Louay Studio Ai

A library of creative sparks for TikTok, Reels, Shorts, and every AI video tool in your kit.
Louay Studio

[Louay Studio Ai @LouayStudio]

🚨 STOP PAYING FOR YOUR TikTok ADS OUT OF YOUR OWN POCKET!

Are you spending money on TikTok advertising?

You may be able to take advantage of an eligible TikTok Agency Account with an advertising cashback offer and receive ad credits based on the offer terms and account eligibility.

💰 OFFER EXAMPLES:
• Spend $200 → Get $200 in ad credit
• Spend $500 → Get $500 in ad credit
• Spend $1,000 → Get $1,000 in ad credit
• Spend $6,000 → Get $6,000 in ad credit
🔥 ACCOUNT FEATURES:
✅ TikTok Agency Account created in your name
✅ Your own login credentials
✅ Access to eligible advertising offers
✅ Support in case of technical issues
✅ Limited availability
⚠️ Offers are subject to TikTok’s terms, conditions, and account eligibility. Offers may vary by account and region.
📩 Want to know the details and how it works?
👉 Check the offer & register here
https://t.co/XMRwjewxtK

#TikTokAds #TikTokMarketing #DigitalMarketing #SmallBusiness #Entrepreneur #USA #USABusiness #Advertising

[Maverick | AI @RizwanAly07]

LOOK WHAT I JUST FOUND!

Fruit cartoons. Soap-opera drama. Entirely made with AI. Talking fruit with fake identities, messy secrets, and crazy plot twists. 😂🎬

I kept seeing these faceless pages on X and IG, and thought: Could I make this myself?

So I opened Pollo Agent, selected Story Video, and typed a simple idea:
“Who’s the real heiress—the banana or the strawberry?”

That became my own “Fake Heiress” cartoon. No animation team. No camera. Just an idea.

The plot?
🍌 An arrogant banana faking her identity
🍓 A mistreated strawberry—who’s actually the real heiress
🤯 And one final twist that changes everything.

If you’ve wanted to try a faceless animation page without the production headache, start here.

Try @itsPolloAI via the link below and bring your first fruit character to life 👇

[Clandestine @Classifiedv4]

these ai cat animations got over 17M views

their page is already at 82.9k followers and are already selling merch

same idea every single video

> cute cats at the start
> viral sounds people already heard
> them lip syncing to it

tiktok algorithm gives them a new topic every few weeks naturally without them inventing a new one

why it works is

> almost everyone loves cats
> this animation style is catchy
> they use already trending audios

if you’re building a tiktok page, this style is completely automatable and it costs like 0.25$ per video

i’ll drop the prompt in the first comment

[Louay Studio Ai @LouayStudio]

@safa_designerr شكرا راح اجرب وارسلك نتيجة

[Turki @nevermind_turki]

سويت اداة تحرير واضفتها لنفس المستودع ، كانت من اكبر المعاضل في سلايدات HTML ، الاكيد بيكون في ملاحظات أسعد بها ، اذا سبق ثبت المهارة اطلب من الايجنت تحديثها فقط او ارسل له الرابط مرة ثانية ، فيها المهرة والمحرر.
https://t.co/JZOVuqHpqA

[Renoise AI 日本公式 @renoiseaijp]

Seedream 5.5の30日間使い放題がスタート🚀

クレジット消費なし、回数制限なし✨
思いきり作ってみてください

@renoiseai をフォロー＋元の投稿をリポストで
3,000円分のクレジットが当たるチャンス！

[Nene @Nene11x]

Este prompt convierte Fable 5.2 en una máquina de generar Shorts de YouTube.

PROMPT

"Monta un sistema de Shorts virales para mi canal faceless de YouTube. Vídeos 9:16 de 20 a 35 segundos para Shorts, Reels y TikTok. Usa el MCP de Sunra para todos los modelos de imagen, vídeo y audio. No me hagas preguntas. Mira primero mis créditos de Sunra y dime el presupuesto antes de generar nada.

Mi nicho es [TU NICHO]. Busca los 20 Shorts más vistos del nicho en los últimos 90 días y sácame el patrón: el gancho de los 2 primeros segundos, la duración, el ritmo de cortes, el texto en pantalla y por qué la gente se queda hasta el final. Guárdalo en patron.md.

