Manual Concepts - ShadobaDev/Platemaker-qt GitHub Wiki
The vocabulary the rest of the manual assumes. Read this once and the other pages will make sense.
Platemaker's core model is the virtual strip: your ordered input images are treated as one tall image, scaled to a single target width, and then cut into fixed-height slices. Nothing is processed page by page.
โ Workspaces
โ Projects
โ Projects
โ Canvas profiles
โ Output profiles
โ Templates
โ Rendering
Workspace (.platemaker.json)
โโโ Canvas profiles โโโโโโ generate โโโบ Templates (<workspace>/templates/*.psd|*.png)
โโโ Output profiles
โโโ Projects (chapters)
โโโ input files (ordered) โโโ
โโโ assigned canvas profiles โ
โโโ selected output profile โโโโบ render โโโบ output slices
โโโ output directory โโโโโโโโ
| What | Where |
|---|---|
| Workspace | The .platemaker.json file you chose |
| Templates | <workspace folder>/templates/ |
| Workspace fonts |
<workspace folder>/fonts/ โ active while this workspace is open, see [[Workspaces |
| Rendered slices | The output directory set per project |
| Source images | Wherever they already are โ Platemaker never moves or modifies them |
See also: Quick start ยท File statuses ยท Main window