Every image and video here was generated on blubiu. Pick a scene to see what its workflow makes.
blubiu is an AI animation generator for one-person studios: script, characters, shots, voice and music in one workflow. Pick a scene, fill in the blanks, get the video, with the same characters every time. Every blubiu workflow exists to pull the four below into one workbench.
Script in a chat app, images in one generator, video in another, voice in a third, then an editor to put it together. Every switch loses context.
One workflow runs script, shots, voice and cut in one place: pick a scene, fill in the blanks, get the video.
The same character comes back with a different face or outfit every generation, so a series never holds together.
Characters, places and voices are set once as assets, and every shot is generated against the same ones.
Dozens of images and clips per episode, and you are the only one keeping track of which ones made the cut.
Every shot is confirmed on one workbench: the keepers move on, the rest stay put.
What comes out is a folder of short clips; stitching, voice, music and captions still happen somewhere else.
Stitching, voice, music and captions happen on the same workbench, and the export is the finished video.
Script, characters, shots, voice and cut are all inside. Set characters, places and voices once, and every video after that matches.
Content with a recurring cast, made episode after episode.
Upload a song, pick a character, and get a full music video cut to the beat.
A narrated story acted out shot by shot by your characters: bedtime, folklore, horror.
A one-to-three-minute short with script, shots, voice and music in one run.
The same cast, places and voices, episode after episode.
Vertical 15-to-60-second clips for YouTube Shorts and TikTok.
Bedtime and learning stories fronted by one recurring character.
2D cartoon episodes with flat colors and clean lines.
Stick-figure animation: the lightest style and the fastest output.
Ten-minute-plus videos, generated in parts and assembled on the same workbench.
A full sticker pack of the same character, as transparent stickers or GIFs, sized for Discord and Twitch.
Turn your own cat or dog into a character, then put it on screen.
A few photos set the look; your pet becomes a cartoon character for stickers and short videos.
Give it a voice and let it talk to camera.
One persona that keeps posting, with no one on camera.
Check-ins, gym days and daily moments, posted with the same face every time.
You write the script; the character presents and narrates it.
Keep a trending video's timing and camera moves, and swap in your character.
One full-body photo moves like a dance clip.
Seedance 2.0Cinematic clips up to 15 s, with optional sound
MiniMax H3Hailuo video up to 2K, follows a reference video
Veo 3.1 FastGoogle's Veo: 8-second clips with sound
GPT ImageBlends references into polished images, up to 4KGPT Image 2.6OpenAI's newest image model: fast Flare or careful SunburstSeedream 5.0ByteDance image model in Pro and Flash tiers
Wan 3.0Up to 15 s with sound and the most references
Kling Motion ControlA photo performs the moves from a videoFFlashVSRUpscale any video to 720P–4KYour character isn't a throwaway prompt. Its sheet, turnaround, places and props are saved as assets, and every later image and video draws on the same ones.
Documentary
Sweet drama
Product
Sticker
Keyframe
Rainy dayLook, outfit, personality and rules, written down as data that every shot is checked against.
Front, three-quarter, side and back views, so the character holds up from any angle.
The places your character keeps coming back to, drawn the same way in every episode.
Signature items and your own products, kept consistent in every shot.
One account and one set of character assets; start whichever way suits you.
Open a scene page, use a prompt template or write your own, and get one image or video.
For: People who know exactly what they want and need one result.
Pick a scene such as a music video or a kids' story; set characters, places and voices once, and the script, shots, voiceover and cut come out of one workflow.
For: People who want to fill in a few blanks rather than break the work into steps themselves.
Drag script, image, video and voiceover nodes onto a canvas and wire the whole pipeline yourself.
For: People who want fine control over every step. Workflows run on this same canvas, so any workflow can be opened and changed.
Yes. Pick a scene and say what you want: the script, the shot list and every frame description are written for you, and each step stops to show you before anything is generated. Don't like something? Say it in plain words: “make shot 2 night-time”.
You pay per generation and see the cost before you run it. Prompts come first for your approval, keyframes next — video, the expensive step, only runs after you've said yes. Failed generations are refunded automatically.
Image: GPT Image, Nano Banana and Seedream. Video: Seedance 2.0, Veo 3.1 Fast, MiniMax H3 (Hailuo) and Wan 3.0, plus Kling Motion Control for motion transfer and FlashVSR for upscaling. You pick the model for each generation, and all of them run on the same credits. No lock-in, no bundled tiers.
Images and video files download directly. Workflows can also be published as apps or called over the API and wired into your own pipeline.
Your character isn't redrawn each time — it's stored as an asset: a 4-view turnaround plus a structured character sheet in your IP library. Every shot and every episode anchors to it, so episode 12 is the same face as episode 1. This is the main thing that separates us from generic video tools.
Yes. What you generate is yours, with no extra licensing fee for commercial projects.
No. The system knows which canvas node each shot maps to and which downstream steps are affected, so only those re-run. That's exactly why we built natural-language rework as a core feature.
Most tools solve “generate one clip.” We solve “the same IP keeps producing new work without drifting” — character assets, the cross-shot consistency chain and per-shot rework are all built around that one problem.
