unofficial disclaimer + FAQ (VRAM/Base/relighting/saturation)

README.md

@@ -30,6 +30,9 @@ that beats [chinoll/Moss-Image-Edit](https://huggingface.co/chinoll/Moss-Image-E
 on quality — every experiment, checkpoint, and side-by-side published,
 including the ones that lost.
 
+*An unofficial community fine-tune of MOSS-Image-SFT. Not an official
+OpenMOSS-Team model, and not affiliated with or endorsed by OpenMOSS.*
+
 ## V3 head-to-head vs chinoll (real portraits)
 
 5 real Pexels portraits (never in training), matched settings (1024px long
@@ -103,6 +106,23 @@ sampler, **shift 5.0**, **50 steps**, **CFG 3.0**, reference at source
 resolution, output up to 1024px. For edits that land too weak, raise CFG to
 3.5–4.
 
+## FAQ
+
+- **Can I run this on 6 GB VRAM?** No. The pipeline needs ~20 GB+ (bf16).
+  Community int8/convrot MOSS builds get lower, but not to 6 GB.
+- **Does it work on MOSS-Image-Base or Kromoss?** It's trained on **SFT**;
+  the native ComfyUI path (Option A) is the reliable one. Base/Kromoss
+  compatibility is unverified — if the ModelPatchLoader crashes on Base,
+  that's why.
+- **Can it relight a scene without changing the people?** Yes — dramatic
+  relighting is one of its strengths (see the lighting row in the grids).
+  For an overexposed photo, try "fix the overexposed lighting, natural
+  balanced exposure, keep everything else the same" at CFG 3.5.
+- **Colors come out more saturated than the source** — a known v3 trait
+  (thanks [@maryann088](https://huggingface.co/maryann088) for the report).
+  v4 (in training) directly targets it: masked loss keeps unedited regions
+  pixel-identical to the source, including their color.
+
 ## Troubleshooting
 
 - **"Load Model Patch failed"** → the LoRA is in the wrong folder. It goes in
@@ -119,8 +139,8 @@ Try it in the [demo Space](https://huggingface.co/spaces/akhaliq/moss-image-edit
 | Version | Recipe | Result |
 |---|---|---|
 | **v1** | warm-start + from-scratch, molbal multi-reference data, 1024px | ❌ Lost the eval gate — chinoll preserves identity better. Kept for reproducibility. |
-| **v2 (armV2-lo)** | warm-start, OmniEdit surgical data, lr 1e-5, 300 steps | ✅ Matched chinoll, won atmosphere/style. Superseded by v3. |
-| **v3 (armV3)** | continue from armV2-lo, **ConceptEdit-12M** surgical data (quality + purity filtered, 6,000 rows), lr 1e-5, 700 steps | ✅ **Published** — wins small surgical edits (sunglasses/accessories/color), ties elsewhere. |
+| **v2 (armV2-lo)** | warm-start, OmniEdit surgical data, lr 1e-5, 300 steps | Superseded by v3. |
+| **v3 (armV3)** | continue from armV2-lo, **ConceptEdit-12M** surgical data (quality + purity filtered, 6,000 rows), lr 1e-5, 700 steps | ✅ **Published** — latest checkpoint; see the head-to-head grids above. |
 
 Older eval grids (v1, v2): [akhaliq/moss-edit-hq-eval](https://huggingface.co/datasets/akhaliq/moss-edit-hq-eval).