeptomcruise
That's great to hear! I mean, the power of your machine is only going to dictate how long the process will take...so, a potato PC will DO it...it'll just take a long time. The bigger issue has always been installing Python, dealing with Github etc...I've really been waiting for someone to come along and simplify that process...and it looks like we're here!
Hi, that was a fantastic video mate, i learned a lot and i wish i new about this before! thank you for showing us how to do this, nice one, but when it comes to the bladerunner edits, i think the very first version of bladerunner that was not messed about with was the best one! i loved it when he was commenting on what was happening in the film loved it! take care and ill look forward to your next set of vids!
Amazing, thank you very much, you are the one guy that goes for here when it comes to those new technologies, the mid-journey face swap I first learn from you, loving it :)
Can’t wait to try this. Probably easier to cut just short scenes out of a movie though ? To make things a bit quicker ? I assume I can use any extension eg. Mkv, mp4
Oh, it’s 100% clip based. I think you’d run into all kinds of errors if you dropped a whole scene in. For best results, it is one shot at a time, and then putting the re-face version back into the final movie. I think you can use any format? Mostly tried with mp4s, but that’s because Premiere doesn’t play nice with MKV files.
100%! I'm starting to dig into AnimateDiff right now for everyone! The tools are still a little iffy in terms of installing, but I know they are being tirelessly worked on right now! I'm sure within the next few weeks we'll have a stable platform to start building on!
Tim, you're awesome. Great vid. I'm just trying to understand how you knew I had so much extra time in my life to play with all of these new toys... But seriously, this looks great and very simple to use. And especially as a Mac user, it's nice to find stuff the runs locally. And now, seeing how simple it is, this is going to make this next year in politics and elections a bit too interesting.
Politics are going to be an AI dumpster fire for sure. I’ve pretty much had to disengage from any of that discourse because…well, there’s just no point anymore. The thing I keep going back to is: I don’t think it matters? We’ve seen people refute ACTUAL video/image/audio (ahem, moon landing)- so is this going to change anything really?
I wasn't expecting the installation of Face Fusion to take so long and be so elaborate. It had to download Visual Studio or something like that. That took a super long time to install. Then some other ffmpy codec shell thing (????). I was nervous... but it looks like after about 20 minutes it's installed now.
Haha, sorry. I think that install time can vary depending on your connection and machine speed. I did say it would take a minute, but yeah- I also cut out the actual install time for the video or RU-vid would have killed me on retention. Glad to hear it is working though! It’ll be worth the wait!
Thank you so much!! Yeah, I try to keep a balance of “this is what it is” without getting too lost in the woods. I always want this stuff to be approachable to everyone. That’s the great part about where we are with AI right now- you can go full power user, or there are slightly easier methods to use as well! Love Pinoko, since it eases up the whole installation process. That alone can be a barrier of entry to some!
Does anyone know why the low resolution Insightfaceswap model hasn't been improved upon yet? All of these solutions use the same base model it seems and to my knowledge Insightfaceswap is actually only 128px which explains the low resolution output in all of these results. Perhaps it's just very hard to get a 256px or 512px model working but it would dramatically improve the outputs.
Ok, wait - 1. 🤯 2. Does it retain the audio? 3. Have you checked to see how 'safe' this browser is? 4. Possibly dumb question - do I just, use something to record the screen when playing a movie first? That would lose the audio though right? 5. 🤯 Thank you, as always. Constant ways to distract me from life! Love it. 👍Nicholson and you = scary but awesome!
So, it does retain audio-- or rather has the option to, or not to. I just kept it off for the video. In terms of safety-- I mean, your call there. If you head over to the Discord, you can ask the Dev about it. It's literally one person...might be getting hammered with questions right now-- but, there are some REALLY smart dev folks in there who are slopping code all over the place. I figure, if those brains are cool with it? For screen recording, you can use Quicktime...I think I ended up just using OBS, which is free. Play the movie on one monitor and just have Obs record it.
Thanks sir. I followed all steps but I don't see the same Facefusion screen you see. I see Files, Launch default, Launch Benchmark, Launch Webcam, Update, Install, Factor Reset. Did I do something wrong?
