SCVeamer version 1.4 prototype - 2026-10-24
=======================================

Open bundled/day1/index.html to view the complete offline presentation.
The bundled starter and template are in starter/ and bundled/template-v2/.
The supplied samples use lossless PNG pages at a width of 2560 pixels.
The English PDF, TeX, game and player localization is preserved.
This is a source prototype; no Windows executable is included or validated.

Requirements: Python 3.11+ (tested on Linux/Python 3.12.14), with Tkinter for GUI.
Run SETUP_WINDOWS.cmd, or: py scveamer.py --setup
It installs the three pinned requirements into this kit's own .venv.
The initial download needs internet access; viewing the generated HTML does not.

After updating talk.pdf in your project folder, run:
py scveamer.py --project "C:/work/my-talk" --pdf-only --image-width 2560

Alternatively select your project in the GUI and build from the existing PDF.
CLI image width overrides presentation.json; absent either, the default is 1920.
Bundled project settings explicitly select 2560.
PDF-only conversion does not require TeX. The TeX compile path is retained,
but individual full-workflow and GUI tests are not documented in the reported Windows 11 use. Executable builds remain unverified; see VALIDATION_PUBLICATION_1_4.txt.

Read VALIDATION_1_4.txt for the versioned-kit checks, CHANGES_1_4.txt for the changes,
and VALIDATION_PDFIUM.txt / validation/ for the preceding backend evaluation.
VALIDATION.txt and VALIDATION_1_2.txt describe historical versions only.
All sample structures/titles and handout selected pages match their originals.
All 17 tests pass without PyMuPDF. Day1 retains 46 slides / 76 states and 14:55-17:10.
Offline Chromium checks verified games, controls and state retention.

Limits: fixed-resolution PNGs soften at high zoom; visited-deck browser PSS was
about 1 GiB rather than the original SVG's 670 MiB. A bounded image-cache experiment
did not give a convincing improvement and was not adopted. Mathematical
alternative-text extraction remains imperfect; displayed formula images are
independent of those Unicode mappings. Arbitrary rotated/cropped PDFs, live
projectors, Mac use and executable packaging still need separate checks. Reported Windows 11 use is described in VALIDATION_PUBLICATION_1_4.txt.

Dependencies: pypdf 6.10.0, pypdfium2 5.14.0 (PDFium 156.0.8076.0), Pillow 12.3.0.
See THIRD_PARTY_NOTICES.txt and licenses/. No new MIT grant or project-code
rights-holder declaration is added. Existing contributions and credits remain.

Version identity
----------------
The application identifies as 1.4-prototype-20261024; HTML product metadata is 1.4.
Settings and fallback workspace use version-1.4-prototype-en. The HTML format tag
is unchanged. Source-page text and historical release records retain their original version labels.

Read CREDITS.html / CREDITS.txt for precautions, development and review scope, publication plans and the reason for the backend change. VALIDATION_PUBLICATION_1_4.txt records the publication-documentation revision and the reported Windows 11 use.

On 24 October 2026, Takayuki Koike reported trying the 1.4 prototype on his Windows 11 computer without noticing particular problems in the operations tried. The individual operations, edition used, and environment details were not recorded comprehensively; this is not a claim of exhaustive testing or separate Windows validation of both editions.
A systematic Windows test covering fresh setup, individual GUI functions and the full TeX workflow, Mac checks, and executable build/run validation has not been recorded.
