Built for the full faceless YouTube pipeline.
From prompt to render, every step is covered. No stitching tools together, no manual copying between apps.
AI script generation
Type a topic, review and edit a researched outline, then watch the full script stream in section by section. Optional multi-pass deep mode generates 20+ section scripts that retain pacing and hook strength end-to-end.
- Hook + body + outro structure
- Configurable target duration (60s - 30 min)
- Tone presets: documentary, listicle, story, educational
- Fact-check panel flags claims and suggests rewrites
- Hook score out of 30 (stakes, specificity, title promise) before voiceover starts
- Or paste your own script - # titles, ## headlines, > quotes, {{source}}, {{speaker}}, {{bullets}} markup
- Or upload your own voiceover (up to 20 files, transcribed with Whisper)
- Local Ollama by default, or any OpenAI-compatible endpoint (OpenAI, OpenRouter, Together, LM Studio, vLLM)
Stock footage that fits the narration
Pexels, Pixabay, Coverr, the Internet Archive, Wikimedia Commons, NASA, DVIDS and your own downloaded clips searched per scene. CLIP scores every candidate against the narration so the b-roll actually matches, and finds the right minute inside long archive films.
- 12 candidates fetched per scene, CLIP-scored
- B-roll cycling - 2–5 clips per scene instead of one
- Per-clip motion-only override (replace footage with motion graphic per slot)
- Zoom + pan keyframes per candidate
- Sub-clip selection from source video
- Brand kit compliance rules block unwanted sources / URLs
- Optional Higgsfield cinematic camera-move shots (hosted, about $0.75 per clip)
AI voiceover: ElevenLabs or ai33.pro
Per-scene MP3 caching, parallelized 4-way for speed. Voice set per channel, brand kit, or video. Use ElevenLabs, ai33.pro, or your own recording.
- ElevenLabs Multilingual v2 + Turbo support
- ai33.pro (OpenSpeaker): the ElevenLabs voice library plus MiniMax, Fish Audio and Vbee voices, for less per character
- Voices you picked on ElevenLabs keep working through ai33.pro
- Parallel synthesis (4× faster than serial)
- Cache key includes voice + speed + model - change one and only that re-renders
- EBU R128 loudness normalization to broadcast standard
- Default voice + speed per brand kit, overridable per video
- Hook preview clips - listen to your opener variants before committing
100+ motion graphic types
Drag and drop from the Animations tab, or let Auto-fill motion suggest them from the narration. Editable per scene. Renders via Remotion (headless Chromium) with up to 3 concurrent jobs.
- Title cards, headlines, quotes - markup-driven
- Stat callouts, bullet lists, progress bars - data overlays
- Counters, timelines, maps - animated infographics
- Charts (bar / donut / line / heat) - auto-styled per brand
- Comparison cards, decision flowcharts, schematics - explainer overlays
- Per-clip placement - motion ribbon overlays footage in the merged V1 timeline
Captions auto-generated, burn-in optional
Whisper transcription per scene → ASS subtitle format → optional ffmpeg burn-in. Caption styles per brand.
- Per-channel caption style (font, color, outline, position)
- GPU-accelerated transcription on NVIDIA, CPU on AMD
- Karaoke-mode word-by-word highlight
- Beat-snapped caption transitions when music sync is on
- Export SRT for YouTube upload alongside the burned-in version
Background music + beat sync
Music library with beat-detection metadata. Scene boundaries can snap to beats automatically.
- Built-in music library (free + Artlist integration optional)
- Beat snap with configurable tolerance
- Auto-duck under the voiceover
- -14 LUFS loudness normalization
- Per-project track + tempo selection
SFX library + per-scene placement
Drop sound effects on the timeline. Search-as-you-type from the built-in library.
- Mixer with per-event gain control
- Time-anchored placement, drag to move
- Bundled library covers transitions, impacts, atmospheres
- Auto-place SFX on cuts
Multi-track timeline editor
Video + motion + captions + voice + music + SFX in one timeline. Drag to reorder, trim with snap-to-scene, ripple delete.
- Real footage thumbnails on V1 clips
- Waveform silhouettes on voice clips
- Motion-graphic ribbon merged INTO the V1 video lane
- Per-boundary transition picker (+ button between scenes)
- Snap motion graphics to scene start/end
- Split, trim, crossfades, enter/exit animations
- Final vs Live preview
- AI editor (Ctrl+K) - natural-language edits with undo snapshots
GPU-accelerated rendering
Final MP4 renders via NVENC on NVIDIA, AMF on AMD, or libx264 on CPU. Auto-detected at startup.
- NVENC: 5–10× faster than CPU, near-zero CPU usage
- GPU CLIP scoring (CUDA on NVIDIA, DirectML on AMD)
- CLIP batch size scales to 32 on GPU vs 8 on CPU
- Per-stage timestamps for build-time profiling
- Parallel TTS (4-wide), parallel Remotion (3-wide)
Brand kits
Per-channel + per-video override system. Logo bug, LUT color grade, intro/outro stingers, default voice and speed, motion-graphic theme and blocklist, sourcing compliance rules.
- Brand kit per channel with default voice / speed / theme
- Logo bug, LUT color grade, intro + outro stingers
- Per-video overrides on the Video settings screen
- Block sensitive sources (gore, copyrighted brands, etc.)
- URL blacklist per brand
- Compliance toggles (commercial-stock, CC public-domain, general web)
AI thumbnails + hook A/B testing
Pikzels integration for thumbnail variants. Hook scorer rates opening-line variants before voiceover.
- 1-8 AI-painted thumbnail variants per run (Pikzels)
- Prompt suggested from your script, edit any variant with a prompt
- LLM-generated hook variants for scene 1
- ElevenLabs preview per hook - listen before committing
- Pick the winner, apply to the project automatically
Long-form and Shorts
Pick 16:9 Landscape for YouTube long-form or 9:16 Shorts for YouTube Shorts, TikTok and Reels. Same pipeline, same editor.
- 16:9 Short (~1 min), Standard (~7-9 min), Long (~15-18 min), Deep dive (~25-30 min)
- 9:16 Shorts up to 3 min (motion graphics off for Shorts)
- Editing format, voice, speed and language (English / German) per video
- Pre-build cost estimate before you approve
Publish straight to YouTube
Built-in YouTube publishing through your own Google Cloud OAuth app. No export-then-upload round trip.
- Title, description, tags, category, language
- Visibility (private by default) + scheduled publish
- Playlist, custom thumbnail, made-for-kids flag
- Multiple YouTube accounts linked to channels
- Publish history per video
- Also exports MP4 (H.264), FCP 7 XML (Resolve / Premiere) and SRT
Research toolkit
Eleven research tools plus channel analytics, so the topic, title and metadata are as strong as the video.
- Topic pulse (Google Trends), Keywords, Competitor teardown
- The Lab script audit, Channel brief, Entity optimizer
- 12 scored titles, thumbnail research wall
- Tags (500-char budget), descriptions with auto chapters, Humanize
- Self-audit, Hook A/B, Pacing dashboard, Script length calculator
- Channels with YouTube history sync and charts
Your data stays on your machine
Scripts, footage, finished videos - never uploaded. Disconnect the internet and renders still work (with cached assets).
- No cloud sync, no telemetry of content
- SQLite-based local database
- All paths configurable (assets / renders directories)
- Per-project cleanup frees disk after completion
- Crash reports anonymized (opt-out available)
Get VidNexus.
Lifetime license. Pay once, render forever.