| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
| |
This reverts commit 308fc78775058d583449b6ffcdc227cf7f1280af, reversing
changes made to 632b33b6ecbc3e2a7c8277effcac9b2de0fff5a9.
|
| |\
| |
| | |
Fix multiple issues with Sixel
|
| | |
| |
| |
| |
| |
| | |
A buffer can be visible in more than one window. With one last-row/last-col and one placement ID per overlay, refreshing the second window overwrites the placement state for the first. After C-x 0, the closed pane's terminal image can no longer be addressed and remains as an artifact.
Record image placement geometry and placement IDs per live window. Refresh and cleanup now delete the placement for the specific window, while existing last-row/last-col properties mirror the window currently being refreshed for backend compatibility.
|
| | |
| |
| |
| |
| |
| | |
Window layout changes move image overlays before kitty-graphics recalculates terminal positions. Clearing last-row/last-col first loses the only geometry the backends can use to delete the old terminal placement, which leaves stale pixels over new split separators.
Delete the current recorded image placement during window-change handling before clearing the cached position. Also make image-mode placement reuse explicit: zoom commands still reuse a PID to avoid flicker, while split/resize rerenders delete first.
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Terminal doc-view rendering used to erase and insert padding/newlines into
the doc-view buffer before placing the page image. doc-view buffers visit
the original PDF, so changing buffer text marks the document modified and
can corrupt it if saved.
Create the Kitty/Sixel page overlay over the existing buffer contents
instead. Compute the page image dimensions, prepare/cache the page image,
and schedule a refresh without changing the visited PDF bytes.
|
| | |
| |
| |
| |
| |
| | |
When kitty-graphics renders doc-view pages in a terminal, doc-view still calls image-mode navigation helpers for scrolling and edge movement. Those helpers expect an Emacs image display and can error or do nothing in the terminal renderer.
Add terminal-only doc-view advice for the image navigation functions so they update window hscroll/vscroll and schedule a graphics refresh instead.
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Compute overlay screen positions in the window currently being refreshed
instead of choosing an arbitrary window that shows the same buffer. This
keeps placements correct when a buffer is visible in more than one pane.
Also derive the terminal column from the overlay's buffer column rather
than posn-col-row. In terminal Emacs, posn-col-row can describe the
right edge of display-property text instead of the logical overlay start,
which makes Sixel placements appear to the right of their reserved cells.
|
| | |
| |
| |
| |
| |
| |
| |
| | |
kitty-gfx--sixel-cleanup can be called without a file when cleanup is
routed through generic backend code. In that case the old implementation
passed nil to md5 while pruning cached Sixel temp files.
Make the cleanup a no-op when FILE is nil.
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
`kitty-gfx--kitty-detect' and `kitty-gfx--sixel-detect' read TERM,
TERM_PROGRAM, TMUX, KITTY_PID, and the WT_* markers via
`(getenv VAR)' -- the process-level lookup. Under
`emacs --daemon' / `emacsclient -t', the daemon process inherits
the launcher's env (often `TERM=dumb' from a systemd unit, no
TMUX, no TERM_PROGRAM); the *attached client* env lives on the
new frame's `environment' parameter. As a result, every detect
call sees daemon env and reports "terminal not supported" even
when the client is attached from a fully sixel-/kitty-capable
terminal inside tmux.
Note `(getenv VAR FRAME)' only consults the frame env (no
process fallback), and `(getenv VAR)' only consults the process
env first (then the *selected* frame's env). Neither alone gives
the desired prefer-frame-then-process semantics. Add a small
helper `kitty-gfx--frame-getenv' that does, and route both detect
functions plus `kitty-gfx--tmux-sixel-supported-p' through it.
For TERM specifically also consult the frame's `tty-type'
parameter, which is set by `server-create-tty-frame' and is the
canonical TERM emacs uses for this frame -- preferred over the
env value when the latter is missing or `dumb'.
