Clandestine (@Classifiedv4) header
Clandestine (@Classifiedv4) profile picture

22 scaling my own AI ugc agency currently at $15k/mo

Joined January 2025
Photo by Clandestine (@Classifiedv4)Photo by Clandestine (@Classifiedv4)Photo by Clandestine (@Classifiedv4)Photo by Clandestine (@Classifiedv4)Photo by Clandestine (@Classifiedv4)Photo by Clandestine (@Classifiedv4)Photo by Clandestine (@Classifiedv4)Photo by Clandestine (@Classifiedv4)Photo by Clandestine (@Classifiedv4)Photo by Clandestine (@Classifiedv4)
ultimate free sauce right here
just found a banger gta 6 format, 5M views on his first 4 videos ill save u a $100 course and make ur claude create this fully automated for $0 give my prompt to ur claude code thank me later <inputs> Ask me for, in one message: 1. The subject (a game, a phone, a car, a film, a city, an app) and the angle, picked from: "5 crazy things you can do in ___ that you couldn't in ___", "5 realistic things in ___", "how ___ works in ___" (one system explained), "everything you need to know before ___". 2. The voice: an ElevenLabs voice id and key if I have them, otherwise a free local voice. If it is paid, tell me the character count (about 1 credit each) and wait for my yes. 3. A music bed if I have one; otherwise pick a free one with no vocals. If I have no subject, pitch me 3 that people are searching for this week and say why each one is hot. Research fresh on the day; never copy another Short's script or order. </inputs> <direction> A 1080x1920, 30 fps YouTube Short / TikTok / Reel of 60-68 s, built entirely on real footage of the subject (official trailers, gameplay, product films), reframed to 9:16. No face-cam, no AI-generated footage. It is relentless: the voice runs at 240-265 words a minute with no pause longer than 0.3 s, a music bed sits about 7 dB under it the whole time so there is never silence, and the picture changes 85-125 times a minute (a new shot or overlay about every 0.5-0.9 s). Text system, all heavy italic sans in uppercase (Montserrat Black Italic) with a 6 px black stroke and a soft shadow: - Top block (y 5-20%): the subject's logo, centred, about 34% of the width. Under it the title of the Short or the current item. - Item header: a green number ("1.") plus the item in white ("THROW DOG POOP") and one emoji-style icon under it, held for the whole item. - Captions (y 70-78%, centred, max 3 words a line, 2 lines): every spoken word appears on its own syllable, white, with the key word of each phrase in neon green #39FF3C for good or exciting and red #FF2E2E for bad or impossible. Every concrete claim gets a mock-up layered on the footage within 0.5 s: a game or app UI element rebuilt in HTML (a star meter, a weapon wheel, a button prompt, a minimap, a phone screen), a big counter that ticks up to the number ("600,000 ANIMATIONS"), a simple map with an animated route or zone, or a tiny cartoon illustration of the action with a comic burst. Each sits in a dark rounded box with a thin green border and a small "ILLUSTRATION" tag when it is not real footage. Banned: a greeting, "hey guys", the channel name, an intro longer than 3 s, any pause in the voice, any frame without text on it, letterbox bars, a 16:9 shot pasted small in the middle, a shot held over 3.5 s, the same shot twice, and any claim with no source behind it. </direction> <structure> Length: 60-68 s, 250-290 words. The voice starts on frame 1. 0-3 s, the hook: "Here's five ___ things in ___ that ___" (or "The ___ in ___ has been fully revamped") over the single most absurd or best-looking clip of the whole Short, with the title building word by word in the top block. Show the comparison visually when there is one (the old version's logo popping up and getting crossed out in red). 3-60 s, the items: five items of about 11-12 s each (45-55 words), always in this shape: "Number one, [label]." (the item header slams in on the word) → the claim in one short sentence → the most specific concrete example, told like a story ("You wait for somebody walking their dog, you hit them so they drop the bag, then...") → what it changes for you ("So you've got to change clothes, switch cars, or split up.") → sometimes a one-line reaction ("No, really." / "It's basically a heist movie."). Order: the most viral item first, the most important second, the strongest visual last. For a one-system explainer, replace the items with 6-8 rules of the system in the order a player