# Would You Rather Video Generator > Type a topic. AutoFeed writes the dilemmas, generates an image for each side, narrates the question, runs the timer and reveals the split — red on top, blue on the bottom, OR in the middle. - Canonical page: https://www.autofeed.ai/generators/would-you-rather-video-generator - Publisher: AutoFeed.ai (https://www.autofeed.ai) - Site overview for language models: https://www.autofeed.ai/llms.txt **Best for:** Comment-driven channels that want an argument in every video. ## What a Would You Rather video is A Would You Rather video is a decision engine. The screen splits — red on top, blue on the bottom, a black bar and an OR circle across the middle — and an AI voice asks "A or B?". A clock ticks while the viewer decides. Then a ding, and the percentages appear: the winning side jumps up in lime green, the losing side stays smaller in red. The viewer has already chosen in their head, which is why the comments fill up. AutoFeed generates the pairs from one topic line, renders an image for each option and stacks up to ten dilemmas in a single video. Everything the model writes stays editable — both option labels and the percentage split — so you can hand-write the entire quiz and use the generator purely as a renderer. ## Key specifications | Spec | Value | | --- | --- | | Dilemmas per video | 1-10, eight by default | | Credits per video | 1,000 at any count | | On-screen font | Titan One, heavy black outline | | Sound design | Ticking timer, then a reveal ding | | Fixed time per dilemma | 1.5s wait + 1.5s reveal | ## Example videos - A Would You Rather video generated by AutoFeed — split screen, narrated question, ticking timer, percentage reveal. (https://www.autofeed.ai/landing/videos/example-3.mp4) ## How the Would You Rather generator works 1. **Set the topic and the count** — Type a comparison topic and choose how many dilemmas to stack: anywhere from 1 to 10, with eight as the shipped default. Each dilemma is its own scene, and a scene is three fixed seconds — 1.5 while the viewer decides, 1.5 on the percentage reveal — on top of however long the voice takes to read the two labels. So the count is what sets the length: eight dilemmas is at least 24 seconds of fixed time before a single word of narration is counted. The credit cost does not move with the count: a Would You Rather is a flat 1,000 credits whether it holds one dilemma or ten. 2. **Generate the pairs** — The model returns a title, a description with hashtags, and exactly the number of versus pairs you asked for. Each label is held to 18 characters and three words at most — "Supra MK1" against "Ferrari 308", never a paragraph of setup — because the labels are burned in at a large fixed size and anything longer shrinks to unreadable or overflows the frame. 3. **Edit anything you want** — Every comparison is an editable card: rewrite Option 1, rewrite Option 2, or set the percentage yourself. Option 2 always takes whatever is left of 100. You can add or remove comparisons up to the limit of ten. 4. **Images and voiceover** — Both option labels go to the image model as prompts, and the question "A or B?" goes to ElevenLabs. The voiceover comes back with per-character timing, so the render knows the exact moment the first label finishes being read — the second image slides in half a second later. 5. **Optional background removal** — Turn on Remove BG and each image is passed through a background remover, so the object sits directly on the red or blue panel instead of inside its own photo. It also repositions both images so the split reads cleanly. Use it for objects; leave it off for scenes. 6. **Render, then let it run** — Choose 9:16, 16:9 or 1:1, pick a music bed, and render. Or attach the format to a series and let AutoFeed publish a new quiz to TikTok and YouTube Shorts on your schedule. ## Everything you can configure ### Comparison topic The theme the pairs are written from. - Any theme — "Marvel villains", "90s cars", "airport food" - Defaults to "Supercars: Lamborghini vs Ferrari vs Bugatti vs McLaren" ### Number of comparisons How many dilemmas are stacked into one video. - 1 to 10 per video, eight by default - Adding or removing a card updates the count automatically - The credit cost is flat at every count ### Option labels and percentages Every pair is editable before it renders. - Both labels editable per comparison - Percentage 1 editable from 0 to 100; percentage 2 is always the remainder - The winning side renders larger and in lime green, the losing side in red ### Remove BG Cuts the generated objects out of their backgrounds. - Off by default - Objects sit directly on the colour panel - Also shifts both images toward the centre line ### Voice Reads "A or B?" over every scene. - Full ElevenLabs voice library - Filter by gender and by use case - No separate language control on this format ### Background music Mixed underneath the narration. - Defaults to Solitude, mixed at 20% volume so the question stays audible - 87 tracks in the shared library, plus None ### Aspect ratio The shape of the finished file. - 9:16 vertical - 16:9 horizontal - 1:1 square ## 30 Would You Rather Dilemmas, Plus 12 Series Topics A dilemma only works if both sides are defensible. These are written as balanced pairs — the left half goes on the red panel, the right half on the blue panel. Stack eight of them, which is the shipped default, and the viewer argues with themselves the