Repro before this patch (Linux, ghostty + tmux, emacs daemon):
*kitty-gfx-debug*:
kitty-detect: nil (KITTY_PID=nil TERM_PROGRAM=tmux)
sixel-detect: nil (TERM=dumb TERM_PROGRAM=tmux TMUX=yes
tmux-ver=3.6 tmux-ok=yes WT=no)
mode: terminal not supported, aborting enable
After the patch the detectors read the frame env, so TERM
resolves to the actual `tmux-256color' / `xterm-kitty' / etc.
and the matching backend is picked.
Behaviour unchanged for non-daemon emacs: when the frame has no
`environment' parameter, `kitty-gfx--frame-getenv' falls back to
the process env directly.
|
| |/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
`kitty-gfx--sixel-detect' currently gates on the outer TERM matching
`xterm|vt[0-9]|foot|contour'. Once `tmux-ok' has confirmed tmux >=
3.4, tmux itself owns the Sixel rendering -- the DCS payload never
reaches the outer terminal. At that point the outer TERM regex is
irrelevant, but the canonical tmux assignment `tmux-256color' (and
the legacy `screen-256color') still trip the check, and detection
returns nil even when Sixel works fine.
Accept `tmux*' and `screen*' TERMs when `tmux-ok' is true. Outside
tmux the existing regex still rules, so no behavioural change for
non-tmux setups.
Repro before this patch:
TERM=tmux-256color emacs -nw -Q -l kitty-graphics.el \
--eval '(setq kitty-gfx-debug t)' \
--eval '(kitty-graphics-mode 1)'
=> "Kitty graphics: terminal not supported"
*kitty-gfx-debug*: sixel-detect: nil (TERM=tmux-256color ... tmux-ok=yes)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bumps `;; Version:' header from 0.4.0 to 0.5.0 and AGENTS.md `Current
version' to match (it was still 0.3.0; missed at the 0.4.0 bump).
Per AGENTS.md versioning rules, minor (0.x.0) is the right step for
new features, and 0.5.0 collects:
- Typst inline equation preview (#5)
- Configurable Sixel encoder + per-invocation timeout (refs #4)
- Sixel inside tmux >= 3.4 (refs #2)
- Windows Terminal detection + GIF routing (closes #6)
- Inline mpv video playback on the Kitty backend
- Loud warning when the Sixel backend cannot find an encoder on PATH
Plus assorted fixes (ECH erase + pre-erase, heading folded-line
skip, Sixel parity, text sizing column guard / CPU / preview-mode),
the new `justfile' task runner, and documentation that hard-states
img2sixel as the recommended Sixel encoder.
CHANGELOG.org gets a new top-level section with the full feature
write-ups, bench numbers for the encoder swap, credits to @timfel
(encoder timeout pattern + Windows Terminal patch + ongoing Sixel
collaboration), and a reworded "what comes next" paragraph that
reflects the Sixel-in-tmux side as done while leaving Kitty APC
passthrough on the roadmap.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Windows Terminal does not expose a stable `TERM' value, so the previous
sixel-detect check missed it. Accept the `WT_SESSION', `WT_PROFILE_ID'
and `WT_WINDOWID' environment markers it injects into every session.
Also add `gif' to `kitty-gfx--image-file-p' so GIF files are routed
through the image pipeline. Animation is still not supported -- only
the first frame is rendered -- but that is better than dropping them
on the floor.
AGENTS.md: list Windows Terminal under supported sixel terminals and
soften the GIF wording from "not supported" to "first frame only".
Closes #6.
Co-authored-by: Tim Felgentreff <timfel@users.noreply.github.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
when none is on PATH
Users hit a silent-failure mode where `kitty-gfx--sixel-detect' returned
true (terminal supports Sixel) but no encoder was on `PATH', so every
`sixel-place' logged `no encoder found' and produced no image. The
echo-area `message' from `sixel-encode' was easy to miss inside tmux,
which made the issue look like a refresh / tmux passthrough bug.
