Tech operator by day. AI/video experimenter by night. Building strange signals from the digital rot.
West Coast
Joined February 2011
- Tweets11.8K
- Following1.6K
- Followers689
- Likes37.9K
Pinned Tweet
A random daily challenge from @ToshiArte sparked an idea.
Weight of Ash
Noooo @grok tell me this ain’t true
O M G - Sensational Bait
Fable 5.2 just wiped out animation studios and artists will need to learn to type.
Haha 🤣
Binary Rot retweeted
i gave this repo to fable 5.2 (via my claude agent fig) and told him to have fun making something interesting...
def feels interesting to me (stay for the turn)
i am continually astounded at the creative choices these advanced models are making
Binary Rot retweeted
Dreamina Drama Studio for AI short dramas is now LIVE!
✍️ Writing | DreamScript | In-house Script Model
▪ Generate scripts across genres for original stories, adaptations, global releases, and commentary.
▪ Break down scripts up to 100,000 words with ease to speed up creation.
🎨 Art Direction | Anycook | In-house Image Model
▪ More lifelike, more distinctive character design tuned for short dramas.
▪ Multi-style scene building to match different story settings.
🎬 Direction | DreamShot | In-house Video Model
▪ Switch freely between canvas and assets.
▪ Keep every shot organized and consistent in style.
▪ Orchestrate drama with 3D Director Studio with flexible camera setup and movement.
▪ Preview key shots early to control continuity and pacing.
🚢 Remaster & Restyle
▪ Upload a short drama to automatically detect characters, scenes, props, and shot structure.
▪ Regenerate visuals based on style, aspect ratio, and target market.
▪ Preserve the original pacing and localize shots for local audiences.
Create hit short dramas in one place — from idea to final cut — with Dreamina Drama Studio.
🕐 Limited-time offer:
▸ Dreamina Basic: just $0.29 for your first month.
▸ Advanced or Ultra: 50% off your first month on a monthly plan, or your first year on an annual plan.
▸ Dreamina Seedance 2.5: up to 50% fewer credits per generation.
🎁 Engagement bonus:
▸ Repost and comment within 12 hours of this post going live. The first 200 participants will receive 1,000 Dreamina credits each, on a first-come, first-served basis.
#DreaminaAI #DreaminaDramaStudio
🤖 Made with AI
Omg.. I have a 2DS hiding somewhere
Fun experiment getting Hermes, active and operational on a Nintendo 3DS.
I can talk to it from the 3DS and it does real work on my Mac. It can searches the web, writes and runs code, and pretty much anything else an agent can do. The 3DS is the remote.
Getting voice in was interesting part. The 3DS browser can't reach the mic, so Hermes and I installed custom firmware and wrote a small homebrew app in C. Hold A to talk and let go to send. The audio goes over to my Mac, Whisper turns it into text, and Hermes takes it from there. The 3DS is completely unaffected still works as intended.
Anything needing approval is surfaced. The screen goes dark and the command shows up. A approves, B denies.
The Hermes Girl animates through every state: listening, thinking, searching, writing code, sending helpers, done, asleep. There are 13 states in total.
Anyways here’s a few simple interactions.
🤖🖤
53
Hi @grok what are the top 5 click bait headlines from salon
Today, after 31 years, we are saddened to announce that Salon is ceasing publication. It has been an extraordinary privilege. salon.com/2026/11/05/goodbye…
2
3
111
This makes me want to get a local rig for genning 😭
Real footage in, 1980s OVA anime out. A style LoRA for MiniMax H3 Reference-to-Video (the real video goes in as the reference), trained locally with AI Toolkit by @ostrisai on a Dell Pro Max Tower T2 with an NVIDIA RTX PRO 6000.
The dataset has two parts. 21 image pairs: a real photo next to the same photo redrawn as OVA anime. That teaches the look. 10 short anime clips, each paired with itself as the answer. That teaches the model to copy the motion instead of making up its own.
The result: feed it a real video and it keeps the poses, outfit and movement, and only changes the style.
About 30 hours, 4,000 steps, rank 16. Loss went flat at step 1,000, but the samples kept improving until 2,500. By 4,000 it was overtrained. Rendered in @ComfyUI.
I will cover this in detail next week at the Dell booth at Adobe MAX, Nov 10 and 11, 2pm.
#DellProPrecision #DellTech #NVIDIA
1
10
270
Fine.
Here’s the @OpenAI crop top fall collection
Btw this is an XL.
wtf I didn’t realize the hoodie I got from @OpenAI dev day was a crop top.
I look so dumb