- {st.enhance.preparing && (
+ {preparing && (
Preparing {UPSCALING[st.enhance.active].label} upscaling
- First time on this computer only · about a minute
+ First time on this computer only
)}
diff --git a/src/player/enhance.ts b/src/player/enhance.ts
index 4889441..4cc506a 100644
--- a/src/player/enhance.ts
+++ b/src/player/enhance.ts
@@ -2,10 +2,11 @@
* Video enhancement: upscaling tiers built from mpv's scalers, debanding and
* the shaders bundled in src-tauri/shaders. Each shader's WHEN condition
* decides whether it runs for the current video and window size, so one list
- * covers every case: the neural upscalers double luma when the video is
- * enlarged (FSRCNNX above 1.3×, ArtCNN at any enlargement, so 1080p in a
- * maximized 1440p window is doubled and scaled down), KrigBilateral rebuilds subsampled colour,
- * CAS sharpens at native size (CAS-prescale also before upscaling).
+ * covers every case: FSRCNNX doubles luma when the video is enlarged (the
+ * small model above 1.3×, the large one at any enlargement, so 1080p in a
+ * maximized 1440p window is doubled and scaled down), KrigBilateral rebuilds
+ * subsampled colour, CAS sharpens at native size (CAS-prescale also before
+ * upscaling) and nlmeans denoises at any size.
*
* The tier the user chose (preferences) and the one actually running (player
* state) are kept apart: a tier whose shaders fail drops to the one below with
@@ -28,7 +29,7 @@ let shaderDir: Promise