- README.org: Requirements section now explicitly lists a Sixel encoder
as required and recommends `img2sixel' (libsixel) over ImageMagick,
with the concrete numbers from PR #10's bench (~2.2x smaller payload,
~10x faster) and a note that this matters most inside tmux because
every refresh re-emits the full DCS payload. Known-Limitations
reworded: missing encoder = silent failure surfaced via the debug log.
- AGENTS.md: Dependencies block reorganised the same way; adds a
pointer to `just sixel-encoder' as the first step when diagnosing
Sixel rendering bugs.
- kitty-graphics.el: when the active backend is Sixel and
`kitty-gfx--sixel-resolve-encoder' returns nil at mode-enable time,
fire a `display-warning' (`*Warnings*' buffer + visible message)
pointing at img2sixel/ImageMagick, so the failure is no longer
buried in /tmp/kitty-gfx.log.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Sixel detection used to hard-disable the moment `$TMUX' was set. That
guard predates tmux 3.4 (released 2024-02-13), which ships native Sixel
rendering, so we were rejecting a working configuration. Drop the hard
guard, gate on a parsed `tmux -V' >= 3.4, and let users opt out via a
new custom var.
- `kitty-gfx-tmux-allow-sixel' (default t). Set to nil if your tmux
build lacks `--enable-sixel' or if you want to avoid the upstream
scroll artifact (tmux's cell buffer is not pixel-aware, so images may
persist after scrolling).
- `kitty-gfx--tmux-version' parses `tmux -V' once per session and
caches the result in `kitty-gfx--tmux-version-cache' (with an
`unset' sentinel so we can distinguish "not probed" from "no tmux").
- `kitty-gfx--tmux-sixel-supported-p' centralises the (in-tmux + opt-in
+ version >= 3.4) check.
- `kitty-gfx--sixel-detect' now uses the helper instead of
`(not in-tmux)'. Debug log gained `tmux-ver=' and `tmux-ok='.
- AGENTS.md / README.org / CHANGELOG.org: limitation reworded from
"no tmux support" to a partial entry that explains exactly what works
and what the remaining caveats are (Kitty APC passthrough is still
unimplemented; scroll artifacts are upstream).
- justfile: new `test-sixel-tmux' recipe wraps `test-sixel-image' in
`tmux new-session' for one-command interactive verification, and
the sibling `test-sixel-image' recipe was switched from a fragile
shell conditional to an elisp `when' so passing an encoder argument
no longer breaks the parse.
Refs #2.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Sixel encoding was hard-coded to ImageMagick (`magick'/`convert') with no
timeout, so a malformed image or hung encoder could freeze Emacs and
users could not switch to libsixel's faster `img2sixel'.
- New customs: `kitty-gfx-sixel-encoder-program' (nil = auto-detect:
prefer img2sixel, then magick, then convert), `-encoder-args' (extra
args list), `-encoder-timeout' (float seconds, default 5.0, nil =
no timeout).
- New helper `kitty-gfx--sixel-resolve-encoder' returns
(KIND . ABS-PATH) where KIND is `img2sixel' or `imagemagick' so the
encode pipeline can build the right argv (`-w/-h <px>' vs
`-geometry WxH sixel:-').
- New helper `kitty-gfx--sixel-run-encoder' runs the encoder via
`make-process' with a separate stderr buffer and a `run-at-time'
watchdog that kills the process on timeout; failures and timeouts
are logged and surface as nil from `kitty-gfx--sixel-encode'
rather than freezing Emacs.
- Existing magick/convert path preserved and still chosen automatically
when img2sixel is absent.
- AGENTS.md: dependency note updated to mention img2sixel and the new
customs.
Inspired by and co-authored with @timfel's sixel-graphics.el fork
(https://github.com/timfel/sixel-graphics.el), which introduced the
configurable encoder + timeout pattern.
