PSD Viewer

A PSD viewer should identify the picture it is showing

A Photoshop document can contain hundreds of layers, masks, effects, text runs, smart objects, adjustment instructions, and color settings. It can also carry an embedded composite: a ready-made raster view stored by the authoring application for software that cannot reproduce the full editing model. This viewer uses that embedded composite and says so on the receipt. It does not quietly substitute a partial layer rendering and label the approximation as the document.

Before a preview appears, the file is identified from its bytes as PSD version 1 or PSB version 2. The inspection reports canvas dimensions, color mode, sample depth, channels, composite encoding, transparency, ICC/XMP presence, and a layer inventory. The layer names and counts are useful facts about the source, but they are not controls: layers cannot be hidden, rearranged, edited, or exported separately here.

The supported PSD and PSB boundary is intentionally narrow

V1 accepts verified RGB, eight-bit PSD and PSB documents whose embedded composite uses Raw or RLE encoding. That is a product boundary rather than a claim that these are the only files the underlying parser could ever recognize. A document in CMYK, Lab, Indexed, Bitmap, or Grayscale mode needs color semantics this route has not proved. Sixteen- and thirty-two-bit samples need a different precision contract. ZIP and ZIP-prediction composites need their own independently checked decode path.

Those cases receive a typed unsupported message before a plausible-looking export can be mistaken for a faithful one. The same applies when the file has no usable embedded composite. A rich layer tree is not a fallback image. Reconstructing it would require independently correct blending, clipping, masks, effects, transforms, text layout, fonts, smart objects, and adjustment layers; this viewer does not claim that editor.

PNG, JPG, and PDF are flattened delivery copies

PNG keeps the full canvas and the verified alpha values from the decoded composite. It is the appropriate choice for interface artwork, overlays, logos, and any view whose transparent edge must remain transparent. The PNG is lossless from those decoded pixels forward, but it is still not an editable PSD and does not carry the source layer model or Photoshop metadata.

JPG creates a new lossy image. Its quality setting is a real encoding choice, and when the composite contains transparency the tool requires a solid background instead of silently turning transparent pixels black or white. PDF creates one 96-dpi page containing one flattened raster view, using 0.75 PDF points per source pixel. Transparency is flattened on white only when it exists. The PDF is useful as a review or handoff copy, not as PDF/A, a layered Photoshop document, or a way to make text and vector shapes editable outside Photoshop.

No resize, crop, or color-managed publishing claim

Every export uses the source canvas dimensions. There is no resize or crop control in this route, because changing geometry would mix viewing the document with editing it and make the receipt less clear. If a smaller delivery image is needed, export first and use the image compressor on the reviewed result. If the visible colors need to become a reusable set, the image color palette tool works from that flattened image.

ICC and XMP presence is reported, but ICC is neither applied nor preserved; the export is untagged RGB, not a color-managed production master. Keep the source document when spot colors, print separations, wide-gamut values, high sample depth, or authoring metadata matters. A screen-ready composite can be useful and visually correct within its declared boundary without being an archival replacement for the source.

The receipt names what survived and what did not

The result identifies the source as PSD or PSB, the embedded composite as the render source, the canvas and output dimensions, color mode and depth, channel and alpha handling, layer inventory, profile and metadata boundary, chosen output, and byte count. Its exact receipt rows are Source format, Render source, Output format, Dimensions, Alpha, Background, Quality, PDF pages, Layers, Color profile, Metadata, File size, and Reopened. The generated file is reopened before its download is offered. Format, dimensions, alpha or background, and representative pixels must agree with the planned export; a filename extension alone is not evidence. Chromium and WebKit may compress an otherwise identical PNG differently, so the promise is deterministic decoded pixels and content, not identical container bytes across browser engines.

Need another delivery format? The inspected source remains available, so PNG, JPG, and PDF can be created without choosing the PSD again. That is why these formats are actions in one viewer rather than three thin pages that repeat the same intake, inspection, preview, and safety boundary. Going the other way is not a switch: a PNG or PDF cannot recreate editable layers that were never present in the flattened file.

The file stays in this browser tab

PSD and PSB are binary documents, so the honest intake is a local file picker or drop. A copied screen image is already a flattened bitmap and is not the same input. The signature and structure determine the format rather than the extension. Large documents receive local memory advice before processing and are not rejected by an arbitrary upload ceiling.

There is no upload or server conversion fallback. The source bytes, layer names, preview, and exports remain in the tab and are not placed in the URL, browser storage, or analytics. The local-processing guide explains how to inspect that network boundary. For ordinary JPEG, PNG, or WebP metadata rather than a layered document preview, use the image metadata viewer. For a PDF-compatible Illustrator document rather than a Photoshop composite, use the AI file viewer; it verifies that narrower representation before preview or export.