POM
@peteromallet
Followers
6K
Following
13K
Media
2K
Statuses
9K
Working with a ragtag group of nerds to accelerate The 2nd Renaissance, @banodoco (https://t.co/K3EvyToS2N)
Berlin/Ireland
Joined December 2011
Sharing new nodes & a workflow for traveling between images w/ Wan! . They help you use VACE to create endless videos w/o break-points + excellent input image adherence. Link here: Many exciting examples of it in action in #wan_resources on Banodoco
13
24
179
RT @mhshapes: Gardens 🌿. #touchdesigner #audiovisual #aiart #animatediff #ai #digitalart #generativeart #stablediffusion .
0
1
0
(Though I feel like it should go without saying, my disgust at his recent antics aside, Elon lifelong contributions clearly make him one of the most net positive impactful people alive).
0
0
2
Every day I find myself growing more grateful that sama machiavelli’d Elon out of OpenAI.
2
0
1
MORE VIDEOS FASTER . FASTER VIDEOS = FASTER GRATIFICATION . FASTER GRATIFICATION = MORE VIDEOS. COMPETITORS LOSE, IMAGES LOSE. VIDEOS WIN. TRY GROK IMAGINATION TODAY.
Grok Imagine is now making *videos* in 1/2 to 1/4 the time that major competitors take to make a single image!.
1
0
5
1 week of live-publishing my weight done. Results = 📈📉❓
I’ve noticed that doing things in the open magically makes them get better so I added a live feed of my weight to my website - including 11 years of measurements. If you ever find yourself wondering “how much does POM weigh right now?” look no further:
0
0
3
There are so many possibilities in using modern video models to upscale old school output:.
1
0
6
And that's actually only covering 1 of 3 repos I was working on related to the project (though this caught most of the work).
0
0
1
Got a summary of my git commits from today. Asked Gemini 2.5 Pro, o3, and Claude 4 Opus how long they estimate that work would take a good developer:. - O3: 5-6 weeks.- Gemini: 3-5 weeks.- Claude: 3-4 months
1
0
2
Walking down the street and I saw someone had spray-painted this onto a building. Now I'm as big a fan of Wan 2.2 as just about anyone but the public vandalism and swearing feels excessive. We as a community should stand against this kind of behaviour
0
2
7
China is absolutely dominating the US in all the leaderboards that matter for our species:.
0
1
19
I asked me dog if she wanted to go for a walk today. She seemed uninterested and I was confused because she always loves her walks. I questioned her to try to understand why she didn't want to go. Turns out Wan 2.2 had already brought her for a walk 🤯.
1
1
7
Wan 2.2 can do that Veo3 image writing thing 🤯.
试了一下 Wan 2.2 的提示词理解和遵循非常强啊. 可以搞定 Veo3 前几天那个图片上标记的提示词，说明本身图生视频上图片理解很厉害
2
19
123
Some people have expressed doubt that the advent of Wan 2.2 will bring a fundamental change to the very nature of how humans relate and communicate. Those doubters immediately become silent after I show them this video.
6
3
41
Wan 2.2 has been out for just 4 hours & its repercussions for our species are still barely understood. We are unprepared. We'll back on our way of life in the pre-Wan 2.2 era & be perplexed & even disgusted by it. May god have mercy on our souls. (credit: JohnDopamine)
13
30
344
#1 sold for the princely sum of $22.50. $22.50 x 1939 = $43,402 - okay for an art competition but not great . Think with more excitement/better explanations could probably get a few times that. More experiments soon!.
Sharing my first NFT as an experiment. It's #1 of 1929 test generations I made while making Steerable Motion for Animatediff, which I believe was the first approach for using batches of images to control a video model. Any proceeds will go towards funding an open source AI art
0
0
2
How cool would it be if I could sell all 1900 of these for $100-200 each?. And give all $250k that comes from that to an art competition that in turn creates a flurry for artists creating beautiful high-effort work with open models?. A glimmer of utopia to combat the slop.
Sharing my first NFT as an experiment. It's #1 of 1929 test generations I made while making Steerable Motion for Animatediff, which I believe was the first approach for using batches of images to control a video model. Any proceeds will go towards funding an open source AI art
0
0
2