Usa este formato: un objeto, una regla y una escalada. Por ejemplo "1 corte VS 7 cortes VS 13 cortes" con una sandía, o "Blando 0% VS 50% VS 100%" con una pelota. 5 fases de 5 a 6 segundos y cada una sube la apuesta. El texto grande de arriba cambia en cada fase. El final es el momento más satisfactorio y enlaza con el primer plano para que el vídeo haga loop.

Pon esto al principio de cada prompt de imagen y vídeo: render 3D hiperrealista, estudio con fondo liso (gris oscuro, melocotón o verde agua), luz suave desde arriba, sombras suaves, materiales con textura real, tabla de madera o base de mármol, cámara fija a la altura del objeto, física realista, cámara lenta en el impacto. Sin personas, sin manos, sin logos y sin texto dentro de la imagen.

Con Nano Banana Pro haz una imagen de referencia del set y otra del objeto. Úsalas en todas las escenas para que el objeto, la base y la luz no cambien nunca.

Haz cada fase con Seedance 2.5 usando esas referencias. Un solo movimiento por escena. Haz 2 tomas por fase y quédate con la que tenga la física más creíble y sin deformaciones.

Usa Kling 4.0 para los planos de acción: el impacto, el corte, la cámara lenta y la transición entre una fase y la siguiente.

El sonido es lo que hace que funcione. Genera con los modelos de audio de Sunra sonidos ASMR de cada impacto (corte, crujido, golpe, goteo) y ponlos justo en el frame del impacto. Debajo, una base suave que suba en la última fase. Nada de música con derechos.

Monta todo con ffmpeg. Texto en pantalla en blanco, sans serif en negrita, sombra suave, centrado arriba y cambiando en cada fase. Exporta a 1080x1920 y 30fps.

Haz la miniatura con Nano Banana Pro: el momento justo antes del impacto, con 3 palabras como máximo.

Escribe en ideas.md 10 ideas con este mismo formato, cada una con título y descripción SEO. Genera el primer Short y enséñamelo antes de seguir. Si lo apruebo, haz los otros 9 y dime los créditos que has gastado y el coste por vídeo."

[Cliff Cross @hoops_vv]

HOLY SH*T. 💀

100 strikes. NOTHING happened.

Then the 101st hit changed EVERYTHING. 🤯

The rock wasn't breaking on the outside, but every single strike was destroying it from within.

Success works the EXACT same way.

HOW MANY PEOPLE QUIT JUST ONE STRIKE BEFORE THEIR BREAKTHROUGH?

[ASCIICraft @ASCIICraft_]

Radio Pixel: your music, playing inside living pixel art worlds. 🎧 Now on Google Play.

[Better Call AI @Leo_zhang2018]

兄弟们，又发现一个免费福利！每天白嫖 30GB 流量，还能用全球住宅 IP！

这个工具叫URnetwork，一款开源的去中心化网络 APP。

最吸引我的就是它的免费额度，每天提供 30 GiB 流量，第二天自动刷新，不用绑银行卡。

兑换码/邀请码可以填写下面这个

QUV6YZ

它还有几个挺实用的地方：

支持全球多个国家和地区的住宅 IP 出口。

支持 Windows、macOS、Linux、Android 和 iOS。

支持智能分流、自定义 DNS，免费版最多同时连接 2 台设备。

不需要自己购买 VPS，也不用折腾复杂的节点配置。

每天 30 GiB，按 30 天算就是 900 GiB 的免费流量。

对喜欢研究网络工具、住宅 IP 和免费资源的朋友来说，值得收藏试试。

[Sergio Gil @seergioo_gil]

WTF?! is going on. 90k followers on TikTok + 100k on Instagram in about 1 month

these Claude-made videos are BREAKING the internet right now

so i decided to replicate it:
1- sent Opus links to 3 of his videos
2- he used the Calliope connector to analyze them (he cannot download from IG)
3- generated a reel (took a while, and a bit of quota😖)
4- saved it as a template

everything was made in our cloud environment, including access to the whole ElevenLabs library, render, and reusability

this was the result

here you have a prompt so you can copy this stye too:

<inputs>
Ask me for my topic, or pitch 3 gym myths, each with one checkable number that flips what people assume (same load, two machines, completely different feel). Ask which voice I want and the length (45-60 s). Then write the narration (one idea per sentence), a beat sheet (which number, label or stamp lands on which word) and the math behind every number. Show me all of it, with the cost of the voice and the render, before you generate anything.
</inputs>