meets them. 60-65 s, the ask: one question that invites a comment ("So which one are you going to try first?") over the hook clip again. The loop: the last words are "And in case you missed it," and they cut, mid-sentence, straight into the first frame, so the end flows into "here's five ___ things..." and the Short replays seamlessly. Voice rules: sentences of 6-12 words, plain words, "you" in almost every sentence, numbers said the way people say them, contractions everywhere, no filler. </structure> <build> 1. Tools (all free, all local). Install what's missing and tell me what you installed: Python 3.10+ with numpy and Pillow, FFmpeg with ffprobe, yt-dlp, Playwright with Chromium (for drawing the text and mock-up layer), faster-whisper large-v3-turbo (int8, CPU), Montserrat Black Italic, and Kokoro if I have no ElevenLabs key. 2. Research: search the web and open the pages. For each item keep the claim, the source, the exact words and the date. Use only what the maker showed or said, or what a named outlet saw in a demo; say "reportedly" out loud when it is one outlet's report, and never use leaks. Show me the five items with sources before writing. 3. Script: write it to the structure, read the word count, and cut until it is 250-290 words. Write out how the voice should say names and numbers ("GTA 6" as "G T A six", "600,000" as "six hundred thousand"). Show me the script before recording. 4. Voice: one take for the whole Short. Transcribe it with faster-whisper and match the words to the script with difflib; take every time from that, never from the voice service's timestamps. Cut every pause down to 0.12-0.25 s, only inside real silence, and speed up by at most 1.12x if it is still under 240 wpm. Level to -15 LUFS integrated, music bed 7 dB under the voice, true peak under -1 dB. 5. Footage: the maker's own uploads only. Split each source into scenes with FFmpeg's scene score, make a numbered sheet, and look at it; drop logos, rating boards, title cards and anything intimate or graphic. For every scene, pick the subject's position on a contact sheet and store a 9:16 crop window (full height, 56.25% of the width) centred on it; let the window drift a little with the action, never jump. You need 60-90 different shots per minute: pull from several trailers and trim each scene into 0.5-1.5 s pieces, and for every item pin the exact shot that shows what the voice says. 6. Timeline, every time keyed to a word: a new shot every 0.5-0.9 s (median about 0.8 s, nothing over 3.5 s), cutting on stressed words. The item header lands on "Number one" and stays until the next item. Captions: each word appears 0.03 s before it is spoken and the line clears at the end of its phrase or at a cut. Mock-ups come in within 0.5 s of the claim they prove, scale up from 92% with a quick overshoot over 0.15 s, and hold until the next claim. Counters tick from about 10% of the value to the value in 0.6 s. 7. Draw the whole text and mock-up layer as one transparent HTML page, export it frame by frame with Playwright, and lay it over the cropped footage with FFmpeg. Render a 540x960 preview first, pull 4 frames a second from the first 15 s and from one 15 s window later, and read every sheet: text never over a face, captions in sync, every claim proven on screen, no bare frame. Then render 1080x1920 and measure: cuts per minute, median shot, words per minute, longest pause, loudness, and the loop seam (play the last 2 s into the first 2 s). Show me the numbers next to the targets and fix what is off. </build> <gotchas> - A 9:16 centre crop of a 16:9 trailer cuts the subject out of half the shots. Pick a crop window per scene from a contact sheet, never one fixed crop. - Trailers are cut for mood, so the most beautiful shot is often the wrong picture for the claim. Pin the shot that literally shows what is being said. - At 250+ wpm the voice must still sound like a person. Tighten pauses first; past a 1.12x speed-up it sounds like a chipmunk. - Speech-to-text writes "80 bucks" as "$80" and respells names. A missing or different word is a real error; redo the take. - Voices read "6" as "six" and "600,000" as digits unless you write it out. Respell for the voice; keep the screen text as written. - If the loop seam has a gap or a level jump, the replay feels broken. The last word must end within 0.05 s of the cut, and the music must not restart. - Age-gated gameplay uploads cannot be fetched