It seems curating+cleaning+captioning the training data is 95% of the whole process when training a model.
Oh boy, another audio-driven video. Here are the input specs for SeeDance 2.0
wavespeed.ai/blog/posts/seed…
#seedance2
It’s weird how I spent an hour thinking about what the next version of Dirix will look like and every single thing I ended up on has been mentioned here.
Replying to @BLVCKLIGHTai
believe me, i'm working on a long form project now and while it's possible it's tedious, there's a few features that would expedite this exponentially
- 3D splats for environments to allow consistent space geometry and camera reframes
- Longer shot lengths to allow proper coverage of scenes rather than generating every shot individually
- Camera reframes of generated videos to create full scene coverage from multiple angles
- Radio Play lip-sync so you can voice act the scene and have the characters match lip-sync across all shots in a scene
Special thanks to @elevenlabs for accepting Dirix into their Grant program and sending us 33 Million credits!
Less than 2GB RAM you say?
RIP Elevenlabs. Alibaba Qwen dropped this new AI text to speech
Free & open source, <2GB vram, super fast.
It has instant voice cloning, emotion control, and even voice design.
Full tutorial: youtu.be/eC8mZceIy5k
Here is what is possible in Dirix's upcoming music video generator + editor.
Welcome Marcuz. First artist from Dirix. Full track on Thursday.
Today we are inching closer to solving one of the biggest obstacles in making long-term projects like films and series:
Character Consistency
If this is something you struggle with, fill out this form to be among the first to try Dirix:
tally.so/r/3yOd5g
Early look at Dirix! 👀
Not completely ready yet, but "if you're not ashamed of launching, you're taking too much time building it". So here I am, both excited & ashamed to share this! 😅
Now accepting Creative Partner applications. Ready to join us?
tally.so/r/3yOd5g
Dr. Sky, someone who has never been out of California, was ordered to go to the center of the universe because of his mistake (which he says is arguable at best) during an experiment.
Created for @CuriousRefuge and @epidemicsound AI Comedy Contest
#CuriousRefuge #EpidemicSound





