You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Add global gallery viewer with scale animation for all images and videos
- Created GalleryProvider context with scale/translate animation
- Implemented GalleryImage and GalleryVideo wrapper components
- Added Notion-style toolbar with close and download buttons
- Videos autoplay/loop/muted as thumbnails, full controls when expanded
- Videos play at 80% volume in expanded mode
- Applied gallery functionality across all blog posts
- Added trailer_4.mp4 to taste blog assets
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <[email protected]>
The number 1 problem you will run into while building something like this is - Latency. Ideal total latency for video games: 20–60 ms (that pretty FAST{" "}
The <Highlight>fundamental difference</Highlight> in my approach here is to write game scripts (pre-meditated or generated by LLM on the fly) for controlling the agent input.
116
116
</p>
117
117
118
118
<p>
119
-
<Imagesrc="/blog/llm_wukong/assets/catjam.gif"alt="Cat Jam"width={20}height={20}className="inline h-5 align-middle"unoptimized/> LLMs never controls the control input directly. LLM only decides what game script to run. The scripts are then injected in the game engine and executed in realtime.
119
+
<GalleryImagesrc="/blog/llm_wukong/assets/catjam.gif"alt="Cat Jam"width={20}height={20}className="inline h-5 align-middle"unoptimized/> LLMs never controls the control input directly. LLM only decides what game script to run. The scripts are then injected in the game engine and executed in realtime.
120
120
</p>
121
121
122
122
<p>
@@ -167,7 +167,7 @@ You must respond with 'action_type' field with one of these values.
0 commit comments