Document input content ratio and transform semantics

This commit is contained in:
Dandaedre 2026-11-16 10:08:28 +01:00
parent 043fd88c61
commit c72835b1f2
2 changed files with 35 additions and 0 deletions

View File

@ -4,6 +4,10 @@
## Unreleased
- Document the manual-verified distinction between input Content ratio and
Transform to, their existing advanced device controls, and the limits of local
preview modeling. No geometry, device writes or save format changes.
- Add independent layer masking: magenta handles, whole-window drag, arrows,
numeric pixel edges, alignment, center/reset and Undo/Escape, without resizing content.
Midra imports and reviewed sends preserve masks; Aquilon rejects unsupported

View File

@ -216,3 +216,34 @@ reveal lower layers. A mask over already cropped or transparent pixels has no
visible effect. Resizing the layer keeps mask proportions. Project files and
embedded snapshots retain masks; plain geometry exports omit them. Midra load/send
supports masking; Aquilon masking remains unavailable.
## Content ratio versus Transform to
The [Midra 4K v3.1 manual, §10.2.6.1, page 65](https://impact-even.com/userfiles/files/telechargement/video/analogway_pls4k_manuel.pdf#page=65)
distinguishes two input corrections:
- **Content ratio** interprets the source's intended proportions, including
non-square pixels. Native is the normal choice. Its fixed presets are 5:4,
4:3, 16:10, 15:9, 16:9, 21:9, 256:135 and 64:27; this selector has no Custom.
- **Transform to** sets the final ratio presented to layers. Signal Aspect Ratio
keeps the signal ratio; alternatives include fixed ratios and Custom.
- **Layer fill option** is separate: it controls fitting when a layer inherits
input settings. Canvas Pilot uses explicit layer fitting instead.
These are input-level corrections shared by every layer using the input, not
layer cropping or masking. They can correct stretched content, but an arbitrary
ratio can also introduce distortion. Leave Native / Signal Aspect Ratio unless
correction is intentional.
Canvas Pilot already exposes both device selectors under **Source → Edit
advanced input settings**, with Custom ratio encoded in thousandths (0.250–10.000).
They are staged device controls, not persistent local image deformation. Merely
opening the controls never writes. A local source resolution is a separate
quantity and should not be changed just to pretend a ratio correction occurred.
The manual establishes the purpose and scope, but does not give the exact
corrected pixel dimensions, rounding or interaction with layer 1:1. A local
preview model must distinguish sampled input pixels from corrected display
proportions and avoid applying a correction twice when loading device-reported
content dimensions. These details need evidence before claiming an exact device
round trip.