Question 1

What is Palmou AI?

Accepted Answer

Palmou AI is a web-based AI product-photography studio at palmou.com built by Palmou Studio in Spain. It runs two image operations end-to-end inside a single credit: intelligent background removal and AI scene generation on the product image. Each operation uses a different best-of-class 2026 model — BRIA RMBG 2.0 for background removal, and Google Gemini 2.5 Flash Image (nano-banana) for scene generation. The product is aimed at Shopify, Amazon, Etsy, and WooCommerce sellers, indie brands, and agencies producing catalog imagery without booking a photographer. Each finished image consumes 1 credit; plans start at $19/month, with a free tier that ships 1 credit on signup. End-to-end latency is under 15 seconds.

Question 2

Does Palmou AI remove backgrounds from product photos?

Accepted Answer

Yes. Intelligent background removal is one of Palmou AI's two core operations and is included in every credit. It is powered by BRIA RMBG 2.0, the 2026 state-of-the-art background remover. Unlike older binary-mask tools (U-2-Net, ModNet, in-house Photoroom/Pixelcut models), BRIA outputs a 256-level alpha channel that preserves soft transitions: hair, fur, glass edges, jewelry chains, watch crystals, knitwear texture, fringe, smoke, ice, partial reflections. If you don't pick a scene in the editor, you receive a transparent PNG with the cutout — ideal for compositing into your own designs. If you do pick a scene, the scene-generation step replaces the background end-to-end without needing the cutout as input.

Question 3

Can Palmou AI generate new scenes for my product photos?

Accepted Answer

Yes. AI scene generation on your product image is the second of Palmou AI's two core operations. It is powered by Google Gemini 2.5 Flash Image (also known as nano-banana), an image-edit model that places your product in a completely new environment without requiring a precomputed mask. Lighting, shadow, and reflections adapt to the actual product instead of being baked into a fixed background — that's the difference between generative scenes and template-based tools like Pebblely. Six scenes ship today: White Studio, Premium Beige, Nature, Luxury Marble, Urban, and Gradient. They cover the most common e-commerce categories (cosmetics, jewelry, fashion, electronics, food, lifestyle). Free-form custom prompts are not exposed to keep output quality predictable.

Question 4

Will AI-generated mockups look fake or generic?

Accepted Answer

Palmou AI doesn't generate mockups. The distinction matters: a mockup synthesizes a fake product from a description (text-to-image), and that's where the "AI/stock photo" look comes from — the product itself is invented and reads as generic. Palmou uses an image-edit model (Google Gemini 2.5 Flash Image) that reads your actual product photo and only modifies the surrounding scene. Your product's shape, color, label text, logos, and texture are preserved exactly. The shadows and lighting in the new scene adapt to the real product, which is the property that makes the output look photographed rather than pasted. If you're worried about the Etsy or Vinted backlash against AI mockups, this is the technical reason Palmou doesn't trigger it — buyers can't tell the background was replaced because the product is the real one.

Question 5

What resolution does Palmou AI output?

Accepted Answer

Background-removal-only jobs preserve the source image's native resolution — if you upload a 4K phone photo, you get back a 4K cutout with the same dimensions. Scene-generation jobs output at roughly 1024 px on the longest side, sized for catalog listings, social ads, and most e-commerce platforms (Amazon's recommended minimum is 1600 px and Shopify renders product images at much smaller sizes). Output format is WebP at quality 92, optimized for web delivery while retaining catalog-grade fidelity. Every plan, including the free tier, ships images with no watermark.

Question 6

What's the best AI tool for product photography in 2026?

