Layer Separation vs Background Removal: Which to Use?

Start with the next edit you need to make

Choose based on the editing task rather than the file extension. If you only need to place a mug on a new backdrop, extracting that mug may be sufficient. If the mug, leaf, orange and display plinth must be moved independently, ask for several layers. A PNG can contain transparency without storing any separate design layers; saving that same PNG as PSD does not automatically separate its objects.

What each Png2Psd workflow returns

Prompt Cutout returns one PNG for the object described in your request. It does not return a PSD stack or a matching clean background. Image to Layers produces inferred raster layers that can be inspected and exported as PSD or individual PNG layers. Both use AI, so fine lettering, product labels, edges and small details can change. An exported PSD contains raster elements, not recovered font or vector objects.

A cutout is different from a reconstructed background

Removing the background around a subject and reconstructing the scene behind that subject are different tasks. A transparent foreground alone does not reveal what was originally hidden. In a separated composition, the base image may still contain the foreground object. Moving it can reveal duplicated content, gaps or generated texture. Inspect the entire composition after moving a layer; neither a PSD extension nor a successful task proves that the hidden background is accurate.

Write a prompt for one result or for several layers

For one object, try: Keep only the teal ceramic mug, including its handle. For a layered composition, try: Separate the ceramic mug, orange, leaf and plinth into individual layers; preserve their positions. These are requests to guide processing, not guaranteed output specifications. Start with the few elements needed for your edit and review the actual result instead of assuming that every named detail becomes an independent layer.

Compare the real sample before processing your own image

The product example below has a self-owned original and five actual output layers. Inspect the isolated mug, orange, leaf and plinth, then compare the base/reference layer. The sample demonstrates object separation; it does not establish a separate clean shadow layer or flawless removal from the base. Public samples can be inspected and downloaded without an account. Processing your own image requires sign-in and sufficient credits.

Check the output against your intended use

Inspect the object on both light and dark backgrounds, check missing parts and halos, and compare labels and geometry with the original. If you plan to move objects, also inspect their old positions in the base layer. Prompt Cutout can return an opaque PNG, which the interface flags. Download and inspect your final composition before using it in a product listing, advertisement or animation preparation workflow.

Compare credit use without assuming a fixed image allowance

Prompt Cutout costs 10 credits for each successful extraction. Layer separation costs 2 credits per output layer up to 2,610,000 input pixels, or 4 credits per layer above that threshold. New accounts receive 50 signup credits once. Both workflows require sign-in; failed tasks do not consume generation credits. Use the pricing calculator for the current rules rather than treating one image as a fixed credit cost.

Practical tips

• One product for a catalogue image: consider a single cutout and inspect its alpha channel
• A product, orange, leaf and plinth that need separate positioning: choose layer separation
• Only need the original flat picture inside a PSD: a normal image editor may be enough
• Need editable font text, vector paths or a Live2D rig: preserve the source file or prepare those parts manually

How it works

1. 1Decide whether you need one object or several independently editable elements
2. 2Inspect the public source and sample layers
3. 3Open Prompt Cutout or Image to Layers and upload an authorized image
4. 4Describe the specific object or layer groups you need
5. 5Check transparency, details and the base before downloading the appropriate output

Frequently asked questions