without a login. Use the open trailers, the maker's site and official screenshots. - Re-uploads carry other channels' watermarks. Check the channel name before downloading. - Green and red keywords lose their meaning if every word is coloured. One coloured word per phrase, at most. - A mock-up of a game UI must look like the real one in spirit but carry an "ILLUSTRATION" tag; never pass it off as a real screenshot. - Exporting the text layer is about 2,000 frames per minute of Short; render the preview first. - This is commentary over quoted footage. Keep a source list and let me decide what gets published. </gotchas> <start> Ask me the three input questions in one message. Then research, and show me the five sourced items and the script before you record or download anything. </start>
1
11
363
i just found gold this channel was created less than 2 weeks ago > 1.7M+ views > 3k subs i recreated their entire format for 0$ using Claude automation and i’m leaking it for you to steal, thank me later ⬇️ Prompt: Build a ~60-70 s vertical (9:16) explainer short in the style of the YouTube channel @zero_effort_lab. Topic: "How does a tunnel boring machine get out of the tunnel?" STUDY THE REFERENCE FIRST - Their top short: "How Do They Get Cranes Down From Skyscrapers?" (1.4M views, 75 s). - LOOK: clean low-poly flat-shaded 3D (simple boxes and cylinders, Unity/Blender toy style), bright blue sky gradient with soft clouds, pale grey city blocks, simple trees, toy-like machines in bright yellow / orange / teal. - CAMERA: gentle moves, wide establishing shots, cutaways and close-ups of the mechanism. Average shot ~3 s. Simple overlays: red "X", red arrows and lines, small labels. - CAPTIONS: 1. A small bold keyword line near the top, white caps with ONE word highlighted yellow or red ("THE CLEVER PART", "HYDRAULIC JACKS", "STUCK", "WHAT'S NEXT?"), shown only on key beats. 2. A white subtitle line with a soft dark outline at ~75% height, showing the current 2-5 word phrase, synced to the voice. No full stops in captions. - SCRIPT STRUCTURE (theirs): hook question -> "To see the problem, let's go back to the start" -> how it works -> "But at the very end, there's a problem" -> "So here's the clever part" -> the solution step by step -> a twist / punchline -> tiny summary ("one skyscraper, three cranes") -> "So what should we explain next?" ~185-190 words per minute, plain curious narrator. SCRIPT (180-210 words, factual, no invented specifics) - Hook: a TBM dug this whole tunnel, so how does it get out? - Start: it can't drive in; it's lowered piece by piece down a launch shaft and built at the bottom. A rotating cutter head digs, a conveyor carries the dirt away, and concrete rings line the tunnel behind it. - Problem: at the end it's deep underground, and the rings are narrower than the cutter head, so it can't reverse out. - Clever part: a second (reception) shaft is dug where the tunnel ends; the machine breaks through and a huge crane lifts it out in pieces, often refurbished and reused. - Twist: sometimes there's no room for a shaft (under the sea), so it's cheaper to leave it. In the Channel Tunnel the British machines were steered out of the way and buried in concrete. - Close: "one shaft in, one shaft out, or no way out at all" -> "So what should we explain next?" SHOTS (~20-26, cut every 2.5-3 s, chunky readable shapes at phone size) city cutaway with the tunnel under it / launch shaft with a crane lowering pieces / the TBM built at the bottom / spinning cutter head close-up / conveyor / concrete rings going in / red lines: rings narrower than cutter head + X on reversing / THE CLEVER PART: reception shaft, breakthrough with debris, crane lifts the cutter head out / on a truck, used again / under the sea: X where a shaft can't go, machine steered down and buried in concrete, ferry passing over / summary diagram / WHAT'S NEXT? VOICE + AUDIO - Voice: Kokoro-82M (free, local), American male "am_michael", ~190 wpm, pauses <= 0.25 s. Word timings from faster-whisper for the captions. - Music: light free track (Mixkit "Curiosity", free licence, no attribution), ~19 dB under the voice. - SFX: downloaded Mixkit free sounds only (pop on the X marks, digging, whoosh, rocks falling, truck, cement mixer). - Mix: -14 LUFS. OUTPUT: 720x1280, 30 fps, H.264 + AAC, 55-75 s. QA: check frames around every cut (no broken geometry or flashes), captions never early, text matches the voice.
