Frequently asked questions
Plain answers about how NotShot's AI catalog rendering works, quality bar, pricing, and supported categories.
What is NotShot?
NotShot is a B2B AI rendering platform for fashion catalogs. Brands and designers upload a garment photo and a model photo and receive production-grade try-on renders — the model wearing the garment — at full catalog quality. Each render is judge-graded for fabric fidelity and framing before being delivered. After the initial render, lighting and texture variations can be applied to the same template at a fraction of the cost.
Which garment categories are supported?
Four live categories at launch: jackets, pants, dresses, and bags (handbag class). Each category has its own labs-validated rendering recipe, garment-model compatibility rules, and judge taxonomy. Shirts, tops, and shoes are on the v1.x pipeline roadmap, gated on labs validation reaching the same quality bar as the live four.
Does NotShot work for ecommerce product photography?
Yes — NotShot is designed for catalog and ecommerce photography at scale. Brands and creative agencies use it to replace or augment studio shoots when a garment's design needs visualizing before sample production, when N color or size variants of a catalog need rendering without re-shooting, or when downstream creative tools need a stable on-model template to feed their own AI pipelines. Output is delivered at full catalog resolution (4K) suitable for ecommerce listings, advertising, and print catalogs.
How does NotShot compare to a traditional studio shoot?
A traditional on-model catalog shoot typically costs in the thousands of euros per garment (studio rental, model day rate, photographer, retoucher) and takes one to several days from scheduling to delivered files. NotShot delivers the same on-model catalog asset for tens of dollars per garment in under an hour, with judge-graded quality control and automatic refund on sub-threshold output. Lighting and texture variations of an existing render are a fraction of the original render cost, where a studio re-shoot would cost the same as the original. NotShot is suited for catalog production at scale; bespoke creative photography (editorial, brand films, hero campaigns) remains a studio strength.
Can multiple users from the same brand collaborate on a tenant?
Yes. Each brand has its own tenant. The first user becomes the tenant admin and can invite additional users via email invitation. Subsequent users join as members; admins control billing, invitations, and tenant settings. All shoots, runs, jobs, and renders are tenant-scoped — no cross-tenant data exposure.
How does the rendering quality bar work?
Every rendered image is scored by an AI judge against a fabric-fidelity, framing, and defect rubric. Renders with a judge score at or above the production threshold ship as the customer's deliverable. Below the threshold, NotShot automatically retries with a different seed; if the retry also fails, the customer is fully refunded. This is the single-shot-with-threshold contract — uncertified iterations never ship.
What lighting variations does NotShot support?
Six labs-validated lighting presets are available as cheap post-render variations on an existing template: studio neutral (free, no relight), soft daylight studio, warm key light, dramatic high-contrast, outdoor overcast, and window light. Switching lighting on an existing template is a fixed-credit operation — designers see the credit cost before they commit. This separates expensive primary rendering from cheap, re-runnable design variation.
How is pricing structured?
Pricing is credit-based, with credits purchased via Stripe in USD. Each shipped render deducts a fixed credit count for the primary 4K on-model generation; each post-render module (relight, retexture, pose, future scene-swap) is a separate transparent credit cost shown before commit. Designers decide per-action whether a paid variation is worth the cost; the free studio-neutral preview remains the default.
What happens to a render that fails the quality bar?
On a single-shot render below the production threshold, NotShot retries once automatically with a different seed. If the retry also fails, the customer is automatically refunded their credits and the job transitions to failed with a structured reason. Defective outputs (cropped feet, missing garment fidelity, unrelated subject) are handled identically — no flawed render is ever delivered as a customer deliverable.
Why are renders persisted even when not shipped?
Every render attempt — including iterations that fall below the ship threshold — is persisted as a forensic-only S3 asset with linked judge scores and metadata. This gives the labs team a full inspectable history of what failed and why, for the same (garment, model) tuples that customers actually use. The forensic bucket is operationally separate from the customer-deliverable bucket; lifecycle, IAM, and BCP differ.
Can NotShot-rendered images be used commercially?
Yes. NotShot is a commercial B2B product. Brands and designers can use the rendered output in catalogs, e-commerce listings, advertising, and downstream AI workflows. Customer ownership of input photos and rendered output is documented in the Terms.
What model identities can be used?
NotShot works with any model photo the customer has the right to use. The platform does not generate synthetic models — every output retains the identity of the supplied model photo, only the garment and lighting are modified. Customers are responsible for ensuring they have rights to the model imagery they upload.
How fast are renders delivered?
Each primary 4K render takes approximately 30 to 90 seconds end-to-end (render + judge + delivery). Post-render lighting and texture passes complete in 20 to 60 seconds each. NotShot processes multiple jobs per worker concurrently so peak throughput at scale is dictated by the upstream rendering capacity rather than worker capacity.