Story Mode

An AI faceless video generator: set up a short in five steps (format, voice, music, art style, captions) and the AI writes a complete story with a hook and a strong ending, then makes it with the same characters and background in every scene. You approve it at three points.

How it works

Go through a short setup: format, language and voice, music, art style, captions, and video and budget. Every step after the first has a suggested default, so you can press Next straight through. The AI then writes the story, picks the place, makes every keyframe and builds the video. You only say yes or no at three points. Video models make a few seconds at a time, and clips made one after another drift, so the AI fixes the look first and makes video only between keyframes you have approved.

Set up in five steps
Approve script and cast
Approve keyframes
Approve final video

It is made for faceless channels on TikTok, Instagram Reels and YouTube Shorts: scary stories, Reddit-style stories, history, mythology, true crime, fun facts, top 5 lists, motivation and day-in-the-life videos. You never appear on camera or record your voice; a narrator reads the script, or captions tell the story.

How the AI writes a story people finish

Short videos live or die in the first two seconds and on whether people watch to the end. The script is written to keep them watching:

Part	What the AI does
Ending first	It plans the twist or payoff before writing anything, then writes every scene to set it up, so the story feels complete instead of trailing off.
Hook	Scene 1 opens on tension, a shocking detail or a question you need answered. No greetings and no "in this video".
Story shape	Each format follows its own arc. A story goes hook, setup, complication, rising stakes, turn, climax, payoff. A Top 5 counts down to the most surprising item. A Reddit-style story builds from the first strange sign to the twist.
Momentum	Every scene follows from the last ("but", "so", never just "and then"), adds something new, and a second hook around the middle keeps attention up.
Ending	The last scene pays off the hook with a twist, reveal or emotional resolution set up by a detail planted early. The final line is short and quotable and, when it fits, echoes the opening so the video loops into a rewatch. Never a cliffhanger, "part 2" or "follow for more".
Pictures	Each scene's picture shows who is in frame, what they are doing, their expression and the light, and neighbouring pictures keep the same people, clothes and place so they blend into each other.

Scripts are written by the same OpenAI model as this assistant. Not happy with it? Use Redo with a note, for example "make the twist darker" or "shorter sentences".

The five setup steps

Step	What you pick
Format	Presets and custom: a story pack, a theme (Horror, Playful and more) and up to 3 characters, then your own idea and character if you like. Format: Story, Top 5 list, Did you know, Reddit-style story, Motivation or Day in the life, and a length from 30 seconds to 5 minutes. Niche: Scary stories, History, Mythology, True crime, Fun facts and more. One pick is enough.
Language and voice	One of 28 languages (English, Hindi, Telugu, Tamil, Kannada, Malayalam, Bengali, Marathi, Gujarati, Punjabi, Urdu, Spanish, Portuguese, French, German, Italian, Russian, Chinese, Japanese, Korean, Arabic, Indonesian, Turkish, Vietnamese, Thai, Filipino, Dutch and Polish) and a narrator. Captions are drawn in the language's own script. Press Play to hear a real sample. Or pick My ElevenLabs account, paste your ElevenLabs API key (stored encrypted, only the last 4 characters are shown), and use any voice in your account, including your own cloned voices, with your choice of model and voice settings. Then the voiceover costs no tokens, and captions follow the exact time each word is spoken.
Background music	A preset mood with a preview, your own uploaded track, or none. Set the volume and whether it gets quieter under the voice.
Art style	Comic, Creepy, Anime, Cinematic, Watercolor, 3D animated, Dark fantasy, Pixel art or Realistic. The suggested one is marked. Then the camera movement (Varied, Zooms, Pans, Gentle, Still) and the transition between pictures (Slow crossfade, Crossfade, Fade to black or white, Dissolve, Slide, Wipe, Zoom, Circle reveal, Blur, Pixelate or Mixed), each with a live preview, and its speed. Moves ease in and out, crossfades use an eased blend, and each scene ends on the exact frame the next one starts on, so joins are seamless.
Captions	On or off, with a live preview. Pick the text colour, a highlight colour for numbers and the spoken word, and the background (none, shade, dark, black, white, yellow, navy or plum). Each has a colour picker for any other colour; a picked background is slightly see-through. Then the font, how each caption comes in (animation: none, pop, fade or slide up), the size, the position, how many words show at once, and ALL CAPS.
Video and budget	Photos, or Photos + AI video. Slide the number of pictures and, for AI video, the number of clips and their length, and pick the models, the clip resolution and whether clips keep the model's own sound. Choose 720p or 1080p, and Voiceover or No voiceover. The full price is shown line by line, and Create story shows the total. The suggested model fits your art style and characters, for example PixVerse for neon or anime looks.

Your first pick suggests the voice, music and style. Once you change one of those by hand, later picks leave it alone.

The three approvals

Approval	What you do
Script and cast	Read the story, the character and the set. Approve, Redo, or Redo with a note, for example "make her older and the setting a desert".
Keyframes	Tap any frame that looks wrong, then Redo marked. The rest are kept. Use Compare to character to check the face. Approve all then asks how to make the video: Stitch the images or AI video with a model you pick, each with its price. Nothing starts until you confirm.
Final video	Watch it in the player. Tap a scene on the strip to redo just that one, or approve. Then download or schedule a post.

The progress strip shows what the AI is doing. You can leave the page; the work picks up where it was.

Why it looks like one take

Join	Result
Seamless	Both scenes use the same keyframe, so the join matches frame for frame.
Cut	The next scene gets its own keyframe with a new angle. It reads as an edit, not a glitch.
Extend	The model continues the motion of the scene before.

Every keyframe is drawn from the character sheet and the set image, never from a previous clip, so nothing gets worse along the chain. The AI chooses the joins; you can change them in Adjust.

Adjust

Adjust opens every manual control: length, visuals, voice, music, style, character text and views, set, joins, keyframe prompts and finish options. Changing something that alters the keyframes asks once, then the AI remakes them.

Saving and tokens

Your setup is kept in this browser until you press Create story. After that the story is saved to your account and the AI works on our servers, so you can close the page and come back. Each step is charged when it starts: the script and characters, then the keyframes, then the video. If a step fails, its tokens come back and the story returns to the last approval. AI video clips are charged one by one when each starts.