13
94
4,327
Another channel made after my prompt was posted and already making $$$
i recreated this entire channel in one prompt on opus 5.5 > +20M views > +2M likes > +100k followers completely automate-able workflow and i’m giving away the prompt for free and you can steal it [References] <Man> corresponds to @Image 2. Use only his appearance: bald round low-poly head, simple dot eyes, short brows, orange t-shirt, seatbelt. <Woman> corresponds to @Image 1, the passenger on the left. Use only her appearance: shoulder-length wavy brown hair, purple t-shirt, seatbelt. <Teal Car> corresponds to @Image 1 and @Image 3. Use its body, color and interior. <Fries Bag> corresponds to @Image 4: a plain brown paper bag with a plain red fries carton full of golden low-poly fries, no printing. @Image 3 defines the parking-lot location and the exterior shot: the car parked between white lines in front of a beige fast food building with a red awning. @Image 1 defines only the windshield two-shot framing, the seats and the characters' positions. Do not use its drive-thru menu board or its brick wall: in this video the background through the windows is the parking lot and the beige building from @Image 3. @Image 2 defines the medium close-up framing of <Man> behind the wheel: his whole head, shoulders, seatbelt and both hands on the steering wheel in frame, with the car roof and windows around him. Do not crop tighter than @Image 2. @Image 4 defines the close-up framing of <Fries Bag> on <Man>'s lap. There are exactly two people in the car: <Man> in the driver seat on the right of the windshield shots, <Woman> in the passenger seat on the left. Do not swap their seats. [Visual Style] Low-poly faceted 3D toy render exactly like the references: flat soft even daylight, matte materials, saturated warm colors, bright clear blue sky, chunky low-poly trees. Simple stylized animation: small head turns, blinks, brow changes, chewing and hand moves only. No realism, no film grain, no depth-of-field blur. No text, letters, logos, captions or speech bubbles anywhere. [Shots] 0-3 seconds: Shot 1, the exterior from @Image 3. <Teal Car> sits parked and still. End state: the car is parked, both people visible small through the windshield. 3-5 seconds: Shot 2, hard cut to the windshield two-shot. <Man> holds <Fries Bag> on his lap and opens its top. <Woman>'s hand is already inside the bag. End state: her hand is in the bag, he stares down at it. 5-10 seconds: Shot 3, hard cut to the close-up from @Image 4. <Woman>'s hand pinches fries out of the carton, pulls them out of frame, then comes back for more. End state: her hand is back in the carton. 10-12 seconds: Shot 4, hard cut to the windshield two-shot. <Woman> chews happily with her eyes closed and a big smile, holding fries; her mouth moves as she talks. <Man> stares at her flatly. End state: she is still chewing and smiling. 12-15 seconds: Shot 5, hard cut to the close-up of <Man> from @Image 2, the parking lot behind. Flat deadpan face, mouth closed, eyes slowly narrowing. End state: he looks unimpressed. 15-20 seconds: Shot 6, hard cut to the windshield two-shot. <Man> slowly reaches toward <Fries Bag>. <Woman> snatches the whole bag and hugs it to her chest, turning away from him, her mouth moving as she talks. End state: she holds the bag against her chest, his hand hangs empty in the air. 20-24 seconds: Shot 7, hard cut to the close-up of <Man>. He lowers his empty hand back onto the steering wheel and stares straight ahead, completely deadpan. Mouth closed. End state: both hands on the wheel, staring forward. [Maintain Consistency] Keep both characters' faces, outfits, seats and seatbelts, the teal car, <Fries Bag>, the parking lot background and the bright daytime lighting identical in every shot. The camera never moves inside a shot; only hard cuts change the view. (No music)
3
11
150
7,813
Masterclass…
Grand Theft Auto VI is now only one week away! Thank you to everyone in the community for your incredible support. There’s still time to pre-order and pre-load so you’re ready on launch day. Code-in-box versions are also available now for pick-up at retailers worldwide.
