Parallel VisionPARALLEL VISIONLAB CREATIVE STUDIO

Opening your workspace

PV LAB
Private workspace
PV LAB / CREDITS

Add credits

Choose a one-time pack.

Available 0 credits

One-time payment. Final total shown at checkout.

INVITE / PV LAB
Invite friends

Give 200. Get 200.

Share PV Lab with a friend. When they make their first purchase of €10 or more, you each receive 200 bonus credits.

PHOTO EDIT / REFERENCE PHOTOS

Nina reference photos

Choose 1–4 photos of Nina to guide Photo edit. This uses reference images, not a trained LoRA. Your BASE image keeps the scene and framing.

Generation connection

Wan 3.0 and Seedance 2.5 handle video. Seedream 5.0 Pro handles images and reference-based edits. Image Upscaler enlarges a finished image. Your existing provider key is used for these tools. Image Generate starts your selected image count at the live provider price. Upscale starts one paid upscaling job per click without a price-review popup. Video keeps price review before payment. Your computer does not render the media.

Provider access, permitted content and model availability must be checked in your own account. A model name alone does not guarantee unrestricted use. No safety filters are bypassed by this Lab.

Open provider · API documentation

Stored encrypted on your private backend. Never committed to GitHub or saved in browser storage.

UTC day. This Lab conservatively counts quoted maximums, including failed or uncertain attempts. Also set limits on the provider API key. Provider invoices remain the billing authority.

Review generation

Your prompt and selected inputs are shared with the provider only after confirmation. Higgsfield shows a live estimate and its final billed amount may differ. A result you dislike can still be charged. PV Lab never automatically submits a second generation.

PV Soul / Characters

Train Qwen or experimental Krea 2 characters from 20–80 photos. The browser prepares compact working copies, the private ZIP is stored temporarily in R2, and FAL runs one explicitly selected 1,000-step LoRA job. Finished weights are archived privately when possible.

Use varied, clear, non-explicit adult identity photos. Training data must comply with fal.ai terms. Krea 2 is text-to-image only and generates through fal.ai, not a base-image Reinterpret editor. Qwen generation remains on SpicyAPI.

Qwen Image 2512 text identity.

Training photos0 / 80

Characters

Saved identities

Repair Region

Paint the area to regenerate. Black stays unchanged; the painted white mask is repaired. This creates a new image and leaves the original untouched.

Input reference

Uploaded input reference, not a generated result
PV LAB / SOUL 2

YOUR CHARACTERS

Create and manage identities you can reuse across new images.

CONSISTENT CHARACTERS

One identity.
Endless possibilities.

Train your character with photographs, then place them in new scenes without losing their identity.

Saved Soul IDsYour characters

Add to folder