Accepted Answer

It depends on the use case. Palmou AI is the strongest option for solo founders and small brands who need fast turnaround and don't want to manage two separate tools. It runs the 2026 SOTA pipeline (BRIA RMBG 2.0 + Google Gemini 2.5 Flash Image) end-to-end with a flat 1 credit per image and under 15 seconds end-to-end latency. Photoroom is broader and mobile-first; Pebblely is template-driven; Pixelcut is mobile-only; Canva Magic Studio is general design with AI bolted on. For desktop catalog work where single-image fidelity matters more than batch volume, Palmou AI typically wins; for phone-first or thousand-SKU pipelines, Photoroom is a better fit.

Question 7

How does Palmou AI compare to Photoroom?

Accepted Answer

Photoroom has a much larger surface area (mobile apps, batch tools, native Shopify integration) and uses in-house background-removal models. Palmou AI is web-only and narrower — one workflow (single image → finished output) — but uses BRIA RMBG 2.0, generally the highest-fidelity background remover on the public market in 2026 thanks to its 256-level alpha channel. For a single-image quality benchmark, Palmou AI tends to win, especially on hair, fur, glass, jewelry, and reflective products. For volume on phone, Photoroom wins. Pricing is comparable: Palmou is flat $19/$49/$99 per month with no per-feature paywall.

Question 8

How does Palmou AI compare to Pebblely or Pixelcut?

Accepted Answer

Pebblely is template-driven — you pick a curated background and the tool composites your cutout into it. Palmou AI takes a different approach: instead of templates, Google Gemini 2.5 Flash Image generates each scene end-to-end, which means lighting and shadows adapt to the actual product instead of being baked into a fixed background. Pixelcut is mobile-first and freemium; Palmou AI is web-first and prioritizes single-image fidelity over mobile UX. If you want template consistency, pick Pebblely. If you want generative scenes that adapt to the product, pick Palmou AI. If you work primarily on a phone, pick Pixelcut.

Question 9

Can I use the images commercially?

Accepted Answer

Yes, on every plan. You own all generated images and can use them anywhere — Amazon listings, Shopify product pages, Etsy listings, ads on Meta or TikTok, print catalogs, billboards. BRIA RMBG 2.0 is licensed for commercial use through Replicate, and Gemini 2.5 Flash Image outputs through Replicate are also licensed for commercial use. There are no per-month image caps after they're generated; once a credit produces an image, the image is yours forever. The only restrictions are the obvious ones: don't use Palmou AI to generate images that violate someone else's IP or dignity (Palmou's content moderation will flag many such attempts and refund the credit).

Question 10

Are my source images used to train AI?

Accepted Answer

No. Source uploads are stored encrypted at rest in Supabase Storage in the EU region and are never sent to any model-training pipeline — neither Palmou's nor any third party's. The only outbound network call is the inference request to Replicate, which produces the output and returns it to your namespace; the source isn't retained on Replicate's side beyond the request. You can delete any image from the gallery at any time, which removes the source, the processed output, and any intermediate masks in a single transaction. Deleting your account purges everything end-to-end.

Question 11

How long does processing take?

Accepted Answer

Under 15 seconds end-to-end for both flows: roughly 6 seconds for background-removal-only jobs (BRIA RMBG 2.0 runs synchronously on a warm-keep cron, no queue) and around 10–15 seconds when a scene is selected (Google Gemini 2.5 Flash Image is one of the fastest image-edit models available in 2026). The editor shows a live elapsed counter and step-by-step progress so you know which stage is running. End-to-end latency is dominated by the inference call, not the application — Replicate model warm-keep crons trim cold-start delay, and the webhook fallback recovers any stuck job within a few minutes if a webhook drop occurs.

Question 12

What if the result isn't satisfactory?

Accepted Answer

You can reprocess the same source for 0.5 credits with a different scene. If the pipeline clearly failed — content moderation false positive, Replicate timeout, transient model error — the credit is refunded automatically and the failure is logged. We don't manually review or judge artistic dissatisfaction (the output is a probabilistic generation, and tastes vary), but the half-credit reprocess option exists exactly so you can iterate cheaply. If you hit a bug or a clearly broken output, email palmoustudio@gmail.com with the image ID and we'll refund manually if it's our fault.

