On-brand AI image generation —
key art to character sheets.
Key art, product stills, and multi-angle character sheets from GPT Image, Nano Banana, FLUX.2, Grok, and Seedream — every still bound to the same Cast identity that runs your video. The frame you sign off is the face the campaign keeps.
Image generation that answers to a brand book
Standalone image tools make impressive one-offs. Campaign stills have to match each other — and match the film. Every route here generates against the same locked identity.
The still and the spot share the face
Cast identity holds across image and video engines alike — the packshot, the character sheet, and the hero spot all resolve to one face, one wardrobe, one grade.
Fourteen routes, one picker
Seedream draft-to-master tiers plus GPT Image, Nano Banana, Grok, Kling, and FLUX.2 — pick per shot and keep one workspace, one wallet, one Cast library.
Reference-driven by design
Up to 14 reference images steer a single Seedream generation — FLUX.2 takes up to 8. Cast identity resolves server-side to the member's verified assets, so the face comes from the roster.
Costs quoted before you run
Seedream stills are a flat 1 VX ($0.10 list). Third-party models meter per output and per reference image — the exact figure sits on the Generate Bar before you commit.
Stills that become shots
Any keeper can seed a video: the still becomes the first frame of an Image to Video generation, and the node canvas chains the handoff automatically.
4K where it counts
Native 4K on the Seedream tiers, GPT Image 2, and Nano Banana Pro — enough for OOH and print masters. Most other routes finish at 2K.
Fourteen image routes, one selector
Model identity comes from the engine registry and per-still VX from the live pricing engine — the numbers below are the numbers the Generate Bar quotes.
| Model | Route | Max output | VX / still |
|---|---|---|---|
| Seedream 4.0 | Iteration tier · BytePlus | 4K | 1 VX |
| Seedream 5.0 Lite | Standard tier · BytePlus | 4K | 1 VX |
| Seedream 4.5 | Cinema tier · BytePlus | 4K | 1 VX |
| GPT Image 2 ★ | OpenAI | 4K | 4.7 VX |
| GPT Image 1.5 | OpenAI | 1080p | 2.7 VX |
| GPT Image 1 | OpenAI | 1080p | 3.5 VX |
| Nano Banana Pro ★ | 4K | 3 VX | |
| Nano Banana Flash | 2K | 2 VX | |
| Nano Banana 2.5 | 2K | 1 VX | |
| Grok Image Quality | xAI | 2K | 1.4 VX |
| Grok Image Fast | xAI | 2K | 1 VX |
| Kling Image | Kling | 2K | 1 VX |
| FLUX.2 Pro | Black Forest Labs | 2K | 1 VX |
| FLUX.2 Max | Black Forest Labs | 2K | 1.8 VX |
Per still at up to 1080p with no reference images; 1 VX = $0.10 list. Seedream is resolution-flat; most third-party routes add a small charge per reference image and scale at 2K and 4K. ★ marks the featured models. The Learning video tier has no image variant.
Key art to keyframes.
Every still a campaign owes, generated against one identity — from the pitch deck to the first frame of the film.
- Included:Campaign key art — hero frames for OOH, social, and web from one brief.
- Included:Product stills — packshots and lifestyle frames that hold one grade across the range.
- Included:Character sheets — multi-angle Cast sheets that lock a face before video ever runs.
- Included:Storyboard frames — board the sequence with finished-looking frames, not grey boxes.
- Included:Reference plates — art-direct a location or wardrobe once, then reference it in every shot.
- Included:Image-to-video keyframes — a still becomes the first frame of a video shot, identity intact.

Brief to key art, four moves
Bind the Cast
Attach a brand-locked identity to the brief. Cast references resolve to the member's verified assets at generate time — the identity lives in the roster.
Pick the route
Draft on Seedream 4.0, ship on Seedream 5.0 Lite — the production default — or master on Seedream 4.5. Route out to GPT Image 2, Nano Banana Pro, Grok, Kling, or FLUX.2 when the shot calls for it.
Generate with references
Steer the frame with up to 14 reference images per Seedream call. The exact VX cost is quoted before you run.
Send it to video
Promote the keeper to a keyframe — it becomes the first frame of an Image to Video shot, and the node canvas chains the handoff automatically.
Image generation, answered
Which models power the VisionX AI image generator?
How much does an AI image cost on VisionX?
How does the AI image generator keep characters consistent?
Can I turn a generated image into a video?
What resolution can I generate images at?
Do unused VX roll over to next month?
The rest of the platform
AI Video Generator
Cinema-grade video across Seedance, Sora, Veo, Kling, and Grok — one Cast, every engine.
ExploreConsistent AI Characters
The moat: a brand-locked character identity that holds across every shot, engine, and cutdown.
ExploreAI Lip Sync
Dialogue that lands on the face — sync any performance to any script, in any language.
ExploreAI Dubbing & Voice Cloning
The Audio Studio: AI dubbing, voice cloning, and text-to-speech tuned for campaign work.
ExploreAI Storyboard Generator
Board the campaign before you burn budget — beats, frames, and shot plans in minutes.
ExploreThe still and the spot, the same face.
Bind a Cast and run your first stills free — 100 VX, no card required.