From d7d67223b8134f2af47a0c7e4d57f7ec7798d35c Mon Sep 17 00:00:00 2001
From: Michele Rossi <mikymod@gmail.com>
Date: Mon, 6 Jul 2026 18:07:26 +0200
Subject: [PATCH] Document edit command and performance in README

---
 README.md | 40 +++++++++++++++++++++++++++++++++++++++-
 1 file changed, 39 insertions(+), 1 deletion(-)

diff --git a/README.md b/README.md
index c92a085..de8f953 100644
--- a/README.md
+++ b/README.md
@@ -82,22 +82,60 @@ Each generation produces:
 
 Directories in `--output` are created automatically.
 
+### Edit (Qwen Image Edit)
+
+```bash
+vnasset edit \
+  --model models/qwen_image_edit_2509_fp8_e4m3fn.safetensors \
+  --input character_base.png \
+  --prompt "make her smile happily" \
+  --steps 4 --cfg 1.0 \
+  --lora models/Qwen-Image-Edit-2509-Lightning-4steps-V1.0-bf16.safetensors \
+  --output character_happy.png
+```
+
+| Option | Default | Description |
+|--------|---------|-------------|
+| `--model` | (required) | Path to Qwen Image Edit `.safetensors` |
+| `--input` | (required) | Input image to edit |
+| `--prompt` | (required) | Edit instruction |
+| `--steps` | `20` | Inference steps (`4` with Lightning LoRA) |
+| `--cfg` | `4.0` | CFG scale (`1.0` with Lightning LoRA) |
+| `--seed` | `random` | RNG seed |
+| `--lora` | (none) | Path to LoRA `.safetensors` |
+| `--output` | `output.png` | Output path |
+
 ## Current State
 
 | Command | Status |
 |---------|--------|
 | `vnasset generate` | ✅ Working |
-| `vnasset edit` | 🚧 Planned |
+| `vnasset edit` | ✅ Working |
 | `vnasset pipeline` | 🚧 Planned |
 
 ## Performance
 
+### Generate (SDXL)
+
 Radeon 8060S (Strix Halo iGPU), bfloat16, 1024×1024:
 - ~1s per step
 - ~20s for 20-step generation
 
 Model loading adds ~5s cold-start overhead.
 
+### Edit (Qwen Image Edit)
+
+Radeon 8060S, bfloat16:
+- ~120s per step at 512×512
+- ~23s model loading
+
+Larger resolutions scale proportionally. The SDPA math fallback in the text
+encoder's visual branch and the transformer's attention blocks is the main
+bottleneck.
+
+**Turbo mode:** Use the Lightning 4-step LoRA with `--steps 4 --cfg 1.0` to
+cut per-step time proportionally (4× fewer steps).
+
 ## Technical Notes
 
 ### bfloat16 required on RDNA 3.5