Refs #4.
Co-authored-by: Tim Felgentreff <timfel@users.noreply.github.com>
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Mirrors the existing LaTeX fragment integration but uses the `typst`
CLI to compile `$...$' math fragments into PNGs that are then displayed
inline through the existing kitty-graphics overlay pipeline.
- New section in kitty-graphics.el: customs (typst-command, ppi,
text-size, preamble), SHA-1 cached PNGs under /tmp/kitty-gfx-typst,
Emacs face-foreground -> typst rgb() hex conversion.
- Two autoloaded commands: `kitty-gfx-typst-preview' and
`kitty-gfx-typst-clear-preview', region-aware.
- Regex skips backslash-escaped dollars (\\$not math\\$).
- justfile with recipes: compile/lint/load + per-feature test-* and
headless typst-render / typst-show.
- tests/test-typst.typ fixture with inline, display, mixed and
edge-case fragments.
- AGENTS.md updated: test instructions, dependency note, issue marked
done.
Closes #7. Parent #5.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Spawn mpv with --vo=kitty to paint video frames into a terminal
region that Emacs controls. Emacs reserves overlay space and
coordinates mpv's screen position via JSON IPC on scroll/resize.
New public API:
- kitty-gfx-play-video: play video file inline at point
- kitty-gfx-stop-video: stop playback and cleanup
- kitty-gfx-toggle-video: pause/resume
Requires: mpv with --vo=kitty support, Kitty terminal.
Gated behind kitty-gfx-enable-video defcustom (nil by default).
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Remove hardcoded column-1 guard that broke heading placement with
display-line-numbers-mode and org-indent-mode. Headings were never
placed (placed=0 hidden=37) because posn-at-point returns terminal
columns offset by body-left, not buffer column 1.
CPU optimizations reduce position queries by 86% and heading
placements by 76%:
- Early window-range exit skips posn-at-point for off-screen overlays
- Emitted flag prevents redundant OSC 66 re-emission
- Single fold-check instead of duplicate calls
- Strip text properties to prevent org-modern payload corruption
Add preview mode that temporarily disables conflicting minor modes
(org-modern, org-appear, org-indent, visual-line, olivetti) and
unfolds all headings for clean OSC 66 rendering. Modes are restored
when heading sizes are toggled off.
Add row-based collision detection to prevent overlapping multicell
blocks when adjacent headings occupy the same terminal rows.
Deferred display property shows heading text as fallback until first
successful OSC 66 emission, preventing invisible headings.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Close gaps identified in Sixel vs Kitty audit:
- image-mode: mode-name now shows Image[Sixel] or Image[Kitty]
based on active backend (was hardcoded to Image[Kitty])
- LaTeX preview: docstring updated to reference kitty-graphics
package name, not Kitty protocol specifically
- sixel-cleanup-all: now erases visible Sixel images from terminal
cells before cleaning disk/memory state (prevents pixel artifacts
on mode disable)
- sixel-place: LRU eviction for .six temp files, capped at
kitty-gfx-cache-size (default 64), oldest evicted on overflow
Remaining by-design differences (protocol limitations, not bugs):
- Sixel re-emits raw pixel data on scroll (no placement IDs)
- Sixel erases by overwriting with spaces (no protocol-level delete)
- Sixel detection is env-var based (no DA1 query yet)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When org folds headings (S-TAB overview mode), the newline after
each heading becomes invisible, causing multiple headings to appear
on the same visual line. posn-at-point then reports col > 1 for
headings that follow the fold ellipsis of a previous heading.
Placing scaled text at col > 1 causes horizontal drift — headings
rendered mid-line overlapping with fold ellipsis and each other.
Fix: org headings always start at buffer column 0. If the computed
terminal column is not 1, the heading is in a folded context where
rendering at scale would be wrong. Skip it (treat as not visible),
erase any previous placement, and let the refresh cycle re-place
it when unfolded.
