Hollywood is cooked.

MikelArteta

Moderator
Staff member
Supporter
Joined
Apr 30, 2012
Messages
264,746
Reputation
35,019
Daps
808,403
Reppin
Goatganda the pearl of Africa
Crazy thing these are the early days imagine 5-10! Years from now?

Prob can just type a prompt like 2 hour action Movie with plot centered around 50 black special forces being sent back to 1770 and slaughtering slave owners

And poof
 

High Art

👑King of The Salon👑
Supporter
Joined
Mar 11, 2022
Messages
9,615
Reputation
5,091
Daps
51,326
[Gone Wild] AI is quietly taking over Youtube. This was the 3rd most watched video last week with 130m views


Posted on Sat Jul 5 11:20:53 2025 UTC




Commented on Sat Jul 5 11:33:47 2025 UTC

How many of the views and likes are also AI?


│ Commented on Sat Jul 5 18:43:52 2025 UTC

https://i.redd.it/mnhjegq5p3bf1.jpeg
mnhjegq5p3bf1.jpeg


│ It says 144 million now. But the same video has hundreds of millions of views on other channels as well. Lol

I wonder which country those views are mostly coming from. :patrice:

Either ways, the Antonov plane reminded me of this:
 

Mystic

Banned
Joined
Nov 10, 2015
Messages
11,686
Reputation
-973
Daps
19,003
Introducing Eleven v3 (alpha) - the most expressive Text to Speech model ever.


Posted on Thu Jun 5 18:24:36 2025 UTC


Eleven V3 is crazy good




JOI's about to crazy with this @RennisDeynolds
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
69,406
Reputation
10,657
Daps
187,576

bnew

Veteran
Joined
Nov 1, 2015
Messages
69,406
Reputation
10,657
Daps
187,576
[Other] I used AI to create this short film on human cloning (600 prompts, 12 days, $500 budget)


Posted on Wed Jul 9 20:48:28 2025 UTC


Kira (Short Film on Human Cloning)

My new AI-assisted short film is . Kira explores human cloning and the search for identity in today’s world.

It took nearly 600 prompts, 12 days, and a $500 budget to bring this project to life. The entire film was created by one person using a range of AI tools, all listed at the end.

The film is around 17 minutes long. Unfortunately, Reddit doesn't allow videos above 15 minutes. I'm leaving the full film here in case you want to see the rest.

Thank you for watching!
 

peppe

Superstar
Joined
Jan 7, 2015
Messages
9,227
Reputation
3,747
Daps
41,085
Onlyfans AI was already popping but now that we have video :wow:
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
69,406
Reputation
10,657
Daps
187,576











1/11
@maviadam
🚀 This open-source AI video generator is INSANE!
🎴 Flood /search?q=#MeiGenMultiTalk
MeiGen-MultiTalk – a game-changer for video creation!
and singing ...
🔥 Check out these mind-blowing examples :
1 - Toon Characters



https://video.twimg.com/amplify_video/1936845987126870017/vid/avc1/960x480/2PDUQ_GgzQrQIMZc.mp4

2/11
@maviadam
2- Live show Conversation



https://video.twimg.com/amplify_video/1936846203192332289/vid/avc1/896x448/5BnaHCVLSkq-I8_N.mp4

3/11
@maviadam
3- Couples



https://video.twimg.com/amplify_video/1936846487792631808/vid/avc1/896x448/3dk2sltcc5UZccNb.mp4

4/11
@maviadam
4-Sing



https://video.twimg.com/amplify_video/1936846845658947585/vid/avc1/704x576/BjyoJcytrQGajJ0K.mp4

5/11
@maviadam
5-Couple sing



https://video.twimg.com/amplify_video/1936846914181345281/vid/avc1/896x448/4vFc9Wu0i30kBLJN.mp4

6/11
@maviadam
6-



https://video.twimg.com/amplify_video/1936847449307406336/vid/avc1/1280x704/iQMXS0K_99xBDadp.mp4

7/11
@maviadam
7- instruction-following videos



https://video.twimg.com/amplify_video/1936847685631250432/vid/avc1/1280x704/jRP2U3FH2vZEHsHB.mp4

8/11
@maviadam
8-



https://video.twimg.com/amplify_video/1936847984500613120/vid/avc1/896x448/iZCMb5ATj4i2fo6p.mp4

9/11
@maviadam
9- Creative videos



https://video.twimg.com/amplify_video/1936848096752746496/vid/avc1/576x704/gD1kxZ5baVbY5Dhw.mp4

10/11
@maviadam
10-



https://video.twimg.com/amplify_video/1936848152637685760/vid/avc1/832x1088/LZ9QJYdppF6awamT.mp4

11/11
@maviadam
Project page: http://meigen-ai.github.io/multi-talk/1194.2K

If you enjoyed this thread:

Follow me @maviadam for more! 🚀
RT the post below to share with your audience! 🔁




To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196




1/21
@angrypenguinPNG
Okay this is actually insane 🤯

A new lip sync method called MultiTalk just dropped and it's the best open-source quality I've ever seen.