7
1
168
9,249
1.7M F*CKING VIEWS ON ONE VIDEO USING THE PROMPT I POSTED YESTERDAY 🫩
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
2
5
39
3,083
this is actually a money printer another one was averaging below 10k and went +300k on the first automated video he made
HOLY F*CK 2 days ago and +200k views with 2k subs creating fully automated videos with claude 3rd channel i saw this week btw
25
1,666
HOLY F*CK 2 days ago and +200k views with 2k subs creating fully automated videos with claude 3rd channel i saw this week btw
4
1
24
3,013
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
1
3
1
34
4,450
3 image prompts, gpt image 2, 1:1, 2x2 grid each. the animation itself is code (cuts on the beat, mouths open on the audio) 1. black cat sheet: Hand-drawn 2D cartoon in a cozy indie animation style: thick slightly wobbly black ink outlines, flat uneven marker and coloured-pencil fills with visible paper grain, almost no shading, cute deadpan derpy faces, on plain warm off-white paper. The image is a 2x2 grid of four equal square panels separated by thin light grey lines, same plain off-white paper background in every panel, the character the same size and in the same spot in every panel so the panels can be swapped like animation frames. No text, no letters, no logo, no watermark, no speech bubbles. The character: a chubby round black cat with big round green eyes, a small white patch on its chest, a tiny pink nose, short stubby arms. Seen from the chest up, centred, facing camera, filling the lower two thirds of each panel. Panel 1 top left: mouth wide open mid-sentence, eyebrows raised, asking a question. Panel 2 top right: mouth closed, neutral flat stare. Panel 3 bottom left: small round 'o' mouth, curious. Panel 4 bottom right: mouth half open saying a word, one paw raised. 2. tabby in box sheet: Hand-drawn 2D cartoon in a cozy indie animation style: thick slightly wobbly black ink outlines, flat uneven marker and coloured-pencil fills with visible paper grain, almost no shading, cute deadpan derpy faces, on plain warm off-white paper. The image is a 2x2 grid of four equal square panels separated by thin light grey lines, same plain off-white paper background in every panel, the character the same size and in the same spot in every panel so the panels can be swapped like animation frames. No text, no letters, no logo, no watermark, no speech bubbles. The character: a fat lazy grey tabby cat with dark stripes and sleepy amber eyes, lying flat and squished inside a beat-up open brown cardboard box, seen slightly from above, the box and cat filling the lower two thirds of each panel, centred. Panel 1 top left: eyes closed, asleep, mouth closed. Panel 2 top right: eyes half open, bored, mouth closed. Panel 3 bottom left: eyes half open, mouth open saying a word. Panel 4 bottom right: eyes half open, mouth open wide, one paw lazily raised. 3. dance sheet (black cat + tabby sheets attached as refs): Hand-drawn 2D cartoon in the exact style of the reference images: thick slightly wobbly black ink outlines, flat uneven marker and coloured-pencil fills with visible paper grain, almost no shading, on plain warm off-white paper. The image is a 2x2 grid of four equal square panels separated by thin light grey lines, same plain off-white paper background in every panel, the camera and framing identical in every panel so the panels can be swapped like animation frames. No text, no letters, no logo, no watermark. The two cats from the references, the chubby black cat with green eyes and the fat grey tabby with amber eyes (out of its box now), both standing on their hind legs, full body, small in the frame, side by side, dancing happily under a single bare light bulb hanging on a long thin cord from the top edge of the panel. Black cat on the left, tabby on the right, both with eyes squeezed shut and big open-mouth smiles like they are singing. Panel 1: both arms up. Panel 2: both leaning left, arms out. Panel 3: both leaning right, one paw on hip. Panel 4: both arms down, hips swung, one leg kicked out.