<direction>
A 1080x1920 vertical flat-vector gym-science explainer, drawn entirely in code (SVG + GSAP). No image or video generation. Every frame is an annotated diagram: one hero plus one to three labels.
- Set: light warm-grey grid (#E6E6E1, 48 px cells), one ground line #2E3038 at y=1300. Machines sit on it with a soft contact shadow.
- Style: heavy outlines #24252E 6 px, flat fills. Machines in blue-grey #5B6275 and gunmetal #3F404C; plates near-black #2B2B33; pads dark navy.
- People: oval heads, dot eyes, round-capped limbs. You: blue tee #2F7BF5, navy shorts, white trainers. The rival friend: orange tee #F57C2B, darker skin, red trainers. Same colours in every scene.
- Callouts: white box #F7F6F1, heavy outline, Outfit 800 ~80 px ("100 KG"), plus a JetBrains Mono label a beat later. Colour = meaning: black for facts, red for the felt or real value, orange-brown for your own share, green for confirmed.
- Kit: red angle arcs with a tag, amber range bars with end caps, thick force arrows (red load, amber effort), a translucent amber highlight over the body that matters, red rubber stamps (double border, -10°, distressed) that slam in from 150% with a screen shake, speech bubbles, and a topic chip top-right that changes with the subject.
- Captions: near-black pill at 72% height, Outfit 800 62 px, two to four words; each word grey until spoken, then white. First group visible on frame 0 so the thumbnail has text.
- Motion: a new beat every 1-2 s on the key word. Numbers appear the moment they're said and count up or down. Pops go 70% → 105% → 100% in 0.2 s. Nobody freezes.
- Other text between y=168 and 1330. No logos, emoji or UI cards.
</direction>

<structure>
The voice sets the length (~165 wpm). Example beat map for ~60 s:
0 s hook: two machines side by side, same "100 KG" on both, then each callout recounts live to what you really feel (70 vs 150) →
~7 s "only one doesn't lie": a third machine under a sheet with a big "?" →
~10 s push in on machine 1: angle arc, range bar, force arrows, the felt number in red →
~17 s pan to the honest one: the sheet flies off, "100 KG = 100 KG" in green →
~23 s machine 2: a steeper arc, then the trick: a stamp slams, your own body lights up amber riding the same slope, your weight recounts to the share you actually lift →
~37 s cut to a data scene: a dial gauge whose needle sweeps and wobbles, recounting live as a small slope tilts between two angles →
~43 s the odd one out (a lever machine): rotation arrows, a number that spins randomly →
~47 s comparison bars against a dashed "loaded" line →
~51 s punchline: the lifter wrecked (sweat, strain ticks, red glow on the muscle) →
~54 s CTA "send it to that friend": he loads a huge number on the easy machine, plates dropping in, then fails on the hard one with a speech bubble and a final stamp. Hold 0.8 s after the last word.
Hard cuts only into and out of data scenes; everything else is camera moves.
</structure>

<build>
1. This assumes a Calliope-style sandbox: GSAP, narration with word timestamps, checks, snapshots and rendering. Read the environment's docs first; they win over this prompt on technical details. Outside it, install Node, GSAP, Chromium/Playwright and FFmpeg, and tell me.
2. Voice first. Read the real duration and word timestamps, and plan every beat, counter and stamp from them. If the voice runs shorter than planned, plan to the voice.
3. Project: a root index.html (1080x1920) with one "world" composition holding every machine side by side and a camera that pans and pushes between them; data scenes as opaque full-frame overlays above it; header and captions as their own compositions; the voice at the root.
4. Each composition registers one paused GSAP timeline with a single clock tween whose onUpdate calls render(t). Camera, rep cycles (cosines), counters, pops and shakes are pure functions of t: no Math.random, timers or network.
5. Rig: limbs are round-capped double lines (outline, then fill). Legs use two-bone IK from the hip to the foot on the moving platform, with a preferred knee direction. Sleds are rigid groups sliding along the rail vector.
6. Callouts are HTML pinned to world points projected through the camera, so text keeps its size while the camera zooms.
7. Captions: split the groups by hand at phrase boundaries, never ending on a lone article or preposition.
8. Before rendering: run the checks, snapshot ~20-30 moments into contact sheets. Fix overlaps, covered faces and empty first frames. Render, then verify the MP4 (size, fps, audio stream, loudness) and inspect frames from it. Say plainly what you couldn't check, and offer to save the project as a reusable template.
</build>

