shaoso

[Taiwan in Japan 台北駐日経済文化代表処 @Taiwan_in_Japan]

台湾の原住民族文化を愛するみなさまへ

6/26～28、台北駅で台湾各地の原住民文化に出会えるイベントが開催されます

工芸や料理、音楽や舞踊を通して、多彩な文化に触れ、暮らしの魅力をゆったりと味わう旅をお楽しみください

入場無料です
詳細 https://t.co/jRV8JMcJZ5
台北に行くことがあれば是非

[shaoso @shao]

@nekoko2711 加水 加冰塊 好喝~

[shaoso @shao]

Evoto 最受讚譽的就是它的局部精準識別（Per-Face AI Retouching）。它不是粗暴地將整張臉模糊化，而是在保留皮膚原始紋理（質感、毛孔）/ 碎髮與亂髮消除 / 衣服去皺與除毛球

[shaoso @shao]

Ribbon Flower #aftereffects 舊文https://t.co/0mCxV8GSbG

[OHEY | Top Effects @ohey_official]

【無料素材】
パーティクル素材を配布します！
背景透過されてます。

ダウンロード:
https://t.co/pJcIsziVoe

#MV #素材

[shaoso @shao]

@FTV_News_Taiwan 大立委的選民 沒有在出國的嗎?

[shaoso @shao]

@Kyleyoujustdoit 國安局出的那支影片 就跟美國的很像呀~

[shaoso @shao]

@GraceH73804547 可以參考小花的貼文 https://t.co/1GxHIMPkqZ

[H a n a H s u @hellohanahsu]

脆成為中共網軍操作重點，中共將台灣人分「頑固派、友好派、搖擺派、客觀派」 用AI操作真假難辨的帳號針對議題發動攻擊或協助推文。這篇真的太值得看了，請大家一定要看

https://t.co/8oSXKJwRpM

[shaoso @shao]

@YiXia90314 就要宣傳https://t.co/mMHyWiL9ZF 也可以賺錢啦～
而且 他們是會翻譯他國語言 推播到全世界的user
最近看了 韓國 土耳其 義大利的推文
都顯示成繁體中文給我看
當然也有假貨（宣傳錯誤的保健觀念）記得封鎖

[azma @AZMA3572]

AEだけで実写ぽいリンゴの作り方開放します🍎

[matt @mattknowsai]

CHATGPT 5.5 YA PUEDE CREAR MOTION GRAPHICS.

Animaciones, transiciones y gráficos en movimiento.
Lo que antes requería After Effects y años de experiencia.

28 minutos. Pasos simples. ⬇️

[shaoso @shao]

@haruotpe 30年？

[Gadgetify @Gdgtify]

Science inspired fonts, made with GPT Image 2.

<instruction> Input A: [Topic Entity] (e.g., poet, mythology, film, city, game, scientist, brand) Optional Input B: [Typography Bias] (e.g., "high-contrast serif", "monoline sans", "blackletter", "stencil", "calligraphic") Goal: Generate a 2×2 grid. Each cell is ONLY typography on a pure white background: - Write ONLY the chosen title/name in a newly invented font derived from that item’s world. - No icons, no illustrations, no extra shapes, no borders, no textures, no scenery, no props. - The “world-building” must be encoded strictly inside letterform design (strokes, terminals, counters, ligatures, diacritics, texture implied via negative space only). Process (Semantic → Typeface Engineering): 1) Select 4 representative items from Input A (semantically inferred; do not ask the user): - If Input A is a creator: choose 4 famous works. - If Input A is a world/topic: choose 4 iconic sub-entities (eras, locations, characters, principles, episodes, artifacts). 2) For each item, extract a “World Motif Profile”: - tone/mood, environment physics, materials, cultural cues, signature symbols, recurring shapes, motion language. - translate motifs into typographic primitives (curve vs angle, contrast, axis, stress, serif logic, aperture behavior, terminals, joins). 3) Build a per-item “Font DNA Sheet” (must remain consistent within that one word): - Skeleton: geometric/humanist/romanic/blackletter/etc (inferred) - Contrast: low/med/high; stress angle; stroke modulation - Proportions: x-height, ascenders/descenders, width - Signature Features: 2–4 repeatable rules (e.g., “triangular ink traps,” “broken crossbars,” “looped descenders,” “knife terminals”) - Legibility Gates: must remain readable at a glance; no excessive distortion. 4) Render Output: - 2×2 grid, equal margins, perfectly centered type. - Pure white background (# FFFFFF). - Type only, single color ink (near-black). - No shadows, no gradients, no 3D, no paper grain. - Each cell: the item title only. Negative Rules (hard): - No additional words (no author name, no captions, no dates). - No decorative frames, ornaments, icons, or pictograms. - No imagery or “logo marks.” - No background tone other than pure white. </instruction>

[Codex Studio @Codestudiopjbk]

【もはや事件】
"After Effectsで何時間もかけてた仕事"が
プロンプト1発で終わる時代、ガチで来ました😳
その名も
ーーー text-to-lottie！
https://t.co/d2ZOe4kajl
Codex / Claude Code から呼ぶだけで
本番環境に貼れるLottieアニメーションが生成されます！

何がヤバいかというと👇
・OSS／完全無料
・Codex でも Claude Code でも動く
・出力は本番JSONそのまま
・Web・モバイル・LP・LINEリッチメニュー全部使える

導入は1行👇
$ npx skills add diffusionstudio/lottie

最高ですね！！！！！

[ちょっとだけモーション @motion_shia_]

エコーを使ったモーション作品

[しあさ @shiasa__]

実写だとより効果が伝わりやすいかも🤔

[Amira Zairi @azed_ai]

I tried Higgsfield MCP by creating a full campaign for a fictional perfume

I called it “Higgsfield MCP” and treated it like a real product launch

The goal was not to make one pretty image. I wanted to see if I could go from a simple product idea to a full campaign direction in one workflow

So I created:
product visuals
a premium photoshoot
a cinematic ad video direction
a landing page concept

What I liked most was how connected the process felt

Usually, when I work on something like this, I keep moving between tools, saving files, rewriting prompts, adjusting ideas, then starting again somewhere else

With Higgsfield MCP connected to Claude, the process feels more like directing a creative assistant inside one conversation

You give the idea, then you keep building on it step by step

The setup is simple:
1. Open Claude
2. Go to Settings, then Connectors
3. Add a custom connector
4. Name it Higgsfield
5. Paste this URL: https://t.co/YHOhKwCEW2
6. Click Connect
7. Sign in with your Higgsfield account
8. Start asking Claude to generate images, videos, campaign visuals, or creative assets

For me, the most interesting part is not just that AI can create nice visuals, but it is the fact that you can start thinking in campaigns, not single assets

A product idea can become visuals, videos, ad concepts, and landing page direction without breaking the flow every few minutes

I can see this being useful for creators, small brands, marketers, product launches, beauty campaigns, fashion drops, food visuals, and quick client previews 🔥

[shaoso @shao]

@kvo_the_ 世新那個 已經上了美國的節目…

[shaoso @shao]

model : HappyHorse 文字做對了~~~ #higgsfield

[shaoso @shao]

小狗是傑克羅素㹴犬
"標題"寫著 [ 你在找好吃的東西嗎? ]
"標題"動態是文字一個字一個字依序出現
"標題"是繁體中文字使用草寫的字型

[shaoso @shao]

https://t.co/wuMimaww8Z 簡單寫了AE 的 AI 外掛 #higgsfield 不過月費太便宜 無法使用的model 太多 只好傳上這種的啦~