Root cause identified via debug log analysis showing:
Before fold: all headings at col=1 (correct)
After S-TAB: headings at col=21/24 on same row (wrong)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Replace space-writing erase with ECH escape (\e[NX) which erases
characters at the cursor without moving it — more efficient and
cleaner terminal state.
Changes:
- kitty-gfx--erase-heading-at: new low-level function using ECH,
disables DECAWM during erase to prevent wrapping artifacts,
erases all rows of multicell blocks (not just topmost)
- kitty-gfx--erase-heading: now delegates to erase-heading-at
- kitty-gfx--place-heading: pre-erases target area before emitting
OSC 66, preventing ghost artifacts from partial overwrites
Adapted from mdfried's "erase-character dance" pattern which erases
the full multicell area using ECH before every OSC 66 emission.
|
| |
|
|
|
|
|
|
| |
- Version bump 0.3.0 → 0.4.0 (text sizing feature complete)
- kitty-gfx-debug-state: show text-sizing-support status in header,
distinguish heading vs image overlays, display heading-specific
properties (text, scale, level, cell-s) instead of image properties
(id, pid, file)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add heading-aware org-cycle handling:
- kitty-gfx--nuke-headings: erases ALL heading multicell blocks
in the current buffer, clearing their cached positions. This
is the "nuke" phase — a clean slate for the terminal.
- kitty-gfx--on-org-cycle: now handles both image and heading
overlays. Images get their placements deleted (existing logic),
headings get their multicell blocks erased (nuke phase).
The subsequent scheduled refresh is the "repaint" phase:
kitty-gfx--overlay-screen-pos checks in-folded-region-p, so
folded headings return nil and won't be re-emitted, while
visible headings get fresh OSC 66 emission at correct positions.
The nuke-and-repaint strategy avoids incremental erase complexity
(which failed in two prior prototypes) by simply erasing everything
and letting the refresh cycle figure out what's still visible.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix scroll/buffer-switch behavior for heading overlays:
- on-window-change: preserve heading overlay position caches
(images clear theirs for re-placement, but headings need
old→new comparison to erase multicell blocks properly)
- on-buffer-change: erase heading multicell blocks when their
buffer becomes hidden, preventing ghost artifacts if the new
buffer's content doesn't fully overwrite the heading cells
- on-buffer-change: preserve heading caches for visible buffers
(same rationale as on-window-change)
The existing two-phase refresh already handles scroll correctly:
phase 1 detects position changes and erases moved headings,
phase 2 re-emits all visible headings. These fixes ensure the
cache state is correct for that comparison to work.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add the OSC 66 emission pipeline (Step 5 of incremental plan):
- kitty-gfx--place-heading: emits \e]66;s=S:n=N:d=D;TEXT\a with
SGR bold + 24-bit foreground color at cached terminal position
- kitty-gfx--erase-heading: erases multicell blocks by writing
spaces at the topmost row (overwrite Rules 2-3)
- kitty-gfx--refresh-heading-overlay: phase 1 computes position,
erases on move/hide, caches new position (no emission)
- kitty-gfx--emit-heading-overlays: phase 2 walks all buffers,
emits OSC 66 for visible heading overlays after all posn-at-point
queries complete (prevents redraw-induced multicell destruction)
Two-phase refresh in kitty-gfx--refresh: phase 1 (walk-windows)
does position queries and image placements, phase 2 (after loop)
emits all heading OSC 66 sequences. Both phases run inside
synchronized output (BSU/ESU) to prevent flicker.