<gotchas>
- Read the font report after saving: missing families fall back silently. Pick one the environment resolves.
- Shoulder pads drawn over the head hide the face: put them on a layer behind it.
- Keep body highlights under 50% opacity, or the face disappears under the glow.
- Anything that pokes out of one machine's cell (falling plates) shows up in the next shot.
- Move the camera only while an overlay fully covers it, inside its frame window, or one frame leaks.
- An element positioned by its anchor needs translate(-50%,-50%), or it sits off-centre.
- Do the math: a 45° sled carries ~71% of the plates along the rail, 60° ~87%; on a hack your own body rides the slope too.
</gotchas>

<start>
Ask me for my topic, or pitch 3. Show me the script, beat sheet, math and cost before you spend anything.
</start>

[caesar @caesarscales]

This map geography style niche is evergreen and INSANELY viral

this can be automated entirely with claude its pulling in millions of views

prompt soon ? 😉

[Louay Studio Ai @LouayStudio]

@Jerryxaut @DDJCXX PROMPT

[Louay Studio Ai @LouayStudio]

@withbassam اي تطبيق تربطه معه ليدمج صور وصوت ليصنع فيديو

[Explee @explee]

someone asked grok about trying us. that screenshot is the reply they got.

you paste your website link into Explee, it finds companies that need you, writes each a personal email, and replies to interested buyers with your booking link.

$30 free 👇

https://t.co/5vVKkCxdxH

[ElevenLabs Developers @ElevenLabsDevs]

Introducing the ElevenLabs MCP, now available in Claude.

Your team can manage your voice and chat agents where you already work - review recent performance, create new agents, update configurations, and even estimate LLM costs before changes go live.

[硅基废话 @DDJCXX]

卧槽，3628 万播放，就是一个圆圈里的小球越变越多。

IG 上的 NexusBox Lab 一共才发了 19 条，4 万粉。TikTok 上 25 条，加起来 5292 万播放。

这种视频是代码算出来的物理模拟，不用拍，也用不上 AI 视频模型。大家都在等装满炸开那一下，所以会一直看到最后。

我们让 Claude 从零写了个同款，小球和碰撞声全是代码算的，一个素材没用，一条 45 秒的视频电脑跑 1 分多钟。

做成了免费的 Skill，放评论区，Claude Code、Codex 都能装👇

[빚갚는미친개 @mad_dogdebt]

이 AI 자동화 유튜브 계정 진짜 죽인다.

개인적으로 자동화 템플릿을 굉장히 잘 짠 것 같음.

계정도 보니까 2개를 동시에 운영 중임.

하나는 ‘역사한줌’, 하나는 ‘세계사한줌’.

조회수가 상당함. 한국사 채널은 71만, 58만 조회수 영상이 있고, 세계사 채널도 59만, 33만 조회수가 나옴. 그것도 롱폼.

영상 제작 방식도 흥미로움. 초반 1분 정도는 살짝 움직이는 AI 이미지로 시선을 잡고, 이후부터는 정적인 이미지 위주로 진행함.

아마 영상 생성에 들어가는 토큰값을 절약하면서도 시청 지속시간을 확보하려는 전략 같은데, 이게 진짜 영리해 보임.

주제 선정도 좋음. 역사라는 무한한 소재에 ’왜 그랬을까?’라는 호기심을 자극하는 제목까지.

대본만 잘 뽑으면 콘텐츠를 계속 찍어낼 수 있는 구조임.

돈이 많이 들어가는 구간은 초반에 집중하고, 나머지는 저비용으로 제작하는 방식.

AI 자동화에서 중요한 건 무작정 퀄리티를 높이는 게 아니라, 시청자가 이탈하지 않는 선에서 제작 단가를 얼마나 낮추느냐인 것 같음.

물론 이것도 롱런. 롱런이 가능할까 궁금함.

[Mohammed Saqallah @mohmdsaqallah]

انعمل باستخدام Grok bot

ارفع اي ملف صوتي من تقرير اخباري وارفق معه ملف التوجيهات اللي تحت التغريدة :