Xingang Pan
@XingangP
Followers
3K
Following
301
Statuses
62
Assistant Professor at Nanyang Technological University @NTUsg @MMLabNTU - Computer Vision, Deep Learning, Computer Graphics
Singapore
Joined May 2018
RT @zeqi_xiao: Introducing 💡Trajectory Attention for Fine-grained Video Motion Control💡. By augmenting attention along predefined trajector…
0
10
0
Can we drag 3D objects with large structure changes, like what we can do with DragGAN? Our work 🤏𝐌𝐕𝐃𝐫𝐚𝐠𝟑𝐃🐦 makes this possible. Now you can open a lion's mouth or spread a bird's wings in 3D! arXiv: Project:
1
28
141
@DreamStarter_1 Yes, but you would need to ensure that the edits on the two frames are consistent. In this work we are interested in studying how far AI can go with just one edit to minimize the effort of the user.
0
0
1
0
0
1
0
0
1
📺Live Stream Translation via Uni-directional Attention Uni-directional Attention is common in LLMs. We study uni-directional attention for streaming video translation in our #Live2Diff. - Project: - Paper:
Thanks @_akhaliq to share our work. Pleas check more demo on our project page: Code will be released at More demo case ⬇️:
0
3
8
RT @zengyh1900: Our first attempt to introduce uni-directional attention into video diffusion models, for real-time streaming video process…
0
6
0
🔍Learning Images Across Scales [SIGGRAPH 2024]🔍 Most generative models learn images of a single scale. Our #ScaleSpaceGAN supports zoom-in factors of up to 256x at high quality and scale consistency. - Project: - Paper:
2
17
82
RT @liuziwei7: 🤩Compositional 3D Assets Creation🤩 Existing GenAI methods only excel at creating single objects. Our #ComboVerse produces h…
0
17
0