Frequently asked questions

You type a topic, pick a style preset, and Reviral writes a hook-driven script, generates a voiceover, picks keyframes via Nano Banana Pro, animates each scene with Seedance i2v, adds animated captions and background music, and exports a ready-to-post vertical video. End-to-end in ~3-7 minutes.

9:16 vertical by default (TikTok, Instagram Reels, YouTube Shorts). You can switch to 16:9 (YouTube long-form / landscape) or 1:1 (Instagram feed) per project.

Yes. When your script has a recurring character, Reviral generates a clean reference image once and locks every subsequent scene to it. The character description is also re-injected verbatim into every scene's prompt — belt-and-suspenders consistency that prevents face/outfit drift on multi-scene renders.

Yes. The create page has an in-browser voice recorder. The recording becomes the master audio for the render, and captions are auto-aligned to your speech via word-level ElevenLabs timestamps.

Character-accurate, not transcription-guessed. The TTS engine (ElevenLabs) returns character-level start/end timings; we collapse those to word timings, so captions snap to syllables instead of drifting.

Costs are shown in credits before you render. Base tier (30s, 6 scenes, AI-generated video) is ~63 credits. Pro tier ~90 credits. Ultra tier ~270 credits. Credits convert at $0.0175/credit on the Growth plan ($35 → 2,000 credits/month).

Stock-footage media type (skips AI video gen) at Base tier — typically under 30 credits per 30-second render. Use this for first-render exploration, switch to AI video for keepers.

Yes. New signups can claim up to 120 free credits across 5 onboarding missions (welcome bonus, first render, voice clone, social connect, 5 videos). Enough for several Base-tier test renders.

Ultra slider plans scale to 12k / 30k / 50k / 100k / 200k credits per month — the per-credit price drops materially on the higher tiers ($0.013 at Ultra 100k vs $0.0175 at Growth).

Keyframes: Google Nano Banana Pro (Gemini 3 Pro Image) at every tier — same model Revid uses for their best output. Video: Seedance 1.0 Pro Fast at Base/Pro, Seedance 2.0 Fast at Ultra (better motion + native 1080p). Both routed through Atlas Cloud when available for cost efficiency.

Pro and Ultra use multi-sample best-pick keyframe generation — 2 (Pro) or 3 (Ultra) parallel image gens per scene, with the best result selected via vision check. Base renders use single-sample for cost efficiency.

No — the script LLM picks one of seven archetypal framings (Underdog / Mentor / Outsider / Insider / Warrior / Wanderer / Trickster) based on the topic. Hook can be a specific promise, a belief contradiction, or a curiosity-gap question (Revid's 756k-video data showed questions outperform statements).

It pauses, not fails. If a provider hits quota or rate-limit, the render goes into a paused state, the admin is alerted via Telegram + email, and the render automatically resumes within 5 minutes once balance is restored. You don't see a 'render failed' screen.

No. The 24-hour pause window ends in a full credit refund with an apology email. You're never charged for a render that didn't deliver.

Yes. Renders are stored in your private project library. You control sharing — there's no public gallery unless you explicitly opt into the community templates feature.

Provider stack: Anthropic (script), ElevenLabs (TTS), Atlas Cloud / fal.ai (image + video gen), Remotion Lambda (compose), Cloudflare R2 (storage). All hosted in US regions.