Generate cinema-quality videos and stunning images from text or image. Powered by Kling 2.1 and GPT Image 2. Free credits to start.
π Click Generate to sign in β 5 free credits on sign-up, no credit card required
Polyfaced is built around a generation-aware credit ledger, real Cloudflare R2 storage with global CDN delivery, and a unified prompt box across every model on the roadmap.
Best-in-class models live at launch β Kling 2.1 for video, GPT Image 2 for image. Three more surfaces (Avatar, Music, Movie Maker) on the way.
Free credits granted at sign-up β enough for a Kling 2.1 short clip and a couple of GPT Image 2 stills before deciding whether to upgrade.
Pro plan price per month. 800 credits refilled on renewal, no watermark, full commercial license, no annual lock, cancel anytime.
Maximum Kling 2.1 video resolution available on the Pro plan. Free plan caps at 720p. 1080p is the default for production work going to social ads or client decks.
Maximum GPT Image 2 resolution available on the Pro plan. Free plan caps at 1024px. 1536px is the right pick for print, hero banners, and large screen displays.
Failed generations charged. The FIFO ledger releases credits automatically when an upstream provider rejects a job or hits a timeout β you only ever pay for successful output.
Each studio surface is tuned for one job. v1 ships text-to-video and text-to-image with the best provider available; v2 adds dedicated model pages for Veo 3.1, Sora 2, Seedance, Wan 2.5 and Pika 2.2.
Generate cinema-quality video from a single prompt or starting frame with Kling 2.1. Up to 10 seconds at 1080p with native motion physics, prompt-aware shot composition, and image-to-video unlocked on Pro. Cinematic ratios 16:9, 9:16 and 1:1 supported out of the box.
Generate high-detail images with GPT Image 2. Text-to-image renders prompts faithfully at 1024px or 1536px; image-to-image transforms an existing reference while preserving subject identity. Native style transfer means brand-aligned outputs without long prompt engineering chains.
Coming soon β generate consistent AI headshots and full-body avatars across styles, lighting setups, and outfits. Maintains face identity across a batch so you get a usable variation set, not 10 random faces. Join the waitlist for early access.
Coming soon β text-to-song composition with vocal direction, genre control, and exportable stems for re-mixing. Built for creators making short-form video ads, podcasts, and game soundtracks. Join the waitlist for the model selection process.
Coming soon β multi-shot AI movie maker with scene composition, character continuity across shots, and one-click voiceover. Ideal for storyboarding, pitch reels, and full short films generated end-to-end from a script outline.
No model API to wire up. No multi-tab provider hopping. Sign in, describe, generate, download β every step is queue-aware and credit-aware so you always know what's happening and what it costs.
One click with Google SSO. 5 free credits land in your balance instantly β enough to test both models before opting into any plan.
Open the studio for video or image, pick your ratio, resolution and duration, then describe your scene in plain English. The cost in credits is shown before you press Generate.
Polyfaced freezes the credit cost upfront, dispatches the job to the provider, and streams progress updates. Most Kling 2.1 videos finish in 60-90 seconds; most GPT Image 2 stills land in under 15 seconds.
Every output is stored on Cloudflare R2 with a permanent share URL. Download in original quality, share the URL, or open from your dashboard history page anytime.
Polyfaced is not a thin wrapper around one provider β it's a creator-first platform that brings the strongest generative video and image models into a single, predictable workflow. The video industry currently has Kling 2.1, Veo 3.1, Sora 2, Seedance, Wan 2.5 and Pika 2.2 all leapfrogging each other every quarter; the image industry has GPT Image 2, Imagen 4, Midjourney v7, Flux 1.1 doing the same. Creators who try to keep up with all of them quickly drown in browser tabs, separate billing pages, separate prompt schemas, and separate file repositories. Polyfaced collapses that complexity into one studio: one prompt box, one credit balance, one Cloudflare R2 history, one commercial license. We pick the best model for each surface at the moment, ship it, and replace it on the day a better one appears. You get the leading edge without managing five vendor relationships.
Every model we ship runs through the official provider channel β Kuaishou for Kling 2.1, OpenAI for GPT Image 2 β via Kie's infrastructure layer. No reverse-engineered APIs, no gray-area scraping, no risk of overnight model deprecation. When OpenAI ships GPT Image 2.5, we route to it within days, not weeks.
The FIFO credit ledger freezes credits on submit, settles them on success, releases them on failure. You never pay for an output you didn't receive. Every freeze, settle, and release is logged as an immutable row β you can audit any charge end-to-end from the dashboard. Compare with the typical SaaS subscription that just charges you regardless of whether the API succeeded.
Every generation is stored on Cloudflare R2 with global CDN delivery and a permanent share URL. Download original quality anytime, link from a tweet, embed in a presentation, drop into Figma. Your generations are yours β if you cancel and don't reactivate, we email you a final export link before purging, not a silent delete.
Today the studio routes you to Kling 2.1 or GPT Image 2 based on which surface you opened (video or image). v2 brings a model selector inside the same prompt box β pick your model, hit Generate, compare side-by-side without losing your prompt state. The same UX scales to Avatar, Music, and Movie Maker as those surfaces land.
Every generation is served from Cloudflare's global edge network. Whether you're in Tokyo, Lagos, Berlin, or SΓ£o Paulo, the download is fast because the file is cached close to you β not pulled fresh from a single origin server. Sharing a generation URL with a client in another timezone? They'll see it in milliseconds, not seconds.
Polyfaced does not train any model on your prompts or your outputs. We do not resell your generations. We do not use them for marketing without an explicit opt-in toggle on the dashboard. If you cancel and don't return within 12 months, we email you a final ZIP-style export link before purging β never a silent delete, never a surprise deletion email, never a 'sorry we lost your files' moment.
A curated set of real generations across both models. We backfill this gallery with new outputs as creators opt in to share. Want yours featured? Toggle the share flag from your dashboard history page after generating.
Mountain sunset dolly shot β real Kling 2.1 video lands after Stage 8 R2 hosting.
Studio product photography style β real GPT Image 2 stills land after Stage 8 R2 hosting.
Cinematic portrait with identity preservation β sample lands after Stage 8.
Fast camera tracking with native motion physics β sample lands after Stage 8.
Neon-accent abstract composition β sample lands after Stage 8.
Rigid-body product spin with crisp motion β sample lands after Stage 8.
Everything new creators ask in their first session on Polyfaced. Have a question that's not answered here? Email hello@polyfaced.com β most replies come within one business day.
Sign in with Google in one click. 5 free credits land in your balance instantly β no card required, no trial pitfalls. Start in the video studio with Kling 2.1, or jump to the image studio with GPT Image 2.