DomainFans

I'm a bad boy!

[DomainFans @ihtml]

@goodhunt Thanks for giving CNB a try! I’d love to hear your honest feedback as you dive in. Let’s make it even better together!

[DomainFans @ihtml]

@MengTo @variantui I saw this example of yours on Variant. Could you please share the prompt you used? I'd like to learn from it.

[DomainFans @ihtml]

@FreeFrontend @function Hello, You can register the domain https://t.co/2iKYPlKz8e; registrars like Dynadot, GoDaddy, and Gandi.

[DomainFans @ihtml]

@mattlefun @solanamobile 空投还有吗？测试一下

[MattleFun @mattlefun]

Seeker Airdrop is Live!

We’re thrilled to announce a special airdrop for @solanamobile users.

All Seeker wallets are eligible to claim up to 2,000 $MATTLE + 300,000 Mattle Points.

Drop your .skr wallet below and follow the simple steps 👇

[DomainFans @ihtml]

@mattlefun @solanamobile xiong.skr

[DomainFans @ihtml]

那就多看看这个世界吧！！！

[lusso @luusssso]

Architect Mickey Muennig’s personal home in Big Sur built in the 1970’s complete with floating bed

[Domain.News 📈 @DomainNews24]

Excited to share we’ve secured 5 domain showcase spots! 🌟 Showcase yours on https://t.co/DZHdP00mr0 for just $20/month. Limited slots available!

Application link:https://t.co/B1MtVjyNvl

#DomainShowcase #Domains

[DomainFans @ihtml]

@prkeshari Can it be open source?

[DomainFans @ihtml]

The domain https://t.co/0mo4NyWHdA is now available for purchase and it's the perfect foundation for your next tool website. With a name like https://t.co/0mo4NyWHdA, you'll instantly establish credibility and authority in your niche. #tools #website #domainforsale #site #info

[DomainFans @ihtml]

@VmDomains https://t.co/RjKk3cCth2 domain is forsale, .forsale domain

[DomainFans @ihtml]

@geekbb 还有人说你白嫖，就他高尚！！！

[DomainFans @ihtml]

@CosmotownDotCOM My domain renew faild, When I renew it, I get an error. Failure to renew successfully has caused the domain name to expire. Please handle it in time and reply to me.
Domain: https://t.co/aCoEkWsCJb
Status: FAILED_TEMPORARY

[DomainFans @ihtml]

@polkaworld_org $DOTA RICH

[DomainFans @ihtml]

很多人的推进不来？

[DOTA | DOT20 on Polkadot @dot20_dota]

🚀 DOTA MINT is officially LIVE!

🌐 https://t.co/5ctclAtprE

✨ The Dota Protocol is the first scalable inscription protocol standard on #Polkadot.

🔥 $DOTA is the first native inscription minted based on #DOTAProtocol

MINT #DOT20 TOKENS NOW!!

[DomainFans @ihtml]

身体机能或许出了很严重的问题。😒

[DomainFans @ihtml]

@ytzong 社保卡可用啊，我家娃就是她的社保卡。

[Barret李靖 @Barret_China]

Depix 是一个针对文字的马赛克还原工具，https://t.co/ymRNzCAF2x，文本类打码最安全的方式还是纯色全覆盖，马赛克被还原的概率很高。

本地跑了一下这个项目，M1 电脑，十几个字符的解码需要长达 10min+ 的计算，效率还是比较低的。它的算法思路是：

1）加载模具图片，最好是相同底色、相同字体、相同字号的文字列表截图，识别时会以模具为参考对照进行文字还原
2）将马赛克内容进行方形切割以分离单个子块，然后在模具中匹配每个子块，找出与之最匹配的矩形，并记录它们在搜索图像中的位置
3）核心算法在于子块匹配，通过对比矩形块的颜色差异，找到最小差异的矩形块作为匹配结果

这个库的实现并没有用到 AI 的能力，属于较为暴力的相似度对比策略，所以效率低是可以理解了，源码中用到的模具都是英文，如要支持中文字符的识别，就需要把所有的中文都放进去，那么效率就会更低了。

看到几篇利用 AI 来实现马赛克还原的论文，主要依靠生成式内容来补充像素，对于像素本就很丰富的图片而言会比较有效，例如老照片还原；针对这种程度的文字类马赛克还原，效果要差很多，不知道市面上是否已经有成熟的工具。