whole way through. *Where to paste:* On the Content tab, press Edit on the post and paste the left half into Option 1 and the right half into Option 2. Trim each side toward 18 characters and three words, which is what the generator itself is held to — longer labels are auto-fitted, so they render smaller. Set the percentage yourself, or leave the generated split. ### Superpowers and sci-fi - Fly at 5 mph — Run at 60 mph - Rewind ten seconds once a day — Pause the world for one minute a week - Talk to animals — Speak every human language - Teleport, but only to places you've already been — Fly, but only at night - Read minds for one hour a day — Know when anyone is lying, forever ### Money and work - $50,000 today — $5,000 a month for a year - Never work a Monday again — Never work an evening again - A four-day week at 80% pay — A five-day week fully remote - Free rent for five years — Free food for life - Win the lottery and tell nobody — Win half of it and be on the news ### Food - Only sweet food for a year — Only savoury for a year - Perfect pizza forever — Perfect sushi forever - Never drink coffee again — Never eat cheese again - Every meal cold but delicious — Every meal hot but bland - Street food anywhere in the world, any night — One Michelin dinner a month ### Travel and lifestyle - A window seat next to a screaming toddler — An aisle seat in the last row - Live somewhere it's always 30°C — Live somewhere it's always 5°C - Fly anywhere free, but never in first class — One first-class flight a year - A house in the mountains — A flat above a great café - Never lose your luggage again — Never queue at security again ### Tech and the internet - Lose every photo you've ever taken — Lose every message you've ever sent - A phone battery that never dies — Signal everywhere on earth - Your search history goes public — Your camera roll goes public - No notifications for a year — No passwords for a year - Give up streaming — Give up online shopping ### Chaos and classics - Fight one horse-sized duck — 100 duck-sized horses - Always be 20 minutes early — Always be 5 minutes late - Sneeze every time you laugh — Hiccup every time you lie - Everyone hears your thoughts for a day — You hear everyone's for a week - Live your best day on repeat — Never repeat a single day ### Topics for a whole series Type one of these into the Comparison Topic box instead and AutoFeed writes the pairs for you — useful when you want a themed channel rather than one-off quizzes. - 90s cars against their modern replacements - Marvel villains - Premier League strikers, any era - Fast food chains - Anime protagonists - Video game weapons - Dream jobs, each with one catch - Holiday destinations in the middle of winter - Dog breeds - Historical figures to have dinner with - Streaming services - Gym exercises you'd delete from existence ## Frequently asked questions ### Where do the percentages come from? They are generated, not measured. AutoFeed rolls a winning share between 52 and 66 per cent, flips a coin for which option gets it, and gives the other side the remainder — deliberately close, because a landslide is a split nobody argues with. Both numbers are editable before you render. Treat them as the format’s punchline rather than as survey data, and if you want them to mean something, set them yourself from the comments on your last video. ### How many dilemmas should one video have? Eight, which is the shipped default. Length follows the count directly: every dilemma is the narrated question plus a 1.5-second wait and a 1.5-second reveal, so each one adds three fixed seconds on top of its own narration. One comparison is a few seconds of video, which is far too short for the format to land; eight is at least 24 seconds before the voiceover is counted at all. The limit is ten, and since the video is a flat 1,000 credits at any count, a longer quiz costs nothing extra. ### Can I write the dilemmas myself? Yes. Every comparison is an editable card with Option 1, Option 2 and a percentage. Generate once to get the structure, then overwrite it — or delete the generated pairs and type your own set from scratch. ### What does Remove BG actually do? Both generated images are passed through a background remover, so the object sits on the red or blue panel with no photo edges around it, and the render moves the two images closer to the centre line. It is worth turning on for objects and products, and worth leaving off for landscapes and scenes. ### Is there a voiceover, or is it silent? There is one. The question "A or B?" is narrated for every pair using the voice you pick, and the timing of that narration is what drives when the second image slides into frame. ### What is the ticking sound in the middle? That is the wait phase. After the question is read, a ticking clock runs while the viewer decides, then a ding plays and the percentages replace the labels — the winner larger and in green, the loser in red. ### Can a Would You Rather series run on autopilot? Yes. Attach the format to a series, pick the days and the whole-hour slots you want, and AutoFeed generates a fresh video and publishes it to your connected TikTok and YouTube Shorts accounts in your own timezone. Plans cap a series at 3, 7 or 14 posts per week. ## Related - [AI Video Maker](https://www.autofeed.ai/features/ai-video-maker) - [All generators](https://www.autofeed.ai/generators) - [Pricing](https://www.autofeed.ai/pricing)