Great video and the software looks amazing but i can't get it to work! I've tried various photos and videos but i get a few seconds into outputting the video and just get an error message but with no explaination beyond that. I've kept all settings standard, as you said not to mess with anything too much but is there something i should be changing?
Haha. The one that I totally no regret not thinking of is Bill Paxton in Aliens: “Game Over, man! Game over!” …i might just have to do that just for fun!
When I'm trying to launch normal mode, it takes forever. It has been loading for many hours and I have tried to turn off my PC and try again but it takes forever. Do you have any suggestions on what I can do?
I downloaded and was able to successfully run Face Fusion. The only problem I have now is none of my videos have fully rendered and been made yet, it only shows the preview on the right box. I tried to tweak around with the settings a bit but I have no idea what any of this shit means. I'm not very tech savy.
Matt Wolfe introduced me to this in one of his videos. I did the install but couldn't get any of the apps to work. I gave up after the third app of course, so I didn't try all of them.
I’ll take a look. That’s one of the things that’s a little tough with Pinoko, and while I often recommend it, it is also a platform that runs locally, so there are so many variables as it if it’ll work well (or even at all!)
@@TheoreticallyMedia Thanks Tim. by the way, I really appreciate your videos. I am an instructional designer who specializes in animated explainer videos. So you are really helping me to expand my capabilities.
Installing this was anything but easy. When installing it, I kept getting error messages because so many things were left uninstalled, and you have to learn anaconda prompt and python to install who knows what that they don't bother to tell you about or install for you.
Yeah, sorry- that can vary from machine to machine. I’ll say, on the Mac side, it ends up being significantly easier to install just because you don’t have to issue a bunch of terminal commands and other software packages beforehand. On the PC side, it seems like the install is best if you didn’t have previous SD installs. All that said, again they’re really early with the platform- and working hard on it. You might want to hop on the discord and leave some feedback. They’re really open to hearing suggestions.
@@TheoreticallyMedia Might sound a dumb question, but how you reopen it? i can find anything it installed to open up? went back on the website and im just getting a white buffer circle. Worked first time round?
I mean...I installed it? But that is totally your call. If you want, head over to the Discord to get a feel. CocktailPeanut is the dev there, you can ask some questions to get a feel.
Maybe you have an older version here, but my version running on Linux has an option under "Face Mask Types" called "occlusion". Turn that on, and I bet the chopsticks would display correctly.
I just discovered this. I ran a quick swap. The faces don't blend. There are visible seams or the source face keeps distorting on the target face. What settings should use? Also if I enable lipsync, the software doesn't work.
I did everything you said. Facefusion says its connecting to a server, and never gets there. this cost me an entire day. NOTHING you said would happen occurred. I'm SO annoyed.
Help me, please. My installation got stuck here: ModuleNotFoundError: No module named 'tensorflow' (env) bash-3.2$ I´m on Macbook M2 Pro Max with Sonoma 14.1
Unless I'm off the mark here I think you need to do a new video. There seems to be about 10 extra steps now and I don't see any simple download on the facefusion site. It appears now you need to go in an enter a bit into terminal with lots of hiccups along the way.
Yeah, the one downside to Pinokio is that it is constantly changing. It's kind of a community driven project, so there are updates all the time, and things added/removed etc. I'll swing back over to them for a refresh video. It's the struggle of doing AI content on RU-vid. The tech changes so quickly, but the videos are locked in time! For the most part, things stay the same, but Pinokio is kind of like a 5 year old on a sugar rush!
@@TheoreticallyMedia some how I got it to load up but once I start to render it stops after about 10 sec... any suggestions? Also I just subbed to you, so you have video on upscaling video? I have some older footage from a film I made I want to try and upscale. Thanks for any suggestions
Haha-- Yeeeeah, I didn't mention that-- sorry. I've actually been juggling a few installs on account of that. That's one downside of running locally-- it takes up SPACE.
Just an FYI. When I installed on a Win10 machine it warned that I had an older version of Python pip. I updated as it was loading and yet it said it couldn't install inswapper because it needed a critical load that required the newer pip. So, although it seems pretty clean my previous older install of Python screwed it up. I assume if I did not have Python already installed it would have installed it for me.
