1 · Lock the product
One clean front shot of the physical item + packaging artwork feed the SKU node. The PDP main image stays a real product photograph — the platform says so, and we agree.
TikTok Shop listing images follow a strict, boring spec — up to 9 square images at 600×600+ (800×800 on some sites), a pure-white front-view main image with the product filling at least 60% of frame, no text, logos or watermarks, no repeated angles, digital renderings not allowed — plus exactly one video per listing capped at 5 MB. The actual war is fought in ads and organic clips, where creative burns out in days. Berrys runs both fronts: disciplined compliance sets for the PDP, and a weekly pipeline of 9:16 hooks generated from packaging-true product nodes.
Reviewed October 2026 · Prepared by Berrys · Technical specifications & Canvas node workflows
| Parameter / Slot | Marketplace & Production Requirement |
|---|---|
| Images | Up to 9 square images per listing; ≥600×600 px (800×800 on some sites) |
| Main image | Front view of the actual product · pure white background · product ≥60% of frame · no text, logos, watermarks, borders or staging with multiple variants |
| Angle rule | Each image a different view (front/back/side/detail); same angle repeated not allowed |
| Representation | Images must show what the customer receives — placeholders and digital renderings are not allowed |
| Video | Exactly 1 video per listing, max 5 MB — short and tightly compressed wins here |
| Branded goods | Brand identifiers (name, logo, slogan) must be clearly visible in images/video |
| Description | Valid description >80 characters + category-specific specs to reach the 'Good' listing tier |
Test your packshot cover image for solid white background (#FFFFFF), product frame-fill percentage and 1:1 aspect ratio before uploading to Seller Centre.
One clean front shot of the physical item + packaging artwork feed the SKU node. The PDP main image stays a real product photograph — the platform says so, and we agree.
Slots 2–9 come from the canvas: detail macros, usage scenes, size-comparison and accessory shots, each a genuinely different angle per the listing-quality tier guide.
Image-to-video on Google Omni Flash or Kling 4.0 gives a 5–10s hero motion from the pack shot; Video Trim plus compression settings fit the 5 MB cap. One clip chosen to work as both PDP video and ad seed.
For Spark/Shop ads: Prompt Helper writes five hook variants (problem-first, demo-first, price-anchored, UGC-style, before/after), each re-running the same scene graph with a different first frame. Refresh cost per week: credits, not shoot days.
Image Compare node side-screens the new set against last week's best performers before anything ships to spend.
The same hook that printed last week is invisible this week. If refresh needs a shoot + edit cycle, you're paying fatigue tax every cycle. A saved canvas workflow re-runs in hours.
That cap forces short, tight, motion-dense clips. Long cinematic renders simply won't upload. Berrys' Video Trim node exists exactly for this: cut to the beat, re-export under budget.
AI video's tell is over-clean lighting. The fix is scene prompting (handheld framing, bedroom/desk contexts, imperfect surfaces) around a product that stays packaging-true — motion is where generic models break geometry, so SKU locking matters more in video than images.
Veo 3.1 and Omni Flash for motion realism, Kling 4.0 and Seedance 2.5 for dynamic scenes, MiniMax H3 and Runway Gen-4.5 when a recipe needs them — switch per job inside one canvas instead of buying four tools.
Agencies run each seller account as its own brand system; nothing bleeds between client SKUs. Berrys Studio adds human art direction on flagship accounts.
TikTok Shop penalties for misleading content are existential. Generated clips must show the product exactly as shipped — that is the whole reason Berrys locks geometry to your real packaging files.
Already working in an AI agent? Connect Berrys to Claude, Cursor, ChatGPT, Codex, or Hermes. Your agent plans the tiktok shop suite and runs the shots on your Berrys canvas, with saved Brand Memory and credit approval in chat.
https://app.berrys.io/api/mcpRemote MCPUsing Berrys, create a TikTok Shop compliant listing set and video hook for my product reference in ./assets: 1. File node: Ingest front-facing real product photo. 2. Brand System node: Set pure white background with ≥60% frame fill. No digital mockups on main cover. 3. Image node (Nano Banana 2.1): Render 1024×1024 main cover plus 3 distinct angle crops (back, side, macro). 4. Video node (Omni Flash): Render 5s 9:16 vertical motion clip compressed under 5MB. Estimate credit cost and request approval before generating.
TikTok Shop's PDP rules explicitly ban placeholders and digital renderings in listing images and demand front-view product truth on white. Anyone telling you to replace your main image with pure AI generation is putting your account at risk. The proven play: photograph the front once, let Berrys handle angles, details, motion and the weekly ad-creative churn.
Listing rules require images/video that accurately represent the product and forbid 'digital renderings' in listing images. The safe architecture — and what we recommend — is a real photograph as the main image, with AI used for secondary angles, motion clips and ad variants that depict the actual product unchanged. Keep everything verifiable; misleading content risks takedowns and account-level penalties.
Most accounts that scale refresh 3–5 hooks weekly per product. With Berrys, each variant is a re-run of a saved graph (minutes), not a reshoot (days).
Google Veo 3.1 and Omni Flash, Kling 4.0, Wan 3.0, Seedance 2.5, MiniMax H3 and Runway Gen-4.5 — routed per shot inside one canvas subscription.
Yes — each brand runs as an isolated Brand System, so client catalogs never cross-contaminate. This is a core Berrys Studio workflow.
Start free on Berrys Canvas — 50 credits, no card required — or send Berrys Studio a brief and get finished campaign visuals with agency art direction.