1
3
673
Sent this to my Claude and it made this 😭
Make Original Ads Great Again
4
263
youtube is a F*CKING goldmine for selling digital products right now everyones is so focused on meta ads and SEO, while this is sitting here completely for FREE these channels do an average of 5M views per month and they both started around 2-4 weeks ago assuming only 5-10% convert to purchases, thats $250k-$500k/m + at least $50k from youtube monetization creating literal AI slop
8
16
258
13,334
holy f*ck this format is printing money and is still unsaturated this channel peaked 100k views before transferring to it What’s crazy is you can completely automate the production too
3
6
42
2,869
i recreated this entire channel in one prompt on opus 5.5 > +20M views > +2M likes > +100k followers completely automate-able workflow and i’m giving away the prompt for free and you can steal it [References] <Man> corresponds to @Image 2. Use only his appearance: bald round low-poly head, simple dot eyes, short brows, orange t-shirt, seatbelt. <Woman> corresponds to @Image 1, the passenger on the left. Use only her appearance: shoulder-length wavy brown hair, purple t-shirt, seatbelt. <Teal Car> corresponds to @Image 1 and @Image 3. Use its body, color and interior. <Fries Bag> corresponds to @Image 4: a plain brown paper bag with a plain red fries carton full of golden low-poly fries, no printing. @Image 3 defines the parking-lot location and the exterior shot: the car parked between white lines in front of a beige fast food building with a red awning. @Image 1 defines only the windshield two-shot framing, the seats and the characters' positions. Do not use its drive-thru menu board or its brick wall: in this video the background through the windows is the parking lot and the beige building from @Image 3. @Image 2 defines the medium close-up framing of <Man> behind the wheel: his whole head, shoulders, seatbelt and both hands on the steering wheel in frame, with the car roof and windows around him. Do not crop tighter than @Image 2. @Image 4 defines the close-up framing of <Fries Bag> on <Man>'s lap. There are exactly two people in the car: <Man> in the driver seat on the right of the windshield shots, <Woman> in the passenger seat on the left. Do not swap their seats. [Visual Style] Low-poly faceted 3D toy render exactly like the references: flat soft even daylight, matte materials, saturated warm colors, bright clear blue sky, chunky low-poly trees. Simple stylized animation: small head turns, blinks, brow changes, chewing and hand moves only. No realism, no film grain, no depth-of-field blur. No text, letters, logos, captions or speech bubbles anywhere. [Shots] 0-3 seconds: Shot 1, the exterior from @Image 3. <Teal Car> sits parked and still. End state: the car is parked, both people visible small through the windshield. 3-5 seconds: Shot 2, hard cut to the windshield two-shot. <Man> holds <Fries Bag> on his lap and opens its top. <Woman>'s hand is already inside the bag. End state: her hand is in the bag, he stares down at it. 5-10 seconds: Shot 3, hard cut to the close-up from @Image 4. <Woman>'s hand pinches fries out of the carton, pulls them out of frame, then comes back for more. End state: her hand is back in the carton. 10-12 seconds: Shot 4, hard cut to the windshield two-shot. <Woman> chews happily with her eyes closed and a big smile, holding fries; her mouth moves as she talks. <Man> stares at her flatly. End state: she is still chewing and smiling. 12-15 seconds: Shot 5, hard cut to the close-up of <Man> from @Image 2, the parking lot behind. Flat deadpan face, mouth closed, eyes slowly narrowing. End state: he looks unimpressed. 15-20 seconds: Shot 6, hard cut to the windshield two-shot. <Man> slowly reaches toward <Fries Bag>. <Woman> snatches the whole bag and hugs it to her chest, turning away from him, her mouth moving as she talks. End state: she holds the bag against her chest, his hand hangs empty in the air. 20-24 seconds: Shot 7, hard cut to the close-up of <Man>. He lowers his empty hand back onto the steering wheel and stares straight ahead, completely deadpan. Mouth closed. End state: both hands on the wheel, staring forward. [Maintain Consistency] Keep both characters' faces, outfits, seats and seatbelts, the teal car, <Fries Bag>, the parking lot background and the bright daytime lighting identical in every shot. The camera never moves inside a shot; only hard cuts change the view. (No music)
1
20
1
185
16,165
holy shit... 31k followers from 3 VIDEOS
1
3
12
843
i grew my agency from $0/mo to $15k/mo in 90 days here’s what i learned outreaching manually > i know this sounds cliche but confidence is key, people only know the version of yourself you choose to display > actual successful people couldn’t care less if you speak formally, what they look for is your skills, hunger and passion for the project > try go get a consistent line of contact, (Whatsapp, Telegram mainly) > offer samples. paid or not the only way to stand out and not be another “joe” in peoples dms is to offer samples and do them as quick as you can > off days are okay, making 0$ for 30 days straight is okay but giving up isn’t > NEVER forget why you started this and stay consistent Luck finds those who seek it, keep grinding kings!