Ahhh--- well, I'd say for sure to report it as a bug with the installer-- That way the error logs go back to the devs. You might want to hop on the discord too, there are a lot of big brains who might have a solution or workaround for you.
Hey man, great stuff. I gave this a try and it worked amazing the first time but after that iam trying for a different set of image and video but iam not getting the output video. It says processing and then just blank grey box, there is no play or download button. Basically after my first, its not working. Iam able to see it in the preview section though frame by frame though. But not in the output box. Please please help me man. I need this for my new short movie
@@wearegroot3114 Did you find a solution? I haven't got this to work at all but am having the exact problem you described, preview frames only, no output. I have a 4090 RTX.
@@lanoi3d yeah I have managed to get it up and running On the left side after selecting the face swapper option, just a few buttons down, you will find a checkbox called cuda. Check that. This makes the generation 2 times faster. And another thing I noticed is that if there is a lot of movement in your video the generation takes a long time and sometimes even shows an error message. Even after following all this if you are not able to download your output video then you can click on the refresh button on the very top of the screen. This will refresh the software soo you will have to select the source and target videos again but this time since it is starting fresh you will be able to download your video. And even if this don't work then go back to the place where you launched the app in the beginning, there will be a stop button on the top, click on that and then click on launch default again. Good luck with your creations
'C:\Users\provi\pinokio\api\facefusion-pinokio.git\facefusion\env\Scripts\activate' is not recognized as an internal or external command, operable program or batch file. This is what I'm getting so cannot get it to work. Any ideas??
Are you upscaling the outputted video with the likes of say Topaz Video AI as I have a m2 studio Mac and the resolution output its very low? how have you made your footage watchable?
To be honest, just Premiere and “optical flow”- although I’ll probably pick up Topaz fairly soon. I am just running in 1080 and not trying to uprez to 4K though.
There should be something in the target section (not at my computer right now) with a drop down menu that will allow you to choose a person on the right or the left of the frame. I haven’t pushed it with a group shot of 3 or more people, I’m not sure if it can do that…
Not sure-- I can give it a go. Hmmm, maybe Mahershala Ali on Wesley Snipes as Blade-- just to get a little preview of what that's gonna look like? Man...that movie needs to be out soon.
I ran into that problem as well and had to reinstall. Check with the discord. I think the devs have put together a ticket system for addressing issues. Fixing is above my level of knowledge, but I can say that there is a fix for this problem!
Hi there, nice video, Everything up and running but for some reason when I start the process it stops after like 1.5 seconds. Just won't process, also I noticed I don't have Keep-Fps in the options box, any ideas? Thanks
Which version are you running? The latest is 2.1.3 (as of Jan 23)-- If there is one issue with Pinokio its that it is a community driven project-- so, they are playing with builds all the time. Fixing things as they discover things aren't working. I'd suggest checking in with their Discord, likely there is a solve there.
As a joke for my wife I'm trying a face swap putting me in the Ricky Business dance video. It works fine when he is sitting at the table, but not at all on the quite active dancing. Any suggestions on settings, or a different application for this?
sorry, maybe I'm a little slow here...I followed instructions and went through the install process. It finished and that next screen with the red rectangular dialog boxes (black background) never showed up. All I see is a white screen, with no obvious way to advance to the display seen in 2:27 of your video. Can anyone please help me?
Check to make sure you’re using the latest version of Pinokio and faceswapper. Same thing happened to me, and I don’t realize there was a whole other 2.0 version out there!
I have a very powerful computer with RTX4080 and i9 13 generation, but this application works very strange, it takes me about 20 tries to make one good output image. All the time or blured photo or cant render or dont swap face. !00 times need to restart server. I tried this app on few PCs usless, same issue
Check in with the Pinokio discord. That does seem odd. They’re got some pretty smart folks in there, and the devs are really active. You’ll get WAY better support than I could possibly offer!
Great, but after trying it out, there's a perpetual trembling of the image (even though it was a relatively static video). Do you have any idea how to avoid this in the settings, or in post-processing with another app? how do you call this kind of specific problem?