Now works in ComfyUI!



https://video.twimg.com/amplify_video/1936543801717428224/vid/avc1/1280x704/ivga-i01Om_PDV13.mp4

2/21
@DreamStarter_1
For real?
I've seen Purz's stream and the results were not THIS good.



3/21
@angrypenguinPNG
i’ll be honest, i’ve yet to test locally on my own comp! this is from research paper :P

will check out Purz stream!



4/21
@1littlecoder
not yet on @FAL i guess? cc @jfischoff



5/21
@zsakib_
wait this isn't just the music video?



6/21
@CoffeeVectors
Ok now!!!



7/21
@codewithimanshu
Game-changer for AI-generated content.



8/21
@kiieford
the detail in the neck vein strain 👌😂 so good



9/21
@hey1tspriyanshu
how do you create long videos like 18 seconds with consistency. curious to know your workflow



10/21
@ttplanet
could you share how you set the promot to move the camera?my video stay solid with same background! Thanks



11/21
@Vijish68859437
Can @HeyGen_Official do this?
Opensource seems better.



12/21
@NewWaveAi2023
Have you got a link to the workflow?



13/21
@JoseMGalaZ
Can you share the model and the wf?



14/21
@paulbri1
ohhh shyt



15/21
@geoppls
What country is it from?



16/21
@SF5321916436736
A‘ūdhu billāhi min ash-shayṭāni r-rajīm



17/21
@nglprz
🔥🔥



18/21
@AgentFola
😰



19/21
@johndpope
we need the training code - Pull requests · MeiGen-AI/MultiTalk



20/21
@degengates
now im about to believe everything last 20 years in TV was all fake AI politicians deepstate military stuff mk ultra... and they now just told us that its all possible but they know it actually a long time ago already



21/21
@niushidazuan
3uR46fHUkJp8T7FxVk6xnM6a5m3zpPyi1d3QHeqCpump 非常棒的项目,真叙事非常好,让每一笔交易都变得有意义,加入我们,一起建设他。




To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196



1/1
@Norris29973102
Our New Open-Source Project: MeiGen-MultiTalk, multi-person conversational video generation
enables multi-person conversations 💬, singing 🎤, interaction control 👬, and cartoon creation 🙊.
🌐 MultiTalk
💻 GitHub - MeiGen-AI/MultiTalk: Let Them Talk: Audio-Driven Multi-Person Conversational Video Generation
Code and model is coming~



https://video.twimg.com/amplify_video/1927995490899218434/vid/avc1/1250x704/g_4gED_Lbqyj7QfS.mp4


To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196



1/4
@Norris29973102
👏 MeiGen-MultiTalk: the code and checkpoints are released. Enjoy!
Github: GitHub - MeiGen-AI/MultiTalk: Let Them Talk: Audio-Driven Multi-Person Conversational Video Generation
Hugging Face: MeiGen-AI/MeiGen-MultiTalk · Hugging Face



https://video.twimg.com/amplify_video/1932370627157192704/vid/avc1/720x940/IIFA-9m7AIA54vS3.mp4

2/4
@TomLikesRobots
Nice - some of these demos look great. Are you planning on a Comfy implementation?
The long video format looks interesting. It seems to use a sliding window with a fixed 3.2-second overlap and an audio cross-attention channel riding alongside.



https://video.twimg.com/amplify_video/1932451514762563584/vid/avc1/1250x704/NJZWUvTJZK7pBwii.mp4

3/4
@OriZilbershtein
Hey Yong, how can one reach you? :D Thanks



4/4
@wangleineo
Good work! What kind of hardware does it require if I want to run this locally?




