STUDIO SERVICES
Production disciplines, forged to a single finishing standard
PixelForge delivers generative-AI visual work as a creative studio — not software you download, not a course you enrol in. Every discipline below includes human art direction, diffusion-model generation where appropriate, and documented usage rights for Canadian client organizations.
1. Generative Brand & Campaign Imagery
Your launch needs twelve distinct hero images but the photo budget covers three locations. We build a generation pipeline — prompt design, reference images, model selection — then art-direct every frame for brand consistency across social, web and print. Text-to-image and image-to-image diffusion models produce candidates; human review eliminates drift before any image reaches a revision round. Typical CAD range: C$6,800–C$18,000 per campaign scope. Turnaround: two to four weeks depending on deliverable count and style-training requirements.
2. Product & Packaging Visuals
A CPG line with forty SKUs cannot afford forty studio days. We generate product visualization from your dielines, material references and brand palette — synthetic photography that reads as shot, not generated. Inpainting handles label placement; retouch and upscale bring files to print resolution. Each SKU passes a human-in-the-loop check for artefacts and unintended similarities. Per-deliverable from C$180; catalogue projects from C$2,400. Revision rounds and usage rights are scoped per project.
3. Synthetic Photography & Composites
Sometimes the scene exists only in the brief — a kitchen at golden hour with your product on the counter, a retail shelf in a store layout you have not built yet. We composite generated environments with product assets, matching colour and lighting through art direction and image editing. Composite builds combine diffusion output with photographed elements where authenticity matters. Projects typically C$3,200–C$9,500. We document which elements are synthetic and which are photographic in deliverable notes.
4. Custom Style Training & Brand Consistency
One-off prompts do not hold a brand across a hundred images. We fine-tune models — LoRA and style-training workflows — on your approved reference library so on-brand generation becomes repeatable. Visual concepting sessions define what "on-brand" means in generative terms: palette, texture, lighting ratio, composition rules. Training deliverables include model documentation and usage guidance for your team. Style-training projects from C$4,500; ongoing retainer arrangements available for Canadian brands with continuous output needs.
5. Retouch, Upscale & Pixel-Level Finishing
Generation gets you eighty percent there; finishing gets you to production. We retouch skin tones, fix edge artefacts, upscale to campaign resolution, and perform pixel-level corrections that diffusion models cannot reliably produce. Image editing covers inpainting problem areas, outpainting for aspect-ratio shifts, and colour grading aligned to your brand guidelines. Finishing-only engagements from C$95 per image; batch rates for catalogue work. Every finished file is human-reviewed before delivery.
6. Art Direction, Usage Rights & Licensing
Responsible AI means clear paperwork. We document usage rights per deliverable: channels, territories, duration and sublicensing terms. We do not create deepfakes, do not use real people's likeness without consent, and flag potential copyright or trademark conflicts before generation begins. Art direction covers the full project — moodboard alignment, revision rounds, final sign-off. This is not legal advice on intellectual property; clients remain responsible for final usage clearance. Licensing documentation included on every CAD project.
Rights disclaimer: Our studio provides generative-AI visual services — image generation, style training, editing, retouching, rendering, concepting and illustration — produced with human art direction and review. Images are created using generative AI tools and are checked by people; AI outputs can contain errors, artefacts or unintended similarities and always require human review. We clarify usage rights and licensing for deliverables and do not knowingly create images that infringe third-party copyright or trademarks, depict real, identifiable people without consent, or present synthetic images as genuine photographs where that would mislead.