Is ImageForge AI a text-to-image app, an AI-art course, or a deepfake service — and do you guarantee the images or that they'll go viral?
No. We are an AI image-generation studio that art-directs, generates, and finishes imagery for client brands — we do not sell a self-serve app, a course, or stock images, and "forge" has nothing to do with crypto or metalwork. We generate from client-owned or licensed references, respect IP and likeness rights, obtain consent where needed, disclose AI use where required, and do NOT create deceptive deepfakes. Generative images can have artefacts, so a human reviews, retouches, and finishes every deliverable. We cannot guarantee a specific style match, reach, or virality; outcomes depend on the brief, references, project scope, and how the imagery is used.
How do engagements work — project vs retainer?
Project engagements have a defined deliverable list, timeline, and CAD budget with milestone payments. Retainers reserve studio capacity monthly — typically a set number of image batches, revision rounds, and art-direction hours. Both models include human review and documented licensing terms. We scope every engagement in writing before generation begins.
What are typical CAD budgets?
Concept sprints often start around C$2,800–C$5,000. Product imagery builds range from C$5,500 to C$22,000 per SKU set. Campaign production may run C$8,000–C$35,000 depending on volume and custom style training needs. Retainers begin near C$6,500 per month. Exact figures depend on your client brief — we provide quotes after reviewing scope, not from a price list alone.
How long does onboarding and turnaround take?
Initial scoping calls happen within one business day of contact during Mon–Fri 09:00–18:00 ET. Onboarding — reference clearing, contract signing, training data intake — typically takes three to five business days. First delivery for a standard product batch often lands in five to nine business days after materials are cleared. Rush timelines are sometimes possible but never guaranteed.
Which models and tools do you use?
We work with cloud-hosted and open-weight diffusion models selected per project requirements — text-to-image, img2img, inpainting, outpainting, and upscaling pipelines. Custom style training uses LoRA fine-tuning on licensed or client-owned material. Tool choice follows the brief, not a single vendor preference. We do not resell model access or checkpoints as a marketplace product.
Who owns the imagery, prompts, and trained styles?
Upon full payment, clients receive agreed usage rights to deliverables as specified in contract — typically broad commercial licence for campaign and product imagery. Prompt libraries and trained style models may transfer or remain studio-managed depending on scope. We retain no claim to your brand trademarks or client-owned reference assets. IP terms are documented before work starts.
How are references and training data licensed?
All reference images and training sets must be client-owned, properly licensed, or cleared for the intended generative use. We maintain a reference log: source, licence type, consent status, and retention period. We reject scraped web imagery, unclear-rights stock, and competitor assets without explicit permission.
How do you handle likeness and consent?
Real-person likeness requires documented consent before generation. We do not create non-consensual likeness, impersonation, or exploitative imagery. Synthetic talent is labelled as such. When your brief includes identifiable individuals, we require signed releases aligned with Canadian privacy expectations and your campaign governance.
How is AI use disclosed?
We support client AI disclosure policies — metadata notes, caption language, and platform-specific labels where required. Our contracts clarify what disclosure obligations sit with the studio versus the client at point of publication. Responsible AI practice includes honest representation of how imagery was produced.
How do you catch mistakes and artefacts?
Every batch passes human review against an image quality control checklist: hands, text, logos, unintended likeness, colour drift, and resolution. Diffusion models hallucinate details; that is why human-in-the-loop retouching is mandatory, not optional. Clients receive revision rounds defined in scope; additional rounds may incur extra fees.
How is personal data handled under PIPEDA?
Contact form data is collected with explicit consent and used only to respond to enquiries and deliver contracted services. See our Privacy Policy for retention, access rights, and Privacy Officer contact. We do not buy or sell personal data or personal images.
What do you NOT do?
We do not guarantee virality, reach, or sales. We do not create deepfakes, non-consensual likeness, or NSFW content. We do not offer get-rich-with-AI schemes, NFT or crypto art drops, courses, stock or prompt marketplaces, or self-serve SaaS subscriptions. We do not buy or sell personal images or data.
Ask a question directly