Also adds org-mode integration hooks (heading-modified debounce,
org-mode-hook for auto-apply, heading scan/remove commands) and
self-tests for place-heading and erase-heading in dry-run mode.
|
| |
|
|
|
|
|
|
|
|
| |
- Add kitty-gfx--query-text-sizing-support using CPR probing
(3 cursor position reports interleaved with OSC 66 tests)
- Detect scale, width-only, or no support
- Wire detection into kitty-graphics-mode enable (Kitty backend only)
- Reset detection cache on mode disable
- Update modeline lighter: KittyGfx[K+T] when text sizing available
- Update enable message to mention text sizing when detected
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
- Add kitty-gfx--dry-run variable and kitty-gfx--terminal-send wrapper
that routes all terminal escape output through a single function,
logging sequences in dry-run mode for batch debugging
- Replace 9 bare send-string-to-terminal call sites with wrapper
- Add kitty-gfx--decompose-scale for OSC 66 parameter computation
- Add kitty-gfx--validate-osc66 for protocol param validation
- Add kitty-gfx--heading-sgr for org-level-N face color extraction
- Add kitty-gfx-run-self-tests with 15+ assertions for batch testing
- Add kitty-gfx-heading-scales and kitty-gfx-heading-sizes-auto defcustoms
- Add kitty-gfx--text-sizing-support state variable
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
After org-toggle-inline-images creates overlays with multi-line display
properties, posn-at-point still sees pre-overlay layout because Emacs
hasn't recomputed pixel positions yet. This caused images to appear ~4
rows too low on initial render (scrolling would fix it).
Adding (redisplay t) at the start of kitty-gfx--refresh forces a full
layout recomputation so position queries return correct values immediately.
Also includes: simplified column calculation using window-body-edges,
stderr suppression in call-process, debug state inspection commands,
and ImageMagick binary preference fixes.
|
| |
|
|
|
| |
Sixel backend support is a new minor feature (new protocol, new
terminals supported, new defcustom). No breaking changes to public API.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Introduce backend dispatch architecture so the package supports multiple
terminal graphics protocols. Existing Kitty protocol code is extracted
into a `kitty` backend; a new `sixel` backend enables ~20 additional
terminals (foot, Konsole, xterm, mlterm, mintty, VS Code Terminal).
Key changes:
- Alist-based backend dispatch (`kitty-gfx--backends`, `kitty-gfx--backend-fn`)
- Kitty backend: detect/prepare/place/delete/cleanup extracted from inline code
- Sixel backend: ImageMagick-based encoding, temp-file cache, DCS emission
- Auto-detection (`kitty-gfx--detect-protocol`): tries Kitty first, then Sixel
- `kitty-gfx-preferred-protocol` defcustom for user override (auto/kitty/sixel)
- Mode-line indicator shows active backend: [K], [S], or [?]
- Fix obsolete `when-let` → `when-let*` (zero byte-compile warnings)
All integration points (org, image-mode, shr, dired, doc-view) dispatch
through the backend interface — no mode-specific changes needed.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
When zooming in image-mode or doc-view, reuse the existing placement ID
instead of deleting and re-placing. This lets the terminal atomically
replace the old placement (same PID, new dimensions) without a delete
step, working around WezTerm rendering artifacts (wezterm#5892).
Kitty handles this perfectly. WezTerm still shows minor artifacts on
zoom-in (larger replacement) but zoom-out is clean. This is a
WezTerm-side bug.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- LRU image cache with configurable size and eviction (kitty-gfx-cache-size)
- Cache stores only image IDs; dimensions computed fresh each time
- Leading-edge debounce for faster initial refresh response
- Targeted placement deletion on buffer/window change (no more delete-all flicker)
- Cell pixel size re-queried on window resize
- Image overflow check prevents visual corruption at window bottom
- convert-to-png returns nil on failure instead of passing non-PNG data
- unwind-protect for temp file cleanup in display-image paths
- kill-buffer-hook checks cross-buffer image sharing before deleting
- ignore-errors around all send-string-to-terminal calls
- Millisecond-precision debug logging throughout
- Dead overlay pruning in refresh cycle
- GIF removed from supported formats (multi-frame issues)
- Window-size-change hook in image-mode for split/unsplit re-render
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
- Use kitty-gfx-image-mode major-mode to avoid evil-collection overriding
zoom keybindings with native image functions that fail in terminal
- Add evil normal-state bindings for +/=/-/0/q when evil is loaded
- Use special-mode-map as parent instead of image-mode-map to prevent
image-mode-window-put and image--change-size errors
- Fix centered image column calculation: detect when posn-col-row is
inflated by overlay display properties and fall back to body-left
- Clear buffer-modified-p after image-mode renders
- Add debug logging to centered display and image-mode-render paths
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Images in image-mode now display centered in the window with
+/-/0 keybindings for zoom in/out/reset.
