İvme-Denoise-v1 (İvmetron-Guanine1): 2.97M-param Bayesian shrinkage denoiser

.gitattributes

@@ -33,3 +33,7 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+assets/crops.jpg filter=lfs diff=lfs merge=lfs -text
+assets/example-park.jpg filter=lfs diff=lfs merge=lfs -text
+assets/gains.png filter=lfs diff=lfs merge=lfs -text
+assets/hero.jpg filter=lfs diff=lfs merge=lfs -text

README.md

@@ -29,10 +29,15 @@ It is also the first release built on a *derived* architecture rather than a tun
 
 The codename is a **margay** — a small nocturnal wild cat with the largest eyes-to-body ratio of any cat. It denoises the dark for a living.
 
+![before / after](assets/hero.jpg)
+
+*A 12.3 MP high-ISO frame, one pass, no per-image tuning. Left: original. Right: `denoise_auto` with tiled inference, 6.5 s.*
+
 ---
 
 ## Contents
 
+- [Examples](#examples)
 - [What this is](#what-this-is)
 - [Model details](#model-details)
 - [The architecture: İvmetron-Guanine1](#the-architecture-ivmetron-guanine1)
@@ -68,6 +73,26 @@ The practical consequence: a network that is handed the correct shrinkage law an
 
 ---
 
+## Examples
+
+All three crops below are 145 % magnifications of the same frame, straight out of the model with a σ map it estimated itself.
+
+![crops](assets/crops.jpg)
+
+**Tunnel wall.** The original carries heavy chroma speckle in a region that is essentially black. It is gone, and the wall is black rather than blue-grey. This is the case that a conventional denoiser either leaves noisy or turns into a checkerboard.
+
+**Train front.** The headlights are blown to clipping and the livery sits in near-darkness. The edge of the front panel survives, the "34" stays legible, and the highlight keeps its shape instead of blooming.
+
+**Tactile strip.** Saturated yellow under sodium light, which is where JPEG chroma subsampling does the most damage. The grain is removed without the plastic-looking flatness you get from a heavy Gaussian blur.
+
+A second frame, a park scene in daylight, denoised at σ = 6 — the value its own noise estimator picked:
+
+![park example](assets/example-park.jpg)
+
+The gentle setting is the interesting one: at σ = 6 the model has almost nothing to do, and it correctly does almost nothing. Detail is preserved because the energy gate never closes.
+
+---
+
 ## Model details
 
 | Parameter | Value |
@@ -160,6 +185,12 @@ Note what is **absent**: no additive bias after the gain, no normalisation, no a
 
 `Proj` is zero-initialised, so every block starts as the identity and the network starts as a no-op. All the dB are learned, not inherited.
 
+Here are the 192 learned gains of one block, plotted against `log e`. This is the *entire* non-linearity of the model — there is nothing else:
+
+![gains](assets/gains.png)
+
+Every curve is a monotone gain that runs from ~1 at low energy to ~0 at high energy, with the family spread out across channels because the learned frame is overcomplete and is not the eigenbasis of the patch covariance. Anything that looks like a familiar activation would have to be justified by the GSM posterior mean; none of them are, so none of them are here.
+
 ### Design decisions that mattered
 
 **No bias in the shrinkage.** The single biggest inductive-bias win. A conv with bias can add energy where the prior says there is none; the GSM estimator cannot.
@@ -263,6 +294,10 @@ At σ = 50 that is **+2.2 dB over BM3D** — and BM3D takes ~3.3 s per image whe
 
 **Accuracy per GFLOP: 0.553 dB/GFLOP** for this model versus 0.225 for Restormer — a **2.5× efficiency advantage**, at 11 % of the parameters and a third of the training time.
 
+![pareto](assets/pareto.png)
+
+Read the plot as: anything to the upper-left dominates. İvme-Denoise-v1 is above DnCNN at 65 % of its compute, and sits well to the left of the transformer cluster — 0.35 dB costs you 2.5× the FLOPs.
+
 Latency, RTX PRO 6000, bf16, batch 1, no tiling:
 
 | resolution | time |