Distort: keep blend steps, saved corners, previews and deep masks right

- A blend step whose interpolated corners and nearer key's corners both
  fail the check is placed as the nearer key is.
- Corners an image can't take are dropped when a file is read, so a
  later transform does not make them its distortion.
- The .ai editing data writes a distorted image at any nesting depth, as
  it does a gradient mesh, so an opacity mask at the limit keeps it.
- A linked image's preview keeps the clipping path its file keeps, so a
  document reopened from its preview still knows the file's clip group
  and Update Links replaces it instead of nesting a new one.
- Docs: `corners` in the plugin API covers only corners that are not a
  parallelogram; Relink's warning for distorted images needs `preserve`.
This commit is contained in:
dotCooCoo committed 2026-11-18 17:44:10 -08:00
1 parent d6bc984cad
commit fb0a82b095
12 files changed
+208 -25

No files matched your search

+4 -3
View File
@@ -678,9 +678,10 @@ every attribute:
of JSON: type whose runs and paragraphs don't fit what the type before it in paint order left comes with
`truncated: true` too.
- An image (`"image"`): `pixelSize` is its width and height in pixels; `colorMode` is `"rgb"`, `"cmyk"`, `"gray"` or
`"unknown"`, read from the image itself. An image changed by Perspective Distort or Free Distort also has `corners`,
its pixel rectangle's corners on the page (top left, top right, bottom right, bottom left), and its `transform` is the
affine placement that matches it at its middle. An embedded image whose pixels take at most 1 MiB in base64 (443 × 443
`"unknown"`, read from the image itself. An image whose corners Perspective Distort or Free Distort moved to a shape
other than a parallelogram also has `corners`, its pixel rectangle's corners on the page (top left, top right, bottom
right, bottom left), and its `transform` is the affine placement that matches it at its middle. Corners that make a
parallelogram are described by `transform` alone. An embedded image whose pixels take at most 1 MiB in base64 (443 × 443
pixels fit) has them in `pixelsBase64`: straight RGBA, 8 bits a channel, row by row from the top, and
`pixelsOmitted: false`. A CMYK image's pixels are its inks through the working CMYK profile, as sRGB. The pixels of
all the images in one input take at most 16 MiB in base64, and an image the host starts to decode takes its size