Šárka and Ondrěj have done incredible work building EbSynth, creating an entire image and video editor from scratch in webgl! Their attention to detail and the care they’ve taken at every aspect of its design is inspiring.
We're launching new EbSynth today 🎉
EbSynth is VFX software that lets you change your video by editing one frame. You can turn any video into your canvas, right in your browser. Go try it out 👇
1
11
2,357
talk is uploaded!
🦠Watch @computerender's first public talk on Mote: an interactive ecosystem simulation!
Mote uses a custom GPU-based physics engine to model hundreds of thousands of organisms, leading to fascinating emergent phenomena: a sandbox that is part game, part research.
Link below⬇️
22
11
143
9,601
I’m giving a talk on my new project next month!
Join us on August 13 for Localhost!
🦠 @computerender will present Mote, an interactive ecosystem simulation with hundreds of thousands of organisms. His custom GPU physics engine models many simple behaviors at a massive scale, producing fascinating emergent phenomena.
RSVP⬇️
11
25
226
30,978
Peter Whidden retweeted
SENSEI can also guide exploration in combination with task rewards. When playing Pokémon Red from pixels, we achieve superior performance to Dreamer (pure task rewards) and Plan2Explore. Only SENSEI manages to obtain the first Gym Badge within 2M steps of exploration 🥇
7/8
2
1
31
3,691
Peter Whidden retweeted
We trained a tiny model to beat Pokemon Red on 1 GPU. This is a qualitatively new result for RL because the game takes ~25 hours to beat! Most tasks used in research run seconds to minutes. Work with @dsrubinstein @DanAdvantage @kywch500 @computerender
17
46
595
37,088
David's been doing awesome work on this! It's been super fun to see this develop in the last couple years. Check out his blog post!
Excited to finally share our progress in developing a reinforcement learning system to beat Pokémon Red. Our system successfully completes the game using a policy under 10M parameters, PPO, and a few novel techniques.
Blog posted below
19
2,710
The work Mykhailo has been doing on this project is incredibly impressive. He's created the first tensor/ml library that supports pytorch nn style code, realtime rendering, and non-nvidia gpus. He's already made a ton of amazing demos. Check out the blog post & star the repo!
I wrote a blog post about my own Python tensor compiler library called TensorFrost that I've been working on for the last 14 months!
michaelmoroz.github.io/Writi…
As a sneak peak here is the last pet project that I've made using it - an implementation of Neural Cellular Automata:
2
1
14
2,968
Peter Whidden retweeted
A YouTuber open sourced an RL algorithm for Pokemon Red and there's a multiplayer map where you can watch different people's models explore the game
8
38
345
29,328
Peter Whidden retweeted
Yet another breakthrough: the AI has just managed to defeat the first gym leader!! 🏅🏅 progress has been accelerating recently, how long will it take to get to the next city?
in-game playtime: 5h
total training time: ~5,500h
4
3
30
4,072
Amazing to see how quickly a community is forming around RL for Pokemon. @dvruette and others have been working on support for gen3 games and are already making a ton of progress! More soon
Baby steps... AI has previously learned to play Pokemon Red, and now it's learning to play Pokemon Emerald as well. What will be the biggest obstacles on its journey to become a Pokemon master?
2
24
3,315
Looking forward to this!
Peter Whidden @computerender will be presenting his work on playing Pokemon with reinforcement learning at LIFE monthly Monday 11/13 at 9 AM PT / Noon ET. DM me your email for an invitation!
1
18
2,780
Peter Whidden retweeted
Pokemon Red agent hacked our reward function. We gave points for gaining health, so it learned to repeatedly deposit and withdraw pokemon...oops
Demo powered by Pufferlib. Project by @computerender. We're all in the Discord -- come hang out and contribute!
2
6
64
9,497
Over the last couple years I've been training a reinforcement learning agent to play pokemon red. I put together a video documenting the AI's learning, and my process creating it. Enjoy!
🔗 piped.video/DcYLT37ImBY
24
40
243
39,499
Can dynamics be distilled from models trained on only static 2D images?
An experiment distilling Stable Diffusion into @zzznah's Neural Cellular Automata model using @poolio's method from DreamFusion.
Prompt: "raindrops on glass"
More info and code: github.com/computerender/nca…
2
4
13
3,387
This approach is intuitive, given that these models are capable of representing a particular subject from multiple perspectives.
Additionally, however, these models may also be able to represent a subject at multiple points in time!
1
3
1,741
Peter Whidden retweeted
Nouvelle expérimentation d'installation interactive qui combine créativité humaine et intelligence artificielle. Créez votre composition avec des formes colorées et laissez l'IA vous surprendre avec son style unique.🎨🤖#AI #TouchDesigner #interactive vimeo.com/796622497
1
2
1,725
New tutorial available showing how to use computerender for vid2vid styling!
github.com/computerender/tut…
#stablediffusion
#vid2vid
3
26
3,501
Peter Whidden retweeted
Working on a component for TouchDesigner to seamlessly generate images with Stable Diffusion using computerender.com. I'm so excited to share the component and tutorial shortly. #stablediffusion #newmediaart #aiart #audioreactivevisuals
2
7
65
7,438