OPINIONATED OMARCHY
TROUBLESHOOTING CORPUS
790
RECORDS
790
AUDITED
0
UNCHECKED
2388
SOURCES
LIVE
display-monitors low common DISPUTE

144 Hz screen animates like 60 Hz once a 60 Hz monitor is plugged in

hyprlandwaylandomarchyomarchy-4archendeavouroscachyosmanjarodesktoplaptopamdnvidiaintel
CORRECTED An auditor found the fix wrong or incomplete and rewrote it.
AUDIT NOTE

The upstream research is excellent and checks out exactly: PR #12418 is real, merged 2025-11-23, and the quoted body ("I removed the logic that referenced m_mostHzMonitor for animation scheduling and switched to using each monitor's own refresh rate.") is verbatim. debug:vfr defaults to true with the wiki's "Heavily recommended to leave enabled to conserve resources", misc:vrr has exactly the documented 0/1/2/3 modes, cursor:no_hardware_cursors is 0/1/2 as described, render:new_render_scheduling is the triple-buffering option, and per-monitor vrr is a real hl.monitor field. Discussion #10969 genuinely confirms no_hardware_cursors = 1 as the working workaround for rotated/multi-monitor cursor artefacts, with multiple users confirming and one posting it in hl.config form. Omarchy's ALPM guard is real (default/libalpm/hooks/00-omarchy-update-guard.hook + bin/omarchy-update-pacman-guard, which aborts when it sees sync and sysupgrade together), so omarchy update is right. Two defects. (1) Four of the six fix steps use hl.set("category:option", value), which is not a real API. The Variables page documents only hl.config({ category = { option = ... } }), and hl.set appears nowhere in the wiki or in Omarchy's quattro tree. Steps 3, 4, 5 and 6 all fail on paste and take the whole config file down with them. (2) The version floor is wrong in a way that matters: the record says "Hyprland 0.54+" but then hands the reader Lua config, and Lua only exists from 0.55 (0.54.x is hyprlang). v0.54.0 shipped 2026-02-27 so it does contain PR #12418, but a 0.54 user following this record writes Lua into a hyprlang parser.

The cause above was not rewritten and may still contain the error described. The fix below is the corrected version.
RISK

debug:* options are documented "Only for developers". Do not leave debug:vfr = false in a config permanently. It burns power and defeats per-output scheduling. cursor:no_hardware_cursors = 1 moves the cursor into the composited frame, which costs a little latency and GPU time on every output.

Symptom

On its own my 144/165/180 Hz monitor is buttery. The moment I plug in the second 60 Hz screen, window animations and workspace switches judder on both. It feels like the whole desktop dropped to the slower panel's rate. Unplugging the 60 Hz display makes it smooth again. Sometimes the fast monitor's own hyprctl monitors output still says @144.00 while it visibly stutters.

Cause

Until late 2025, Hyprland scheduled *all* animations from a single monitor, an internal m_mostHzMonitor, rather than from each output's own refresh rate, so a mixed-rate setup animated everything on one shared clock. That was removed in hyprwm/Hyprland PR #12418 ("animation: improvement animation on multi refresh rate monitors"), merged 23 Nov 2025: *"I removed the logic that referenced m_mostHzMonitor for animation scheduling and switched to using each monitor's own refresh rate."* If you are on a build older than that, the judder is the bug itself. On current builds the residual causes are almost always local: debug:vfr turned off (which is a developer-only knob that forces a fixed render loop), VRR set to always-on so the refresh rate flaps on every cursor move, hardware-cursor plane bugs across differently-transformed/refreshed outputs, or the fast monitor having silently negotiated down to 60 Hz because of cable/bandwidth limits.

Fix

Steps 1, 2 and the hl.monitor snippets are correct as written. Fix the version floor and replace every hl.set with the documented hl.config form.

**1. Update first.** The floor for the animation fix is a build newer than 2025-11-23 (v0.54.0+), but everything below is Lua config, which requires **0.55+**:


hyprctl version   # need 0.55 or newer for the Lua config shown here

On Omarchy 4 use omarchy update (a direct pacman -Syu is aborted by the ALPM guard), and elsewhere sudo pacman -Syu.

**2.** Unchanged. Check the *reported* mode with hyprctl monitors -j | jq ... and pin the mode with hl.monitor({...}) if the fast panel negotiated down.

**3-6. The option settings.** All four go in one call in ~/.config/hypr/looknfeel.lua:


hl.config({
  debug  = { vfr = true },                  -- default true; "heavily recommended to leave enabled"
  misc   = { vrr = 2 },                     -- 0 off, 1 on, 2 fullscreen only, 3 fullscreen video/game
  cursor = { no_hardware_cursors = 1 },     -- 0 hw if possible, 1 never, 2 auto
  render = { new_render_scheduling = true },-- triple buffering on weaker GPUs
})

Apply them one at a time rather than all at once, so you learn which one actually mattered. hyprctl eval takes the same Lua and does not touch your files:


hyprctl eval 'hl.config({ cursor = { no_hardware_cursors = 1 } })'

The per-monitor vrr variant in the record is correct and is the better option if only one display should get VRR:


hl.monitor({ output = "DP-1", mode = "2560x1440@165", position = "0x0", scale = 1, vrr = 2 })
hl.monitor({ output = "HDMI-A-1", mode = "1920x1080@60", position = "2560x0", scale = 1, vrr = 0 })

Then hyprctl reload.

Verify

With both monitors connected, drag a window or switch workspaces on the fast display. It should be smooth while the 60 Hz screen animates at its own rate. hyprctl monitors -j | jq -r '.[] | "\(.name) @\(.refreshRate)"' should show the two different rates steadily, not one of them oscillating.

Sources

mixed-refresh-rate-monitors-animation-judder