How To Create Deep Fakes Tutorial
VloĆŸit
- Äas pĆidĂĄn 18. 01. 2020
- I recently had some fun experimenting with deepfacelab.
We can use deepfacelab built on tensorflow to create awesome deepfakes.
In this episode I will be replacing my face with web dev simplified.
All the code will be available on patreon.
đSupport me on Patreon for exclusive episodes, discord and more!
/ dev_ed
đŽ Follow me on:
Twitch: / developedbyed
Twitter: / developedbyed
Instagram: / developedbyed
Github: github.com/developedbyed/
Microphones I Use
Audio-Technica AT2020 - geni.us/Re78 (Amazon)
Deity V-Mic D3 Pro - geni.us/y0HjQbz (Amazon)
BEHRINGER Audio Interface - geni.us/AcbCpd9 (Amazon)
Camera Gear
Fujifilm X-T3 - geni.us/7IM1 (Amazon)
Fujinon XF18-55mmF2.8-4 - geni.us/sztaN (Amazon)
PC Specs
Kingston SQ500S37/480G 480GB - geni.us/s7HWm (Amazon)
Gigabyte GeForce RTX 2070 - geni.us/uRw71gN (Amazon)
AMD Ryzen 7 2700X - geni.us/NaBSC (Amazon)
Corsair Vengeance LPX 16GB - geni.us/JDqK1KK (Amazon)
ASRock B450M PRO4 - geni.us/YAtI (Amazon)
DeepCool ATX Mid Tower - geni.us/U8xJY (Amazon)
Dell Ultrasharp U2718Q 27-Inch 4K - geni.us/kXHE (Amazon)
Dell Ultra Sharp LED-Lit Monitor 25 2k - geni.us/bilekX (Amazon)
Logitech G305 - geni.us/PIjyn (Amazon)
Logitech MX Keys Advanced - geni.us/YBsCVX0 (Amazon)
DISCLAIMERS:
I am a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.
#programming - VÄda a technologie
I never realized I could look so scary!
Your job is simplifying the web for us âș
That's sounds interesting so make sure to subscribe to this channel to get more videos just like that
đ
Hahahaha
how do i add new clips
The one you posted with me on Twitter was hilarious đ I can see this causing some big problems in the future though especially as itâs perfected
People created deepFake, and later created deepFake identifier, the world is strange
@@indrakhatiwada9860 Its because Destruction comes first Creation comes as a consequence.
There's currently a 1 million dollar kaggle competition to detect DeepFakes
traversy the legend himself
ooh two legends on same platform and even know each other !
Thanks traversy you are the first who teach me html +css very well
and Thanks to the dev Ed who encourage my skills to advance level :)
Step 1 : *You Need A Good Pc*
Me : Understandable Have A Great Day...
Cloud virtual machines ;)
you can do cpu training but the proccess is slow
imagine not having a decent pc
@@peptobepto im part of poor gang
Instructions not clear.
Daddy Traversy's face is on my left big toe.
Finally. I can now do Takoyama singing Dame Da Ne
Hahahahahaha
Best comment so far
its baka mitai not dame dane
Thatâs not how you do it you need to use first order motion model
You are the one who is bringing and supporting all the community members together. Thanks I hope you will keep supporting everyone even if you grow in millions cause you have a big heart Ed â€ïžâ€ïžâ€ïž
Thank you! One of the only videos that goes step by step. Easy to follow along. Subscribed!
I tried it but ran into unknown errors. Pixbim animate photos does what I need. Nice video.
Thank you!!! This helped so much! You took the time to explain every detail so itâs easy to understand
Looking forward to the colourizing vid đ
Brad one was quite hilarious...Innovation at its best ~đđđ You must have applied more than 150k iterations for Brad because it looked so closer to real Brad. Awesome stuff. So cool
OMG my first reaction.. đ
Holy Molly, that Kyle and Dev Ed fusion looked so strange!
This was helpful. Thank you.
I've just started working on it. WOW!
How long did it take
Excellent tut. Gonna play this through a few times as I have my first play with deepfake creation đ
Every one of your videos is pure art
The ending was more hilarious than the intro part xD
Pratyay Banerjee I would suggest to learn regular JavaScript and then move on to es6 or higher. As long as you understand the basics you should be fine.
@@carlosvasquez-pt4sb who asked
carlos vasquez uuuuhhhmm
Great tutorial, thank you! Your tone also is very soft and nice and that's make easier to get selfsecure to make such a difficult processes. Also would like to know where you get the music from? Thanks in advance.
this is incredible simple processed
creative developer :) good job
Kinda reminds me trying to convert video formats back in the late 90's and early 2000's. I'm sure this process will be come MUCH simpler and efficient in the future.
I love these kinda videos! đđđ
bro you are killing it!!!
Really appreciates your efforts.
đđđ you are đ„
Thatâs why we love you! đ„°
You're peanut?
Nuts?
Florin Pop u r from twitter? I follow u there đ
Akshay yes, I am đ
@@prasadlahane5486 *nutter butter*
You made it easy to us... Tnx
"I need therapy" is something I identify myself, great video very interesting, can it be implemented the video result in some web/CSS/html/Java of your tutorials? Maybe something with scrolling?
Thanks a lot !
You did excellently!!! I have request to you. will you share or explain on Facial Expression Recognition using CNN+ LSTM/ GAN/ DBN (I mean hybrid machine learning model)?
Thank you som much
Thanks for that video.
Does it hard to use AI to turn pictures styles? like the Van Gogh's style that we see around the internet?
Just wondering , would a Bunch of good High quality face pictures work to do the transfer to the face in the video...? or does it need to be still shots from that same type of video..??
please reply , Thank You so much, Great Work.
nicely done! no MAC?
