Log inSign up
Wei Cheng
60 posts
Wei Cheng profile banner
@wchengad

Wei Cheng

@wchengad
AI researcher at StepFun
wchengad.github.io
Joined August 2015
293
Following
138
Followers
RepliesRepliesRepostsRepostsMediaMedia

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
  • @wchengad
    Wei Cheng
    @wchengad
    Aug 22
    Thanks for sharing, @HuggingPapers @_akhaliq! Congrats to Hengyuan on the new work! He’s taken WithAnyone one step further with WithEveryone — a new SOTA model for multi-human face identity-preserving generation. From WithAnyone → WithEveryone — an exciting evolution! 🚀
    @HuggingPapers
    DailyPapers
    @HuggingPapers
    Aug 21
    WithEveryone: group image generation with up to 10 identities A new framework from Tencent Hunyuan and Fudan that unifies planning and identity grounding, improving face similarity and cutting copy-paste artifacts.
    Image
  • @wchengad
    Wei Cheng
    @wchengad
    Jun 29
    Big Congrats to Sijin 👍👍
    @ch3cook_csj
    Sijin Chen (CH3COOK)
    @ch3cook_csj
    Jun 29
    🤖 Translation as a Bridging Action — Learning Novel Robotic Manipulation Skills from Human Motions We’d like to share our research on leveraging human action data: can bimanual robots learn to perform new tasks from human demonstration data? Human motion data is low-cost to
    Image
    00:00
    Image
    Image
    Image
  • @wchengad
    Wei Cheng
    @wchengad
    Mar 29
    Thanks @HuggingPapers & @_akhaliq for sharing PixelSmile (based on Qwen-Image-Edit2511) Unlike traditional image editing models that only achieve coarse expression changes, PixelSmile enables: • Smooth, continuous control over expression intensity (reducing semantic blending)
    Image
    00:00
    @HuggingPapers
    DailyPapers
    @HuggingPapers
    Mar 27
    Image
    PixelSmile A diffusion framework for fine-grained facial expression editing with continuous intensity control and robust identity preservation across human and anime portraits. Supports zero-shot expression blending and introduces FFE-Bench for comprehensive evaluation.
  • @wchengad
    Wei Cheng
    @wchengad
    Dec 31, 2025
    🚀 **OmniSVG Training Code is OUT!** 🎉 📅 Date: December 31, 2025 🎁 New Year's Gift for the Community! --- We're thrilled to release the **official training code** of OmniSVG — you can: ✅ Train your own SVG generation models ✅ Fine-tune on custom datasets ✅ Reproduce our
    github.com
    GitHub - OpenVGLab/OmniSVG-train: This is the official training code of OmniSVG
    This is the official training code of OmniSVG. Contribute to OpenVGLab/OmniSVG-train development by creating an account on GitHub.
  • @wchengad
    Wei Cheng
    @wchengad
    Dec 3, 2025
    🎉 OmniSVG @ NeurIPS 2025 — Come see our poster! •📅 Date: Thursday, December 4, 2025 •🕚 Time: 11:00 AM – 2:00 PM PST •📍 Location: Exhibit Hall C, D, E — Poster #5512 📢 Exciting News from OmniSVG! We've listened to the community and upgraded for stronger generalization
    Image
    OmniSVG/OmniSVG1.1_8B · Hugging Face
    From huggingface.co
Advertisement
Advertisement