1
2
373
i got paid $500 to make this video you can make your own personal brand and get paid for sponsors too, here's exactly how i made mine > find a cartoon nostalgia character > ask claude to make an unhinged claymation scripts options and choose the one you like > ask it to generate start frames for each scene using GPT2 img model > finally tell it to create video prompts for each scene using kling 3.0 > copy and paste it then done this whole process will take less than 15mins, and once you made a video for one brand you'll keep making for others it's literally an infinite money glitch here's the prompt i used for this exact video: Create a short Tubbyroadmen comedy video featuring a (Censored)collaboration. Treat the absurd characters and situation like a serious British crime drama, with natural UK roadman dialogue and understated reactions. STYLE: Photorealistic characters, detailed fur, realistic faces, glossy black puffer jackets, worn British interiors, warm amber practical lighting, cool blue shadows and subtle 35mm film grain. Controlled handheld camera work. Vertical 9:16, except the generated video displayed on the laptop, which is 16:9. CHARACTERS: Dipsy: green creature with a straight antenna, cow-print bucket hat and black puffer. Black British male voice, mid-20s; flat, slightly nasal East London accent, natural roadman delivery. Po: short red creature with a ring antenna, backward black snapback and black puffer. Young adult Black British female voice; South London accent, dry and unimpressed. STORY: Dipsy sits at his laptop in a dark bedroom. A lotion bottle and tissues beside it create a suspicious visual misunderstanding. Po is already visible in the doorway. Leave the first second silent for an added sound effect. Po: “I told you to stop clicking them dodgy ads.” Dipsy: “Nah, bruv. I’m using Migoo to make our next video. Look.” Po approaches. Cut to a laptop POV showing the video he has created, composited onto the screen in editing. LAPTOP VIDEO: A shorter version of Po stands at a chicken-shop counter, holding a small kitchen knife low beside her thigh. She reaches up with her empty hand and rises onto her toes. Po: “Bossman. Empty the fucking till.” She notices the fried chicken, becoming distracted. Po: “Actually… you got any wings left?” Play the change of priorities completely straight. No attack or exaggerated slapstick. BACK IN THE BEDROOM: Po: “Make me taller.” Dipsy: “It makes videos, bruv. Not miracles.” Po: “Make yourself funny, then.” ENDING: Po: “How much you spending on this?” Dipsy, looking directly into the camera: “Half what you’d pay for Seedance. Get on Migoo, bruv—link’s in the bio.” PRODUCTION: Generate separate clips from approved start frames, maximum 15 seconds each. Every required character and prop must already exist in the frame. Maintain consistent faces, clothing, voices, lighting and physical continuity. Only the speaker moves their mouth. Natural lip sync, stable eyes and normal blinking. No extra dialogue, random vocalisations, morphing or appearing objects. Keep each individual Kling prompt under 2,500 characters.
1
3
789
This is F*CKING INSANE just found this 2 weeks old channel made with opus 5.5 code this is literally the 3rd one i found today he runs this on automation for basically $0 please dont buy a course for this, im working on a free guide ill post tomorrow
2
5
310