This tech is going to be absolutely terrifying in 10 years.
I literally had Web Dev Simplified open on another tab when this video started playing hahahahahahah
hey ed im a huge fan i need to know the differences in web designing and devloping for coders
Hi! So, amazing video, im definitely suscribing !! i have a question: if i want to deepfake myself into a scene, is it better if my video is 60fps instead of 30? Or is it the same?
can't wait to see WDS to react !!! :D
Nice , subscribed to your channel .. i like your personality and the content you share
Cool now i understand the basics concepts of deepfake or deepface thankssssssssssss man
how do I load a saved backup? .. so that iterations start from where it was left?
Oh man, I was scared!
everytime i finish a project and merge mp4 i keep getting just a image not a video wassup with that dev ed
Interesting to see this older version of DFL. I started making Deepfakes last year just before the RTX 3000 series cards came out. Settings are totally different now.
You making millions off of it yet?
I wanna know from where do you find these cool stuf
please help !!!!!
when i typ on google drive on the site he write me error 404 (cant find) The requested URL was not found on this server.
please help me !!!
Keep going bro â„ïž
Dev Ed you are amazing
will i need to separate my clip if it includes cut scenes and blocked faces at momments
Hey Kyle nice work, great educational information on your channel, quick question..........i tried the deep fake and did all steps , do you know the reason why my Train (step 6) wont load, i tried all version, it gives all error any ideas
Great video!
Please make more videos on machine learning. :3
Thank youđâđ
do u have the software for amd graphic card cant find it thanks
when I convert files to mp4 the result is the source video, why? I use 6/7 DF
when do we implement coding on a ux/ui design build and how? how do implement databases with these web pages that we build cause people need to download files u know
Assuming you already have the video files ready, how long would it take on average to build a five minute video?
Please increase font size... I use a phone to watch your tutorials and I can't see the font :(
Whoops my bad on this one
@@developedbyed đ
Same.
@@developedbyed
Hey, in 18:50 u compared 50k vs 150k iterations but in what resolution were they both? Both 256 and the difference is only with the iter or they have have different resolution as well? What the resolution in both?????
@@developedbyed please answer....
"in case you want to replace your body with someone better...." đđEd why?!?!
thanks to your tutorial I managed to understand something
Where can i find the full tutorial and code? Link please.
Does the Opencl version use the processor or the AMD graphics card ?? Is there anyway to use both the processor and graphics card for maxmum speed ?
0:05 i am still laughing
Dev Ed - How to apply effects ( blur etc. ) to all frames, when converting ( step. 7 )? Thank
Hahahahahaha! Super hilarious intro! :^)
Imma bout to prank my sister by putting my face on a movie as one of the background characters.
Amazing!
what youtube downloader you use? I can't find one that do the job, every web I went won't let me convert to mp4
ytmp3.cc and switch to mp4
when i go to git hub and try to find the ''windows desktop app'' it doesn't show up so i cant download it, did they change it? / at 1:06
I forgot the title because I had a lot of tabs open
At the beginning I was gonna comment what happend to your beard
Hi! Can anyone tell me why am I getting this error while training: "Unable to allocate device-local memory cl_mem_object_allocation_failure"
My GPU is not enough? It's ASUS Radeon R7 370 4GB GDDR5 256bit
Wow its realy great
Nice one
when i do point 6 it tells to me "no saved models found. enter a name of a new model" can you help me?
You skipped a step. After you grab the frames for both files you need to run them through cropping. check your data dst and src folders, If there are not frames in both the main folder and the aligned folder for both you skipped a process step.
tnx
In stage 6 did not work with me as required. and In the end result, there was no change. What is the solution please?
Can you make a tutorial about colorize black and white pictures?
does the duration of the video affect the process time?
1:52 old hamster ay lol
Hey, in 18:50 u compared 50k vs 150k iterations but in what resolution were they both? Both 256 and the difference is only with the iter or they have have different resolution as well? What the resolution in both?????
Please answer đđ»đđ»đđ»đđ»đđ»
when i go to train it says no saved models found
Im am doing this on a rtx 2070 and still getting out of memory error even tried batch of 1. No way I am doing this right. Any advice?
i wont try this, but my $100 laptop wouldnt even get 0.1
Can someone tell me what to do if there are multiple faces in a video and i only want one face to be deepfaked?
no images being saved at aligned folder.. whyyyyyyyy please help me
before training it took me 28 minute and i'm using a laptop it's gtx 1060 3gb and i left it 20 minute because i wanted to just try it
Could you use photos as source if you have no videos?
There is no "windows desktop app " to click from github link
Will you ever do voice deepfake tutorials?
brother what would be the pc specs for Deep fake, and I like to go for low budget computer
Wizard Ed :)
Great Dev Namaste ...
I'm not getting he same menu on github so i have no idea what I'm supposed to be downloading.
it keeps crashing me in the TRAIN SAE step. what can i do? (i have a 1060 6gb card).
with my specs i ran through it like butter. with the default videos took me less then 10 min cool
which specs?
@@JoaoLucas-dd4xm 5950x amd rtx 2080 ti 128 ram gen4 ssd
You likely did the basic non-HD model. Try doing the SAEHD model with ~320 resolution or higher. Takes months, on hardware as good/better than yours. Results look pretty impressive though.
after train h128 u used convert h128 not sae anymore right?
The script 4 "4) data_src extract faces S3FD best GPU" is not recognizing faces, so I get an error "... just allow square images"
When I run data_src faceset extract it ask me for a GPU index. What value should I insert there?
Just press enter for default results. It will probably use the GPU.
Finally found someone uses CZcams Premium
Kayl goes on Police xD
I have a 2080ti and 8 Gig VRAM and it still crashed me at 1 batch size? Any suggestions?