Questions we hear in first calls

No — we are a studio. VisualSynth is a generative-AI visual production company based in Toronto. Clients hire us for art-directed campaign imagery, product visuals, and synthetic photography with documented usage rights. We are not a consumer text-to-image app, not a royalty-free stock site, and we do not create non-consensual deepfakes or deceptive impersonations. If you need a dozen on-brand visuals for a product launch with a licence your legal team can read, that is us. If you need unlimited free downloads of random images, it is not.

Consumer tools are excellent for exploration; they are weaker at repeatable brand production. We bring prompt architecture, custom style training, retouch and colour pipelines, version control, and contractual deliverables. Your in-house team keeps access to their tools; we absorb the production volume and liability documentation when deadlines or consistency requirements exceed what informal prompting can guarantee.

Ownership and licence scope are defined per statement of work. Most commercial clients receive broad usage licences for agreed channels and territories; work-for-hire arrangements are available when required. We do not recycle your approved masters for other clients. Training adapters on your assets remains separate — it requires explicit permission and may involve additional fees and deletion schedules.

Small batches with existing style pipelines can turn in a few business days. New brand systems, custom training, or large variant sets typically need two to four weeks including review rounds. Rush work is possible with stated overtime rates. We will not promise overnight delivery for a twelve-scene campaign without a brief — that is how off-brand work happens.

Yes. Agencies embed our producers during pitches and production bursts under NDA. Invoices can reference your client or VisualSynth depending on tax and procurement needs. We remain invisible in client-facing decks when requested.

Contact submissions are processed under our Privacy Policy with PIPEDA-aligned consent. Client brand assets used for generation are stored in access-controlled environments with retention limits defined in contracts. We do not sell contact data or use your imagery to train public models.

Often, yes — via reference-controlled generation and optional custom adapters trained on your approved library. We evaluate feasibility in briefing: some styles depend on proprietary lighting rigs that diffusion models approximate but do not perfectly replicate. We will tell you where hybrid compositing or traditional reshoots are more honest.

Common exports include TIFF, PSD with layers where scoped, high-resolution JPEG or WebP for digital, and PDF proofs for print review. Video loops for DOOH are quoted separately. Naming conventions and DAM metadata can follow your taxonomy on retainer accounts.