<3

can you create a model from pictures instead of a video?

Wtf co to Robi na planecie faktów

Just downloaded what I think was represented as complete downloads - as GITHUB owner has made his files read-only. The extraction went ok BUT FOR the entire "Workspace" folder is missing. Windows 11 did not report any setup files. I have an NVIDIA A3000 (laptop) perhaps this and the DirectX alternate will not run OR some unknown security issue is causing the folder to vanish. THERE IS a .bat file which I ran and it opens to "DeepFaceLive". Is that it - the same as "Workspace"?

Much more difficult when you use longer videos with way more angles. This is cake in comparison

Thanks a million for this great tutorial. Eternally grateful!

The part where i should enter the Hugging Face Token does not appear.

Does this still work ? V5 or v

awesome. thank you <3

How do I set another face with the face I want to replace in the aligned folder?

Really cool Video and very good described. I wanted to make a still image alive, but it seems not to work. I don' see any proceed to making the image looking recocnizable in the Preview Window. Not in SAEHD nor in Quick96 Mode. But I am traings later. ;-) ----- EDIT: To avoid Missunderstanding: I made the still Immage to a MP4 Video with about 300 Frames. The Destination Video had about the sames Frame Numbers. I also switched the to dst / src each other.

thank you for making this video. I wish I could find another video that discusses how each specific setting would do in an advanced situation.

I love the video thank you, but please how can you create my own models and use it on DeepFaceLive. Please teach me

Great video! However, I'm running into a wall at the training section. When I run it, it says something went wrong after 40 seconds. (My goal is to make AI created images of my wife and I - and then eventually others. So if there's a newer free option, I'm open to it!) Here's where it gets to in the execution: Traceback (most recent call last): File "/usr/local/bin/accelerate", line 8, in <module> sys.exit(main()) File "/usr/local/lib/python3.10/dist-packages/accelerate/commands/accelerate_cli.py", line 43, in main args.func(args) File "/usr/local/lib/python3.10/dist-packages/accelerate/commands/launch.py", line 837, in launch_command simple_launcher(args) File "/usr/local/lib/python3.10/dist-packages/accelerate/commands/launch.py", line 354, in simple_launcher raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd) subprocess.CalledProcessError: Command '['/usr/bin/python3', '/content/diffusers/examples/dreambooth/train_dreambooth.py', '--image_captions_filename', '--train_only_unet', '--save_starting_step=500', '--save_n_steps=0', '--Session_dir=/content/gdrive/MyDrive/Fast-Dreambooth/Sessions/jcpdb', '--pretrained_model_name_or_path=/content/stable-diffusion-custom', '--instance_data_dir=/content/gdrive/MyDrive/Fast-Dreambooth/Sessions/jcpdb/instance_images', '--output_dir=/content/models/jcpdb', '--captions_dir=/content/gdrive/MyDrive/Fast-Dreambooth/Sessions/jcpdb/captions', '--instance_prompt=', '--seed=565794', '--resolution=960', '--mixed_precision=fp16', '--train_batch_size=1', '--gradient_accumulation_steps=1', '--gradient_checkpointing', '--use_8bit_adam', '--learning_rate=2e-06', '--lr_scheduler=linear', '--lr_warmup_steps=0', '--max_train_steps=3000']' returned non-zero exit status 1. Something went wrong Some potential issues: I'm trying to follow along via a newer version of fast-DreamBooth.ipynb - there wasn't a place to put the token in from huggingface. I'm doing this on a Mac - I didn't have Stable Diffusion installed yet. (I'm using this tutorial to install it if I should have it... uxplanet.org/install-stable-diffusion-ui-on-mac-beginners-guide-351e40a9e8e2) Training Image size - I The runtime - I need to be connected to hosted runtime for colab. The only GPU I have available to me is T4 GPU. When I'm running these steps, I assume I select that one. My image files are 1000x1000px. Maybe that's causing an issue? Thanks if you can help!

How do you put in your own videos and face images? All the tutorials I've watched show the included 'Robert Downey' files used. I confused! :-/

Thanks a lot, indeed fascinating!!what If we want to only faceswap in one of the characters in the movie? can we replace Elon's face with a few pictures of other individuals' faces?

just click model download on that step since is not need to get a hugging face token

Very nice,,how to 8natall pyton

What does data sort bat do?

How did the Tom Cruise deepfake achieve such incredible quality? Think they trained on tonnes of different video data. How do you do that?

Thank you for explaining it in a clear way

can you do this with source video, but instead of destination video you just got pictures ?

None of download links working for me.

Version 5.2 is able to give us seed number just by clicking Apps on the right above, then DM results Great video btw, keep it up buddy

No envelope symbol in my version! :-(

fantastic! thank you so much!!

can we save the target iteration anc continue later

perfect tutorial

Thanks for the tutorial, I’m using RTX 1060 i don’t know why my training window won’t open after running the SHED trainer

I am getting an error with training SAEHD. "ImportError: DLL load failed: The paging file is too small for this operation to complete." How it be fixed?

🎯 Key Takeaways for quick navigation: 00:00 🎨 Use Midjourney to create consistent characters for comic books, video games, or any project. 00:27 📝 Use tags and prompts to generate desired character variations in Midjourney. 01:24 🔗 In Midjourney 4, upscale the image and use the envelope icon to get the seed information for the character. In Midjourney 5, use the envelope reaction on the grid of four images. 02:46 ✍️ Write a formula in the prompt to instruct Midjourney to use a specific character with the desired seed. 04:01 ➡️ Change the character's actions, locations, and appearance by modifying the prompt and using different tags. 05:44 🙏 Thank you to viewers and subscribers for the support and engagement on the channel. Made with HARPA AI

Part that's missing and a big one is the SRC. When you extract faces how many people have a src video that has one face in it. After extracting the aligned faces from the SRC and DST you need to remove any that aren't the faces you are targeting.

how to do it if video has multiple people in frame?

Excellent video! Would love to see a Photoshop Crash Course Tutorial for beginners if you're up to it!

You do a great job with your video presentations! No fluff, long intros or bs weird humor (which I find cringe btw!) It's quite refreshing to have someone get down to the basics and straight to the point! These are the type of videos and media I lean towards in general. Subscribed!

Very clear - Only thing IM unclear on is during Training do we have to contiously click P so the graph changes or can we just leave training on for hours ( currently my iterate is 1 and ive left it on for 10 minutes)

Something to try with "full body" would be to change your AR to be taller. Your wide screen AR is what is causing the AI to do half body. If you changed just your AI and put (full body photo) in parentheses you might generate the full thing. Then you can out paint if you want the image back to a wide 16:9. Just an idea.

Amazing. I'm stoked. Worked like a charm with only using my i5 11th Gen CPU! No more paying $20 for 1 video to these scammy websites 👏🥂🤝

Thanks! This is refreshingly clear.

How to extract a specific face from a video where there are many faces and replace a face in another video

Thanks, this was a great intro video to get started and hit the ground running!

Awesome work.

what are the prerequisites before using deepfacelab ? do i need to install python and tensor flow ? and any other prerequisites ?

This was a clear explanation, I also hope that some people would explain like this with more specific informations. The onle problem for me is that you're going to do more of the processing deepfake editing and it's going to take more time and patience just to get the best quality. It's really hard. Thank you!

Could you please explain how to select a specific face in a multi-face video (source and dest).

how do you keep the exact outfit tho, say your trying to make an AI photoshoot , you want images to look like they're from the same place

how about same outfit, same face, but different scenes and poses?

One of the only good tutorials on this matter, nice work!

Thank you. This was very helpful.

can you make one for after detailer extension as well