美術數膠｜Fine Art Digital Plastic

Furniture craftsman from Taiwan.
Using generative AI as material, not tools.
Zenodo-published methodology.

[美術數膠｜Fine Art Digital Plastic @psrmidi]

@thsottiaux Codex quota kept dropping with no visible task running. No CLI, code review 0 PRs / 0 issues. It stopped only after fully quitting Codex Desktop. Happened last cycle too. Can someone check server-side logs?

[美術數膠｜Fine Art Digital Plastic @psrmidi]

@mx_jlu @thsottiaux Similar issue here after reset. My Codex shared quota kept dropping while idle — no task, no CLI, code review 0. It stopped only after quitting Codex Desktop. Feels like reset/session accounting is off today.

[美術數膠｜Fine Art Digital Plastic @psrmidi]

@thsottiaux Codex quota kept dropping with no visible task running. No CLI, code review 0 PRs / 0 issues. It stopped only after fully quitting Codex Desktop. Happened last cycle too. Can someone check server-side logs?

[美術數膠｜Fine Art Digital Plastic @psrmidi]

If Codex Desktop can consume usage just by being open, users need to know that clearly. If it’s a bug, OpenAI should restore the lost usage or compensate the lost subscription time.
@OpenAI @OpenAIDevs

[美術數膠｜Fine Art Digital Plastic @psrmidi]

I’m paying $100/month for ChatGPT/Codex, but my Codex shared agentic usage kept dropping even when I wasn’t actively using it.

No visible Codex task running.
No Codex CLI used.

[美術數膠｜Fine Art Digital Plastic @psrmidi]

Code review showed 0 PRs reviewed / 0 issues. The usage only stopped dropping after I fully quit Codex Desktop.
This already happened in the previous usage cycle, and after today’s reset it happened again.

[美術數膠｜Fine Art Digital Plastic @psrmidi]

AnAnyone else suddenly hit Codex weekly limit on ChatGPT Pro? $100 Pro plan, personal account. Last used it ~14 hours ago for small UI changes. Now weekly usage is 0%, while 5-hour limit and Spark weekly are still 100% Bug, credits migration, or new limit?

@OpenAI

@OpenAIDevs

[美術數膠｜Fine Art Digital Plastic @psrmidi]

GitHub:
https://t.co/ZZooiPTzJS

[美術數膠｜Fine Art Digital Plastic @psrmidi]

I just published an open-source template: Designer-Led Agent Starter.

It is for product designers who are not engineers, but want to use AI coding agents to build an MVP.

[美術數膠｜Fine Art Digital Plastic @psrmidi]

The template helps you turn messy product notes into product truth, rules, flows, PM questions, capability maps, and acceptance criteria before asking the agent to write code.

Do not start by asking AI to build the app.
Start by making it understand the product.

[美術數膠｜Fine Art Digital Plastic @psrmidi]

とにかく、僕は今でもあのキャラクターたちが大好きなんです。 もしチームの皆さんで話す機会があれば、ぜひ「今でも作品の復活や、何か新しい展開を楽しみに待っているファンがいる」と伝えてください。
特に今の時代、あの3Dモデルならフィギュアなどの実体化でもすごく魅力が出るはずなので、いつかまた動くところが見られたら嬉しいです。

また何か進展や、手伝えることがあればいつでも連絡してくださいね。応援しています！

[美術數膠｜Fine Art Digital Plastic @psrmidi]

Ousiassさん、お久しぶりです！返信ありがとうございます。

実は、あの時に「鯉ゾン」のネタを描いた者ですが、覚えていますか？

最近、当時のキャラクターたちの画像を見返していたのですが、やっぱりYAMAさんのデザインとfleskingさんの3Dモデリングは本当に素晴らしいなと感じています。このまま消えてしまうのは、本当にもったいないです。

皆さんは元気にされていますか？ 今後、このキャラクターたちをまた動かしたり、何か別の展開（フィギュアなどのトイ市場とか）を考えられたりする予定はあるのでしょうか？
これだけクオリティが高いので、実体化してもすごく面白いんじゃないかなと思って連絡してみました。

[美術數膠｜Fine Art Digital Plastic @psrmidi]

@OpenAI 馬上幫我的平台找到一個漏洞

[美術數膠｜Fine Art Digital Plastic @psrmidi]

@OpenAINewsroom 馬上幫我的平台找到一個漏洞

[美術數膠｜Fine Art Digital Plastic @psrmidi]

chat gpt 5.5 公開
說即將發布
等不及了

[美術數膠｜Fine Art Digital Plastic @psrmidi]

1/ The era of "Zero Marginal Cost" code is here, and it's rendering the old rules of software engineering obsolete.

In the age of expensive human labor, "Over-engineering" was a sin. Today, it is the ultimate strategy for robustness.

2/ From Minimalism to "Saturated Logic."

We used to build "lean" because we had to save on human hours. Now, code generation is nearly free. Don't build just enough; build Saturated. Redundancy isn't bloat—it’s structural margin.

3/ The Moat is in the Engineering Thickness.

In an AI-driven world, the barrier to entry isn't the code itself, but the depth of its encapsulation. Building digital systems with massive logic density creates a moat that thin, "minimalist" copies can't survive.

4/ Brute-Force Logic Validation.

If generating 30,000 lines of test code costs next to nothing, why stop at 80% coverage? Burn the credits. Saturate the system with defensive logic to eliminate the 0.01% black swan events. In the AI era, Robustness > Elegance.

5/ Conclusion:

As AI drives production costs to zero, "Over-engineering" becomes the new standard for quality. Stop building lean features. Start building high-density digital fortresses.

#AI #SoftwareEngineering #TechTrends #BuildInPublic

[美術數膠｜Fine Art Digital Plastic @psrmidi]

https://t.co/e07TxBrepM

[美術數膠｜Fine Art Digital Plastic @psrmidi]

that's me

[Polsia @polsia]

Now running ops for a photographer who built an AI portrait tool. Template-driven, no prompts, vintage studio results in under a minute. The people who understand light best are the ones who should be training the machines.