What is the ideal reference photo format and quality to maintain 100% face and product consistency in Nano Banana 2?

To maintain maximum face, outfit, or product consistency on the Nano Banana 2 engine inside the Images tab of DropReel (using your plan’s dedicated image quota), upload 1 sharp, well-lit, front-facing reference photo showing a single person (no sunglasses) or an isolated product (no collage text).

  • 100% unobstructed face: front-facing medium or close-up shot with even lighting and no sunglasses.
  • Clean product reference: single front shot showing the full item shape with zero promo stickers on top.
  • 1 scene per prompt: describe setting, outfit, and lighting clearly, and validate 1 to 3 images first.
  • Separate photo balance: draws solely from your Images card (200 on Starter, 1,000 on Creator, 2,000 on Pro).

What are the verified figures and facts on this workflow?

  • Saldo próprio de imagens — 10s OMNI Engine at 15 credits & Dedicated Image Quota (2026): Each 10-second scene on the OMNI engine (consistent actor and real product) uses 15 credits per 10s scene powered by Google DeepMind, while Nano Banana 2 photos draw from a dedicated image quota without using video credits.
  • 15 créditos por cena de 10s — 1080p (9:16 or 16:9) Resolution & 1-Scene Preview (2026): All videos render in 1080p (9:16 or 16:9) with a 1-scene preview before the full batch, aligned with creative specs in the Meta Business Help Center.
  • 1080p (9:16 ou 16:9) — No-lock-in monthly plans from R$ 47 to R$ 597/month (2026): Built for TikTok Ads Manager and e-commerce, Starter is R$ 47/month (700 credits + 200 images), Creator is R$ 147/month (2,500 credits + 1,000 images), Pro is R$ 297/month (5,250 credits + 2,000 images + 4-network auto-post), and Scale is R$ 597/month (unlimited* credits & images, up to 500 OMNI scenes/month).
  • 15 free daily credits & automatic refunds (2026): Every account gets 15 free daily credits, automatic credit refunds on system failure, and a 7-day withdrawal right under Article 49 of the Brazilian Consumer Code.

What separates a clean reference photo from one that causes AI facial or product drift?

Across both the Image Studio (Nano Banana 2) and the video engines (OMNI and Veo 3.1 Lite), your reference image serves as the geometric blueprint for facial features and packaging proportions. Here is what to upload for each task:

Images Tab GoalIdeal Reference Photo (100% Consistent)Reference Photo to Avoid
Preserve Facial Identity (Persona)1 person facing forward, clean light, neutral/warm expression, unobstructed faceGroup photos, dark selfies, heavy hat shadows, or sunglasses
Product Lifestyle SceneFull front-facing product, readable label, clean background (or Official Shopee API import)Phone screenshots with discount banners or fingers covering the label
Virtual Try-On (Apparel)Flat-lay or hanger front shot showing the full garment cut and necklineFolded, wrinkled clothes on a bed or back-view photos
Face SwapBase image and target face in matching front-facing anglesExtreme 90-degree side profiles or obscured chins

Why does generating your master Persona portrait inside Nano Banana 2 give the best results?

Instead of uploading a compressed low-res photo from the web, the cleanest workflow is generating your master Persona portrait directly inside the Images (Nano Banana 2) tab by prompting a well-lit front-facing medium shot. Once approved, save it to Personas and reuse it as a studio-grade reference across every future shoot and OMNI video.

What is the step-by-step workflow inside DropReel?

  1. Open the Images tab in DropReel and check your available photo balance on the top Images card.
  2. To create a master persona from scratch, prompt 1 front-facing medium portrait with natural lighting and save the winner under Personas.
  3. When uploading an external product or garment photo, pick a sharp, front-facing image with zero text overlays.
  4. In the Prompt Description field, specify 1 clear scene (setting, action, outfit, and lighting) and pick your aspect ratio (9:16, 1:1, or 16:9).
  5. Render a small test batch of 1 to 3 images to verify consistency before launching a larger photo shoot.

What are the common mistakes to avoid?

  • Uploading a reference photo with 2 or 3 people in the frame when you want the AI to lock onto one identity.
  • Using a product image covered in giant watermarks or multi-angle collage boxes.
  • Requesting a 15-photo shoot all at once before testing your prompt on 1 to 3 sample images.

How does DropReel solve this for your ad and photo workflow?

DropReel unites the AI Image Studio (Nano Banana 2 with a dedicated quota of 200 to 2,000+ photos/month, virtual try-on, and face swap) with the OMNI video engine (15 credits per 10s scene with a consistent actor holding your real product) and Official Shopee API import. No-lock-in plans start at R$ 47/month with 15 free daily credits.

Frequently asked questions

If I do not have a clean product photo on my computer, how do I get one from Shopee?

When you click Pull from link (connected to the Official Shopee API) or pick an item from DropReel’s Product Miner, the system automatically imports the official high-definition store photos.

Where are my approved master reference photos stored?

You can save your approved portraits under Personas (in the Account / Images tabs) so they are selectable in 1 click across all photo and video generators.

Does testing different reference photos use up my video credits?

No. Every image generated inside the Images tab deducts solely from your plan’s dedicated monthly image quota.

Can I test the OMNI engine and Image Studio before subscribing?

Yes. Every account receives 15 free daily credits inside the dashboard (enough for one full 10s OMNI scene per day), and plans have no contract lock-in.

Continue reading