FAQ

Slow or unresponsive

Work out which kind of slow

Slow to start: the first launch loads AI models. Tens of seconds is normal

One feature is slow: AI features (portrait, colourise, repair, cut-out) are slow by nature —

a few seconds to tens of seconds per image on CPU

Everything is slow: check whether the canvas is oversized (beyond 6000 px) or has too many layers

Things to try

Close panels you are not using

Downscale very large images before processing

Merge layers you do not need right now

This app runs AI on the CPU with no GPU acceleration. Being slower than a cloud service is the trade-off —

and the payoff is that your photos are never uploaded anywhere

AI features do nothing

Step one: check whether the models are present

Open Help → System info, which lists the bundled AI models.

Any model shown as missing means that feature cannot run.

Common causes

Model files were not copied across (easy to miss with a manual portable copy)

Face features find no face: the face is too small, in profile, or covered

Not enough memory: this app runs on CPU, which needs RAM

What to do

Missing models: extract or install the package again, completely

No face detected: try a frontal photo with clearly visible features

The system info page has a copy button — handy when asking someone else for help

The brush paints nothing

Check the layer: is a hidden or locked layer selected?

Check opacity and flow: maybe they were dragged to 0

Check the selection: with a selection active the brush only paints inside it

Check rulers: an active ruler constrains strokes, which can look like the brush misbehaving

Check the mask: black painted on a layer mask hides what you draw

Work through it in order: layer → selection → ruler → parameters. Nine times out of ten it is one of these

Fill leaves a white rim or does not fill

White rim: tolerance too low. Use 30~50 on line art

Will not fill: the line art has a gap and colour leaks out. Close the gap with a brush first

Leaks outside the lines: tolerance too high. Turn it down

Anti-alias: always tick it when colouring line art

To check for gaps: zoom the line art to 400% and follow the outline. A one-pixel gap is enough to break the fill

Export looks different from the screen

Colours look washed out: the export is sRGB and your viewer does no colour management. Compare in another viewer

Wrong size: check the scaling setting at export time

Transparency turned black: it was saved as JPG. JPG has no alpha — use PNG for transparency

Blurry: export quality was too low, or the dimensions were reduced

The most common case is actually "transparency turned black": pick PNG or WEBP, and hide the background layer before exporting

The app crashed — is my work safe?

The app autosaves and offers to restore after a crash

Autosave files live in %LOCALAPPDATA%\WenCangImageEditor\Autosave

But this feature has not been thoroughly tested. Do not rely on it

What actually keeps you safe

Press Ctrl+S to save .wci as you go. Make it a habit

Keep two copies of important work: one .wci and one export

Autosave is a safety net, not a vault. Saving manually at key points is still the reliable way

Interface looks small on a high-DPI screen

The main interface follows system scaling, with crisp text and icons

A few standalone windows (the colour picker, for instance) keep a compact size. They look small but text is sharp and clicks land accurately

This is deliberate: the target machine is 1080p, where scaling the whole window to 150% eats most of the screen

If the small size makes things hard to read, try system scaling at 125%,

or tell us and we will consider an interface size setting

Shortcuts do not work

Check whether a selection or a panel has taken focus

A Chinese IME state swallows single-key shortcuts. Switch to English and retry

Shortcuts can be edited in Edit → Settings → Shortcuts, and this help updates to match

The shortcut table in help is generated live from your current settings, not written by hand.

So it can never disagree with what the app actually does

Suggestions and bug reports

Include this and you save three round trips

Help → System info, click copy, and paste the contents

Steps to reproduce: what you clicked, and at which step it broke

Screenshots if possible, ideally including the original image

For crashes, say what the last action before the crash was

How to describe a problem well

Do not say "it does not work". Say "I clicked X, expected Y, got Z".

"Auto beautify does nothing" and "auto beautify popped up saying a model is missing" are completely different problems.

The second one we can locate within ten minutes

What this version cannot do

Stating the limits so you do not waste time:

Group-photo beautify: per-person tuning works — use the "Next face" button at the bottom of the panel,

one parameter set per face (see the portrait chapter). But there is no "beautify everyone at once" button;

a crowded group photo means going person by person

Profiles and heavily occluded faces: landmarks cannot all be found, so results degrade

Importing video files: mp4 and other video files cannot be opened as footage. You can still click Video at the top to open the timeline, make frame-by-frame animation and export MP4, GIF or PNG sequences (see Feature 16)

Very large images: anything in the tens of thousands of pixels struggles. Downsize first

Vector output: exports are raster (PNG/JPG), not SVG

These are not bugs, they are the current boundaries. If you need them, tell us and we will prioritise