Past Perfect Support

Support information for the Past Perfect Mac app, including system requirements, key workflows and feature notes.

System requirements

A Mac running macOS 14 or later.
Local storage space for your original images, edits and exported files.
An internet connection is only needed for downloading or updating the app. Editing and restoration run entirely on your device.

Overview

Past Perfect is a photo editor for scans, family photos and everyday pictures. Editing runs entirely on the device, so your photos are not uploaded.

Past Perfect works non-destructively: your edits are saved automatically while the original image is preserved. The finished image is only created when you export. If Save copies is turned off, the original is overwritten.

Getting started

Add one or more folders to the library using the sidebar. You can expand subfolders and then select a photo from the strip below the preview.

Right-click a folder in the library to remove it from the library or start batch processing. Files on disk are not deleted when a folder is removed from the library.

Useful keyboard shortcuts: ← / → for the previous or next photo, Cmd-E to export and Cmd-+ / Cmd-- to zoom.

Editing tools

You will find the editing tools in the inspector on the right. They are grouped into tabs by task:

Light adjusts brightness and tonal values, including exposure, contrast, highlights, shadows, black point and white point. It also includes midtones, noise, clarity and sharpness. Automatically improve analyzes the image and suggests suitable slider values.
Color provides temperature, color cast, saturation and vibrance, plus a Color Mixer for eight color ranges. For each range, you can adjust hue, saturation and brightness independently. You can also import LUT files and use them as a color look.
Looks contains ready-made image looks, including black-and-white, color and dithering looks. Looks can be mixed with the original image. The farther the slider is moved to the right, the stronger the look becomes. You will also find vignetting and frames here.
Restore contains optional tools for cleaning up and restoring old or damaged images: dust and scratches, denoise, face restoration and upscaling. The individual steps can be enabled independently.
Remove background removes the background using Apple Vision directly on the device. If transparency should be preserved, save or export the result as PNG.
Transform lets you rotate and mirror images, straighten the horizon and crop the image to the desired aspect ratio.

Dust, scratches and denoise

Under Restore, you can optionally clean up an image before face restoration. These functions are disabled by default and are only applied when you enable them:

Dust and scratches detects and reduces typical small defects such as dust spots and scratches. A higher strength removes more, but can also smooth out fine image details.
Reduce noise is a stronger denoise operation applied to the entire image, particularly useful for noisy or scanned images.

If several Restore steps are enabled, they are processed in a fixed order: first dust and scratches, then denoise, followed by face restoration and finally optional upscaling.

The noise reduction under Light is a gentle adjustment that changes the image directly. Full-image denoise under Restore is a separate processing step and can also be included in a profile.

Restore faces

Three GFPGAN versions are available for face restoration. No version is best for every photo, so if the result does not look natural, it is worth comparing the models:

GFPGAN 1.2 usually looks stronger and sharper, but can also appear more artificial.
GFPGAN 1.3 is more restrained and often more natural. This is the default model.
GFPGAN 1.4 often provides slightly more detail than 1.3 and can be useful for heavily damaged or old photos.

GFPGAN detects faces and then automatically enhances them. It can convincingly reconstruct lost details and typical damage in old or small portraits. Very small faces contain little actual image information, however, while very large faces can make AI changes more noticeable. Always compare the result with the original.

Practical tip: With very high-resolution portraits, face restoration can produce an overly smooth or artificial result. In such cases, another GFPGAN model may work better. Version 1.3 is particularly suitable when the correction should be as subtle as possible.

Face restoration can be enabled or disabled independently of upscaling. For 2x or 4x upscaling, you can use Lanczos and, if the model is available, Real-ESRGAN. At 1x, no additional upscaling is performed.

Profiles

Profiles save a combination of editing steps and export settings. This lets you reuse a proven workflow with just a few clicks. Manage profiles via Manage profiles or the gear icon next to the profile buttons.

Each profile can define:

Processing steps, for example cleanup, denoise, face restoration, automatic corrections, crop, rotation, fixed export size, look and vignette.
File rules, where Save copies defines the destination folder, naming scheme and output format. If the option is turned off, export overwrites the original.

When you select a profile, its settings are applied to the current photo. With No Profile, you can return to manual editing.

Batch processing

To process several images at once, right-click a folder in the library, choose Batch Processing and then select a profile. Past Perfect processes the readable images in the folder one by one and saves the results according to the profile's rules.

A progress window shows which image is currently being processed. Batch processing can be cancelled between two images.

For batch processing, the destination must be defined in advance. Use a fixed export folder or Next to original in the profile. Ask every time cannot be used for batch processing.

The output format is also defined by the profile, for example same as source, HEIC, JPEG or PNG.

Export and files

Edits are saved automatically. There is no separate Save command; when you move to the next photo, your editing state is preserved.

Press Cmd-E to export the current image. The active profile's settings are used. With Save copies, a new file is created; without it, the original is overwritten. If transparency should be preserved, PNG is the appropriate format.

Camera RAW files such as ARW, NEF, DNG or CR2/CR3 can be opened and edited. They are not overwritten; for export, use a profile with Save copies.

Reset All removes the saved editing steps and restores the original editing state. Once the original has been overwritten, the previous version cannot be restored.

Glossary

Automatically improve: Analyzes the image and sets suitable values for the main Light controls. It is not a fixed look, and the values can still be adjusted afterwards.
Floyd-Steinberg: A dithering method that builds an image from black and white dots. The dot size affects the visible grain.
GFPGAN: An AI-based face restoration method that runs entirely on the device. Past Perfect includes versions 1.2, 1.3 and 1.4. If the result looks too artificial, compare the models.
Real-ESRGAN: An AI-based upscaling method. It is available at 2x and 4x if the corresponding model is present; alternatively, Lanczos can be used.
Dust and scratches: Optional automatic cleanup of typical small defects such as dust and scratches. The strength determines how strongly the image is processed.
Denoise (full image): Optional denoise for the entire image. It is stronger and more fundamental than the regular noise slider under Light and can also be included in a profile.
Fit vs. Fill: Fit shows the entire image within the target format, which can leave empty borders. Fill and Crop fills the entire target format and therefore crops part of the image.
Vignetting: A gentle darkening of the image edges that draws the eye more toward the center.
Non-destructive: Edits are saved separately while the original pixels remain unchanged. The changes are only written to a finished image file on export, or directly to the original if Save copies is turned off.