Mechatronics Engineer | AI Engineer | Artist
Joined October 2021
- Tweets1.1K
- Following138
- Followers504
- Likes4.6K
Pinned Tweet
𝐂𝐇𝐎𝐎𝐒𝐄 𝐘𝐎𝐔𝐑 𝐂𝐇𝐀𝐑𝐀𝐂𝐓𝐄𝐑
After hours of research, experimenting and many failures, i managed to get the smoothest AI video2video i have ever done!
Its incredible how this technology lets me turn into any character without any animation or video effects background.
The tools i used:
-Stable Diffusion text2image
-Controlnet Extension
-EBsynth
There is still work to do.
As you can notice, the animations are not perfect.
There are still some "hallucinations" and temporal inconsistencies. This could be solved with higher keyframes grid.
But it also requires more computational power.
If you liked the video make sure to follow, ill be posting more cool ai stuff!
Sebastian retweeted
In the last 30 days AI busted reality and if you're smart you can get the work done of 30 talented men.
I or a single smart employee at @hyros_official is getting shit done in days that used to take take a month by a team.
This is a gift from zeus for young people. 3 dudes with zesty energy can realistically build a 100 million $$ business with just a Claude (support, operations) , Cursor (engineering) , Figma (design) and Lovable (front end site) subscription (all powered by opus 4.5).
The world just changed faster than I've ever seen it change in my life. Move fast.
Sebastian retweeted
I will be giving away 100,000 dollars to somebody from my email list on Christmas day.
Rules.
1) You must be signed up to my email list to win, I will randomly select an email address and email the winner on Xmas day.
Sign up for free on cobratate.com/newsletter.
2) You must retweet this tweet. If you are selected, and you didnt retweet, I will choose someone else.
Top G.
The G stands for Generous.
GOOD LUCK!
Amazing things can be done with Stable Diffusion AI image generation.
Than cannot be done by Midjurney or Dalle-3.
AI VFX experiment: AnimateDiff + ControlNet
This combines a simple logo animation (left) and #AnimateDiff with the QR Monster controlnet into a loop. Instant water simulation!
#aianimation #stablediffusion #VFX #mograph
3
358
DALLE-3 image consistency is crazy!! 🤯
Take a look between the phone camera and the actual scene.
Dalle-3 completely nailed it !!
Although image is not 100% perfect, this was done with:
- No prompt engineering
- No negative prompt
- No special model
For sure, this introduces a new era in AI image generation, where we no longer need to fine tune our prompts.
Here is the prompt I used:
➡️"A picture being taken of a cryptid sighting of Rick and Morty as they run into the bushes. Rick and Morty have gone completely insane. They turns their heads and creepily look into the camera as he makes his getaway. There's a thick fog, and the scene is dimly lit."
1
5
307
Humans
Lets remember this video is generated with only 1 input image, so there is limited information for the Ai to start generating consistent movement.
This one maintained temporal consistency really well, but had some decays in details.
8/10
1
2
202
