Vision-Language Models 👁️🤖

Tried some vision language models (VLM). Aliyun Bailian, Last year, I tried VTON models. Alibaba, being an e-commerce giant, clearly nailed it. You can see, that is much better than last year, though some detail on my trousers are lost.

After that I tried to use the video generate tool, the video is a little creepy.

Then, the cool part, Tencent’s Hunyuan world generation. What is surprising me it that, the world is not only a 3DGS you can walk around, but also, the buildings are with collider.

Input	Output