- Add kitty-gfx--image-scale buffer-local variable
- Add kitty-gfx--image-mode-render for centered display at scale
- Add kitty-gfx-image-increase-size (1.25x per press)
- Add kitty-gfx-image-decrease-size (0.8x per press)
- Add kitty-gfx-image-reset-size (back to 1.0)
- Bind +/-/0 in the image-mode keymap
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
PDFs now render via Kitty graphics in terminal doc-view-mode.
Pages display centered in the window with n/p navigation and
+/-/0 zoom controls.
- Advise doc-view-mode-p to bypass display-graphic-p guard
while preserving per-type tool availability checks
- Advise doc-view-insert-image to render pages via Kitty
using the centered display helper with zoom scale
- Advise doc-view-enlarge and doc-view-scale-reset to update
the buffer-local scale factor and re-render
|
| |
|
|
|
|
|
|
| |
- Add kitty-gfx--display-image-centered: centers images in the
window with optional zoom scale parameter for full-window modes
- Bump kitty-gfx-max-width 80->120, kitty-gfx-max-height 24->40
- Relax refresh visibility check to only require overlay start
on screen (allows images larger than window for zoom-in)
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
C-c C-x C-l now renders LaTeX fragments as images in terminal Emacs.
Bypasses org's display-graphic-p guard since the latex+dvipng pipeline
works fine without a GUI.
- Advise org-latex-preview to handle terminal color fallbacks
(unspecified-fg -> Black) and suppress clear-image-cache
- Advise org--make-preview-overlay to display generated PNGs via
kitty-gfx-display-image instead of Emacs image specs
- Add kitty-gfx--org-clear-latex-preview for C-u cleanup
- Add LaTeX test fragments to test-kitty-gfx.org
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Add fold-aware visibility detection so images disappear when their
containing heading is folded and reappear when expanded.
- Add kitty-gfx--in-folded-region-p: checks org-fold (9.6+) then
falls back to overlay-based invisibility, ignoring cosmetic
org-link bracket hiding
- Update kitty-gfx--overlay-screen-pos to skip folded overlays
- Add kitty-gfx--on-org-cycle hook: deletes terminal placements
before clearing cache, then schedules refresh
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
- Query terminal cell size via CSI 16t (XTWINOPS) instead of hardcoded
8x16 fallback, giving accurate image aspect ratios
- Wrap all placement output in synchronized output (DEC mode 2026) to
eliminate flicker during scroll and redisplay
- Add native dirvish preview dispatcher (kitty-image) that hooks into
dirvish-preview-dispatchers for automatic image preview
- Improve shr/eww advice to detect actual image type from spec
- Extract kitty-gfx--refresh-overlay for cleaner refresh cycle
- Reset position cache on buffer/window change so images re-place after d=a
- Bump version to 0.2.0, update package metadata and README
|
| |
|
|
|
|
| |
- kitty-gfx-dired-preview: preview images in a side window from dired
- image-mode: add q to quit, set buffer-read-only
- Add dired-get-file-for-visit forward declaration
|
|
|
Display images in terminal Emacs via Kitty graphics protocol
with direct placements. Integrates with org-mode (including
org-link-preview for org 10.0+), image-mode, and shr/eww.
|
