Mark Entropy as the preferred candidate and show its mask

README.md

@@ -93,11 +93,15 @@ The mask sheets show all seven latent-time layouts for each prompt. Each sheet c
 | Control | [MP4](examples/control/p3_dense_lora.mp4) | [MP4](examples/control/p3_detail50_warmup.mp4) | [MP4](examples/control/p3_uniform50_1x2.mp4) |
 | Preserve | [MP4](examples/preserve/p3_dense_lora.mp4) | [MP4](examples/preserve/p3_detail50_warmup.mp4) | [MP4](examples/preserve/p3_uniform50_1x2.mp4) |
 
-## 🟨 ENTROPY selector screen — validation only
+## 🟨 ENTROPY — best current candidate
 
 Entropy was among the stronger sparse candidates in the initial layout screen: at step 0, entropy50's mean teacher gap was 0.429, compared with 0.416 for detail50 and 0.565 for uniform2 (lower is better). Detail50 was slightly better on that metric. This separate inference test compares matched 50% token-budget entropy placement with a root-shuffled entropy control at warm-up σ=0.9 and 30 steps. It uses four validation prompts only; it did not receive an untouched confirmation set and is not a promoted result.
 
-Each sheet highlights **ENTROPY** and places its tile pattern next to the uniform and shuffled controls and a frame from the corresponding entropy video. The seven rows show the seven latent-time layouts.
+**ENTROPY is the best current candidate and the preferred algorithm for this project.** The available selector run is a four-prompt validation screen, not an untouched confirmation. The initial teacher-gap screen slightly favored detail50 (0.416 vs. 0.429 for entropy50), so this preference should not be read as a statistically confirmed benchmark win.
+
+The four-column overview places the reference, ENTROPY output, its actual t=0 tile mask, and shuffled-ENTROPY output side by side. Each prompt sheet below shows all seven latent-time layouts alongside uniform and shuffled controls.
+
+![Four-column ENTROPY comparison with the tile mask](examples/entropy-selector/comparison.png)
 
 ### Prompt p0