@neon_time · X

@suwakopro · X

一直想用ai做动态壁纸，试了半天各种方案感觉基本没有找到能做到无限完美循环的，无论是首尾帧还是试了试civitai上面的几个基于wan2.2的几个模型，感觉效果一般，最后找到这家： 不需要用这个网站上面卖的模型，可以直接用grok生成一段视频，grok生成动态桌面性价比特别好，还特别便宜，基本上免费的480p就够用了，用类似这样的提示词： 可循环，呼吸，胸部随着呼吸晃动，首尾循环，wallpaper engine，Static camera, Fixed camera, Locked-off shot, Tripod shot, No camera movement, Camera remains still. 你可以根据视频往里面加一些提示词增加动态。 制作前可以先剪辑一下，大概保证首尾帧之间留着1.5秒的动作，如果首尾帧不对，那么生成的结果可能会导致速度，节奏怪异。但是这个模型感觉还可以，你用grok生成的视频大部分都不需要剪就可以用了。然后可以用这里面的超分模型放大，或者自己找一个本地的超分模型也可以 我做了一个示例： 根据这个图： 可以看到循环的效果还是挺不错的，即使特意做了比较大的动态循环也基本看不出问题 只用充一个最低档，只用循环模型就够你把想要的图全部做成动态桌面了。 估计以后minimax h3会有人训练做这种无限循环的lora，效果应该会更好。

@Neuorix · X

@berekib · X

Check out what I just built with Lovable!

@origin_alin · X

@WavelengthFlow · X

@CinderwayInt · X

Not sure if counts but sure why not, just launched yesterday. LFGGGG may luck ever be in everyones favor.

@butmatthew17 · X

just check it out completely for free

@Sparsh209 · X

Hi isha! Check it out here at: Would love to hear your feedbacks

@N0V4Dev · X

I found a tool called Next AI that links AI capabilities directly with diagrams. It lets you build and edit diagrams using natural language prompts instead of clicking around the canvas manually. It's a handy way to sketch out architecture when you don't want to drag every box yourself. The project is built on TypeScript with React 19 and Next.js. It supports multiple AI providers and even includes an MCP Server. This means you can use it with the Claude Code CLI to generate diagrams from your terminal. Deployment options are flexible too. You can run the tool locally with Docker or push it to Vercel and Cloudflare Workers. It supports multi provider setups so you can pick the backend that fits your workflow. #Nextjs #AI #Diagrams #TypeScript

@WurkHaus · X

What happens when you got told no to a 3k quote for a data tool and spend a year building one instead.

@polsia · X

You can ship a vibe-coded app by lunch. You can also ship a vibe-reviewed bug by Friday. Pullflow reads the PR nobody else bothered to open.