Script, voice, images, AI video and editing run on your NVIDIA card: $0 per video, and your media never leaves your PC. No big GPU? Rent the power from RunPod, at true cost.
From idea to published short without leaving the app — and in control at every step.
Type an idea: the Studio writes the script, records the voice, finds or generates the shots and edits the video in 9:16, 16:9 or 1:1. You approve every step, nothing ships without you. Your own title or a simple hook, plus a built-in profanity filter.
Wan 2.2 animates your scenes at native 704×1280. Short, sharp shots, never mushy.
A custom image per scene, brought to life in the edit with a camera move.
✕
✕
✓AI looks at every shot before keeping it. Nothing fits? It searches again.
See and tweak every scene before you render, and drop stickers on them (pack included). No bad surprises.
Free local voices or HD ones. A dictionary fixes pronunciation without touching the captions.
Word by word, synced to the voice, with your link stamped at the end.
Queue your ideas at night: the night queue renders them one by one, ready when you wake up.
Drop your videos: the AI watches each clip, places it on the right scene and explains why. Nothing is generated, everything stays on your PC. New in 1.8.
One photo and a text: your avatar speaks with synced lips, computed on your card. Included since 1.7.
Take someone's place in a video, from a single photo. Included since 1.7.
Your AI shots on cloud GPUs, paid per shot with your RunPod key. No subscription, no markup.
On launch, the app profiles your card (model, VRAM, driver) and applies the best setting it can actually handle, without crashing it.
Your machine can't handle visual AI? No problem: plug in your RunPod key, top up whatever you want, and the app generates your shots on high-end cloud GPUs. No subscription, no markup. Script, voice and captions stay local and free (3.8 GB lightweight engine), and so does Stock mode.
Z-Image-Turbo in Q4, 576×1024, brought to life in the edit with camera moves. $0 per image.
Wan 2.2 in Q4, 17 frames max per shot. To stay sharp, the app officially recommends AI images, and AI video via RunPod.
Wan 2.2 5B in Q8, native 704×1280, 25 sharp frames per shot extended with smooth slow-motion. ~225 s per shot, $0.
AI images in Q8 at 720×1280, 25 frames per shot, 8B language model in Q8.
49-frame video shots, no compromise on sharpness.
We'd rather give you a surgically sharp 1–2 second shot than a mushy 5-second clip with smoothed-out faces and melting hands. The local recipe is tuned for realism, not length.
FFmpeg editing, transitions and filters, captions, audio mix normalized to EBU R128 at -14 LUFS. Even without a graphics card, your video is assembled at home.
RunPod, MiniMax, ElevenLabs, Kimi K3 or Claude: with your keys, direct requests to the provider, no LastParadox server in between, no markup.
Laptop, office PC, old graphics card: you still get photorealistic AI video. The difference with the platforms: you pay RunPod directly, at true cost, only when you generate.
Open-weights models that run at home, and the cloud only if you choose it, billed at the provider's price.
Retro interface, GPU profile, local packs and live progress. Screenshots of version 1.8 on an RTX 3060 profile.
AI video platforms profit from credits that melt with every second of compute. We sell you the tool once.
A Windows PC. For local AI, an NVIDIA card of the Pascal architecture or newer, with an up-to-date driver. The app profiles your card and unlocks what it can actually handle:
Models download on first launch, in the version suited to your card:
Plan for ≈ 24 GB of disk space for the full setup on an RTX 3060, ≈ 32 GB on a 16 GB+ card. Without a suitable card, any recent Windows PC is enough for Stock mode and Cloud GPU Studio mode; editing always happens locally.
We won't sell you a dream: generating AI video on a CPU or an old graphics card would take hours or crash, so the app doesn't try. But you're not locked out of AI video:
Either way, the app installs a lightweight local AI engine (≈ 3.8 GB): script, visual checking, Kokoro voice and captions run on your PC for free, as do editing and the audio mix.
On an RTX card, everything runs on your PC, with no API key and nothing sent over the Internet:
Cost: $0 per video. The trade-off we own: a short made entirely of AI video takes roughly 15 to 20 minutes on an RTX 3060, because we favor sharpness over speed.
Only what you turn on, with your own keys (BYOK): Kimi K3 or Claude Sonnet 5 for the script, RunPod for cloud GPU video (Tier 4 FLUX + Wan 2.6, MiniMax Hailuo 02, Pruna), MiniMax Speech 02 HD or ElevenLabs for voices, Pexels/Pixabay for stock footage, Microsoft for free online voices. Keys are stored in a local file and requests go straight to each provider, never through a LastParadox server. You pay the provider's true price, no markup: ~$0.02 of script per short, ~$0.50 per RunPod shot at maximum quality. The cost is shown before every call.
Golden rule: zero compromise on pro quality. Past 1–2 seconds on a consumer card, video models produce smoothed-out faces and melting hands. So the app generates short shots at native 704×1280, then extends them with smooth slow-motion and a slow cinematic zoom. The result looks real, not like a cartoon.
Both modes are available since 1.7 (NVIDIA card with 12 GB or more, packs installed from Settings). The update is included in the license.
Both require your consent (your own face, or the person's written permission) and stamp an “AI-generated” label on the video.
Yes. Stories are 100% original, footage comes from your local models, from royalty-free banks cleared for commercial use (Pexels/Pixabay) or from generative models run with your keys (RunPod), and voices are cleared for commercial use. No watermark, no restrictions: monetize freely on YouTube Shorts, TikTok, and Instagram Reels.
Yes — English scripts (local model, Kimi K3 or Claude), English voices (local Kokoro, free Microsoft, ElevenLabs or MiniMax HD), synced dynamic captions. French and other languages are also supported natively.
You can add your personal images and clips right from the first step: the AI reserves the right scene for them in the edit. You also get a built-in Viral Stamp CTA generator to automatically stamp an elegant neon badge with your website URL or social handle.
€39 once. Your graphics card does the rest — or your own keys, down to the cent.
Buy — €39