Image created with gemini-3.1-flash-image-preview with claude-sonnet-4-5. Image prompt: A weathered Chinese photographer with vintage film camera on tripod photographing people against a half-demolished concrete wall, dusty transitional neighborhood setting, fire horse standing naturally near rubble in background, scattered Polaroids on folding table in foreground, overcast flat daylight, muted desaturated colors, observational realism composition, large white text overlay reading IMAGES in Chinese cinema poster style, river-gray and concrete tones, documentary stillness, human-scale intimacy
BitDance is our lunar new year gift: a 14B parameters autoregressive image generation model – that is autoregressive in bits, not codebooks It’s fast given 14B parameters, try it: https://x.com/multimodalart/status/2023797260057014372
ByteDance releases BitDance: Scaling Autoregressive Generative Models with Binary Tokens “”We present BitDance, a scalable autoregressive (AR) image generator that predicts binary visual tokens instead of codebook indices. With high-entropy binary latents, BitDance lets each”” https://x.com/iScienceLuvr/status/2023707945104458097
Interfaces are slowly leaving the screen. Instead of pressing buttons, you interact with things that behave like real objects in space. In this case, music stops being a timeline and becomes something you can walk around, touch, and reshape from any angle. And for my dear”” https://x.com/IlirAliu_/status/2021948191784960024
The scariest part of AI slop is its growing believability. People are less likely to question or investigate a fake that fits their worldview. It’s flooding every video platform. It will be weaponized for cognitive warfare at an unthinkable scale and precision. It’s a total”” https://x.com/TheHumanoidHub/status/2024194678871220415
⛷️Here’s my entry for the fast generative model olympics🥇 The Sphere Encoder is an autocoder so powerful that it produces high quality images quickly and without diffusion. At training time, we learn an encoder that maps natural images uniformly onto the surface of a sphere.”” https://x.com/tomgoldsteincs/status/2023796756366963032
Back in the day when we figured out that large-scale pretraining seems to be a great thing, I then very quickly decided to spend a significant amount of time and effort towards removing all our downstream transfer/eval task’s images from our pretraining data. But I didn’t just”” https://x.com/giffmana/status/2023481657177911383
DuckDuckGo rolls out AI-powered image editing on Duck.ai – 9to5Mac https://9to5mac.com/2026/02/19/duckduckgo-rolls-out-ai-powered-image-editing-on-duck-ai/
Generate an SVG of a pelican riding a car in France with a cat sitting beside it. Background has Eiffel tower.”””” https://x.com/OriolVinyalsML/status/2024519608833810496
Moondream’s inference engine got so fast image decoding became a bottleneck. So we shipped a SIMD image decoding library that’s faster than all the Python options I know of. Plus it’s statically linked so not a pain in the ass to install.”” https://x.com/vikhyatk/status/2024005498874306984
Use @Photoshop inside @ChatGPTapp now. No install required. It gives you slider-level control over effects like Cyber Glitch and Time Travel Warp. You can even open in Photoshop for web if you want to go deeper. #AdobePartner”” https://x.com/bilawalsidhu/status/2024178413511266802
I continue to be struck by how much less annoying GenAI video is than image gen. I think it’s because this actually looks like a sample from the training distribution rather than some weird overproduced output from aesthetic guidance. Not everything needs to be “”beautiful””.”” https://x.com/jd_pressman/status/2023256826431852852
Leave a Reply