To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196



1/1
@SD_Tutorial
Generate your talking avatar with:
Multitalk+ Wan2.1 🧐👇

1. Setup Kijai's Wan setup:
https://www.stablediffusiontutorials.com/2025/02/wan-video-generation.html

2. Download Kijai's released model:
https://huggingface.co/Kijai/WanVid...ideo_2_1_Multitalk_14B_fp8_e4m3fn.safetensors




To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196



1/1
@kurtqian
multitalk text to speech which built on eSpeak NG TTS synthesizer



Gu51NC_XUAAO249.jpg



To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196



1/1
@mix_buchi_
こちらはMultitalkによるlipsync

Anime style
Flux.1 -> Wan2.1 I2V with multitalk

/search?q=#AIart
/search?q=#FLUX1
/search?q=#AImovie
/search?q=#Wan21
/search?q=#multitalk
/search?q=#ComfyUI



https://video.twimg.com/amplify_video/1945439821503279105/vid/avc1/480x640/aJ1SnbIHy30P2c_e.mp4


To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196






1/1
@creator_kachun
Research - gen A.I. character singing + pose control with WanVideo MultiTalk + VACE

Song: 許美靜 - 傾城



https://video.twimg.com/amplify_video/1945708528817356800/vid/avc1/720x480/ooPdwKgzCZQW8_Ys.mp4


To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196




1/8
@fffiloni
Meigen MultiTalk @gradio demo is available on @huggingface 🤗

Duplicate on L40S for personal and unlimited inference, enjoy !

*Compatible with multi-GPU too 😉



GuSz_AqWgAAowlo.jpg


2/8
@fffiloni
Meigen MultiTalk HF Space: https://huggingface.co/spaces/fffiloni/Meigen-MultiTalk



3/8
@codewithimanshu
Personal, unlimited inference on L40S? Impressive!



4/8
@naundob
How long are these examples supposed to take? I’m over 12min on A100 and still no result.



5/8
@zaesarius
Oh



6/8
@theinsiderzclub
Looks interesting

Curious to see how MultiTalk performs



7/8
@LouisUn1t
Important context re: agentic shift-AMD + mimik's new edge AI fabric could enably more distributed LLM running. What will hyper-local model execution impact most?



8/8
@Mind_0S
Exciting progress-but how do we extend these agentic capabilities to disconnected, local, or privacy-first contexts beyond centralized infrastructure?




To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196




1/3
@mesmerlord
being a shorts slop merchant is expensive.... and most of the costs is from the capcut sub(29.99/mo from next week) I had to pay for. the video itself only cost me like $0.3 in FAL api creds and my own runpod pipeline for multitalk



https://video.twimg.com/amplify_video/1946117532294782976/vid/avc1/720x1280/VtESOxE1BMsWmHQP.mp4

2/3
@EsotericCofe
Have you tried Hedra too? Just curious



3/3
@mesmerlord
wayyy back in the day when they were not that good, haven't gone back since maybe they're better than multitalk now?

I just like the option of making all these on my own site atm lol, costs me like 0.02-0.03 per hook vid and I make like 10 and choose best 3 or so



GwIBBr6WUAA4Zol.jpg



To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196
 

bnew

Veteran
Joined
Nov 1, 2015
Messages
69,406
Reputation
10,657
Daps
187,576


1/2
@maxinnerly
A serious leap just happened in AI video production:
MultiTalk, a new open-source model by MeiGen-AI, can generate multi-speaker, lip-synced, animated dialogues with stunning precision. Think crypto Twitter threads turned into ultra-clean, deepfake panels, in 15 seconds flat.

And the kicker? It works with just a 4090 GPU, supports up to 15s videos, and syncs multiple voices to multiple faces… without needing post-editing.


This isn’t just a tech demo. It’s a direct threat to expensive video shoots, particularly in the crypto marketing industry.

•Launch trailer for your L2? Skip the studio.
•Partner spotlight? Render it with AI faces.
•AMA clip in 7 languages? Done by lunch.
•Explainer for DeFi flows? Generate it with multilingual characters.

MultiTalk solves the last pain point in video generation: accurate, cheap, scalable lip-sync for dialogued content.

The real innovation?
> Label Rotary Position Embedding (L-RoPE) = tracks voice-to-face identity with ridiculous accuracy
> Built for multi-person scenes (unlike Hedra or Runway)
> Works with animated or real faces
> Free on Hugging Face, integrated with ComfyUI

Crypto teams should immediately test use cases across:

- Token utility demos (animated personas)
- Announcements (subtitled in 5+ languages)
- Founder explainers (using AI-generated avatars)
- Community-driven storytelling (UGC turned into visual panels)

Combine with tools like Topaz, Real-ESRGAN, or Luma AI to upscale to 4K and control camera motion.



https://video.twimg.com/amplify_video/1944080317079597060/vid/avc1/850x570/_cgAE7tXmoKYkvsp.mp4

2/2
@maxinnerly
try it out for free: MeiGen-AI/MeiGen-MultiTalk · Hugging Face




To post tweets in this format, more info here: https://www.thecoli.com/threads/tips-and-tricks-for-posting-the-coli-megathread.984734/post-52211196
 
Top