Question 13

Is there an API for bulk processing?

Accepted Answer

Yes, on the Agency plan ($99/month, 1,500 credits/month). The REST API has two endpoints: POST /api/v1/images submits a job (accepts a public image URL and an optional scene_id) and returns 202 with an id and status_url; GET /api/v1/images/:id polls status. Authentication is a Bearer API key generated at /api-keys (only the SHA-256 hash is stored, plaintext shown once). Rate limit is 60 requests per minute per key. Full reference at palmou.com/docs/api. The API is the right choice for migrating an existing Shopify or Amazon catalog or for integrating Palmou into your own internal tooling.

Question 14

Does it work with reflective, transparent, or textured products?

Accepted Answer

Yes, and this is one of the strongest cases for picking Palmou AI over older alternatives. BRIA RMBG 2.0's 256-level alpha channel preserves the soft transitions that older binary-mask models destroy: glass edges, jewelry chains, watch crystals, transparent packaging, hair, fur, fringe, knitwear texture. Reflective metals (chrome, polished steel, mirrored finishes) come out clean because BRIA models the partial transparency of the highlight rather than treating it as either fully opaque or fully transparent. The downstream Gemini 2.5 Flash Image scene-generation step preserves those soft edges through the rest of the pipeline.

Question 15

Is the output Amazon-compliant?

Accepted Answer

Yes. The White Studio scene produces a pure-white seamless background that meets Amazon's main image requirement (RGB 255, 255, 255). The Premium Beige and Luxury Marble scenes are appropriate for Amazon A+ content secondary images and Shopify hero imagery. Output resolution from the scene-generation step (~1024 px on the longest side) sits below Amazon's recommended 1600 px main-image minimum, so for Amazon main images we recommend the background-removal-only flow which preserves your source resolution. We don't apply text overlays, brand watermarks, or any visual element that would violate Amazon's image guidelines — just the product, the background, and clean shadows.

Question 16

Where is the company based and is it GDPR-compliant?

Accepted Answer

Palmou AI is built by Palmou Studio, an independent software studio based in Spain. Because we're EU-based, we operate under the EU General Data Protection Regulation (GDPR) and Spain's LOPDGDD. Data residency is in Supabase's EU region; Stripe handles payments under their own DPA; transactional email is Resend. The legal pages at /legal/privacy, /legal/terms, and /legal/cookies cover data subject rights (access, rectification, deletion, portability, objection), the legal bases for processing, and the cookie inventory. You can download a copy of all your data from /billing (right to portability under GDPR Art. 20) or email palmoustudio@gmail.com to request full account deletion, which we process within 30 days.

Feature	Palmou AI	Photoroom	Pebblely	Pixelcut
2026 SOTA models	✓BRIA RMBG 2.0 + Gemini 2.5 Flash Image	—In-house (older gen)	—Template-based	—In-house (mobile)
End-to-end latency	✓Under 15 seconds	—10–30 seconds	—30–60 seconds	—10–30 seconds
Free-form prompts	✓Yes (Pro)	—Limited	—No (templates only)	—Limited
Flat per-image pricing	✓1 credit / image	—Subscription tiers	—Subscription tiers	—Subscription tiers
Commercial license	✓Yes, all plans	✓Paid plans	✓Paid plans	✓Paid plans
API access	✓Yes (Agency)	✓Yes (Enterprise)	✓Yes	—Limited

Your product photo
studio, in a tab.

Drag to see the difference.

Three 2026 SOTA models.
One credit per image.

Edges nobody else gets right.

Your product stays your product.

Catalog-ready in under 15 seconds.

Palmou AI vs Photoroom, Pebblely, Pixelcut

We don't invent your product. We keep it exactly.

Simple credit-based pricing

Starter

Pro

Agency

Frequently asked questions

Start processing today.

Your product photostudio, in a tab.