mirror of
https://github.com/storytold/vectorcraft.git
synced 2026-10-25 18:27:02 -06:00
M4.49: PNG-8 and GIF export with a palette quantizer and dithering
document.export {format: "png8" | "gif"} reduces the art to a palette of at most `colors`
(2–256) entries: median cut over a 15-bit histogram refined by k-means (perceptual,
selective or adaptive weighting), or the web-safe, black-and-white or grey palettes; art
with that many colours or fewer keeps them exactly. Dithering is Floyd–Steinberg diffusion,
an 8×8 ordered pattern or noise, with an amount. Pixels under half opacity become one
transparent entry; partly transparent edges blend over the matte. PNG-8 is an indexed PNG
(PLTE, tRNS, 1–8 bits per pixel, Adam7); GIF is a GIF89a (optionally interlaced). Both are in
document.formats, Export As (PNG-8 / GIF Options dialogs) and Export for Screens.
document.formats' writable list and the MCP export tool's format enum gain gif and png8
(their tests list them).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
co-authored by
Claude Opus 5.5
parent
ba4ed2ff93
commit
ac77b52ad3
@@ -27,6 +27,7 @@
|
||||
| `engine.execute` `file.place` | `{paths?}` / `{path \| name+dataBase64, …file.place params}` | with no file, picks files and opens the `place` dialog; with `paths`, opens it for them (fields `files`, `link`, `template`, `replace`; `__replace` says whether Replace applies, `__info` describes each file). `ui.dialog.confirm` places one file centred in the view (or in place of the selected object) and loads the place cursor with several (`file.place.queue`). With a file, it runs `file.place` centred in the view. Files dropped on the canvas are placed at the drop point (linked; Shift embeds), elsewhere or with no document open they open. With the cursor loaded (`ui.inspect` → `tool: "place"`, `toolOptions: {count, current, name, names}`), `ui.pointer` clicks and drags place files and `ui.key` ←/→/↑/↓ and Escape cycle and discard them |
|
||||
| `app.export` raster options | `{path?, format: png \| jpg \| webp, ppi?, background?, antiAlias?, interlaced?, quality?}` | `ppi` sets the pixel size (72 = one pixel per point) and is stored in the file; `background` is `transparent`, `white`, `black` or `"#rrggbb"`; `antiAlias` is `none`, `art` or `type` (text snapped to pixels); `interlaced` writes an Adam7 PNG. The PNG Options dialog (`pngOptions`; `jpgOptions` / `webpOptions` for the other raster formats) has the same fields: fill them with `ui.dialog.set` and press OK with `ui.dialog.confirm` |
|
||||
| `app.export` JPEG options | `{path?, format: jpg, quality?: 0–100, colorModel?: rgb \| cmyk \| gray, method?: baseline \| optimized \| progressive, scans?: 3–5, embedIcc?, imageMap?: none \| client \| server}` | `cmyk` writes ink amounts in the working CMYK space (CMYK colours keep their inks; RGB ones are separated with the colour settings); `embedIcc` (default true) embeds sRGB, the working CMYK profile or a grey profile, generated by the CMS; `imageMap` also writes `<stem>.html` (client-side `<map>`) or `<stem>.map` (NCSA) beside the image for the objects with a URL and an Image Map shape (`attributes.set`), returned in `linked`. The `jpgOptions` dialog has these fields too (`quality` is shown 0–10; a CMYK document defaults to `cmyk`) |
|
||||
| `app.export` PNG-8 and GIF | `{path?, format: png8 \| gif, colors?: 2–256, reduction?: perceptual \| selective \| adaptive \| web \| blackWhite \| gray, dither?: none \| diffusion \| pattern \| noise, ditherAmount?: 0–100, transparency?, matte?, interlaced?, …raster options}` | a palette of at most `colors` entries (art with that many colours or fewer keeps them exactly); pixels under half opacity become one transparent entry unless `transparency` is false, partly transparent edges are blended over `matte` (`none` keeps their colour). `png8` writes an indexed `.png` (1–8 bits per pixel); Export for Screens rows take both. The `png8Options` / `gifOptions` dialogs have these fields |
|
||||
|
||||
| `ui.dialog.*` on `importPdf` | `{field: "page" \| "allPages" \| "range" \| "cropTo" \| "password", value}` | opening (`app.open`, `file.open`) a PDF or `.ai` with several pages or a password, or placing one (`file.place` without `page`), opens the Import PDF dialog (`mode`: `open` or `place`; `pageCount`). While `__locked` is true, set `password` and `ui.dialog.confirm` to unlock it (a one-page file then opens or is placed). Then set `allPages: false` and `range` ("2-3, 5") to open some pages, or `page` to place one, and `cropTo` (`bounding`, `art`, `crop`, `trim`, `bleed`, `media`); `ui.dialog.confirm` opens or places them (placing keeps the other `file.place` params, such as `link`, `template` and `at`, in `__place`). `document.open {pages, cropTo, password}` and `file.place {page, crop, password}` do the same without the dialog |
|
||||
|
||||
|
||||
Reference in New Issue
Block a user