Drop ISNet and other obsolete pipeline mentions from docs.

README.md

@@ -17,8 +17,7 @@ Open-source background removal and alpha matting from RGB images. This repositor
 hosts the **OSS variant** exported as a self-contained ONNX graph for ONNX Runtime.
 
 The graph includes DepthAnythingV2 and a ConvNeXt-fused matting head — no
-PyTorch checkpoints are needed at inference time. ISNet, DINOv3 ViT foreground
-segmentation, and the refiner are **not** included.
+PyTorch checkpoints are needed at inference time.
 
 - **Try it live:** [withoutBG on Hugging Face Spaces](https://huggingface.co/spaces/withoutbg/withoutbg)
 - **Website:** [withoutbg.com/open-weights-model](https://withoutbg.com/open-weights-model)
@@ -37,7 +36,6 @@ segmentation, and the refiner are **not** included.
 | ONNX output tensor | 448 × 448 |
 | Depth | DepthAnythingV2 `vits` (`dav2s`) |
 | Matting | ConvNeXtFusedMattingUNet (DINOv3 ConvNeXt `base`) |
-| Refiner | disabled |
 | Transformer opt | disabled |
 | ORT offline opt | extended |
 | Size | ~455 MB |
@@ -56,12 +54,11 @@ and SHA256.
 
 ## Architecture
 
-v10 matting-only pipeline:
+v10 pipeline:
 
 - **Depth:** DepthAnythingV2 `vits` (`dav2s`)
 - **Matting:** ConvNeXtFusedMattingUNet — frozen DINOv3 **ConvNeXt** backbone
   fused into a U-Net on 4-channel RGB+depth at **448²**
-- **Not included:** ISNet edge maps, DINOv3 ViT foreground segmentation, OSS refiner
 
 Consumers letterbox RGB to the fixed ONNX input tensor (`canvas_size` in the
 sidecar) and run a single inference session. Depth is computed inside the graph.

withoutbg-onnx.md

@@ -23,8 +23,8 @@ withoutbg-open-weights.onnx.json
 ```
 
 The ONNX graph is self-contained for inference. It includes DepthAnythingV2 and
-the ConvNeXt-fused matting head (matting-only; no refiner). The serving
-container does not need PyTorch checkpoints at runtime.
+the ConvNeXt-fused matting head. The serving container does not need PyTorch
+checkpoints at runtime.
 
 Do not distribute local training checkpoints unless the release explicitly calls
 for source-weight distribution.
@@ -40,12 +40,11 @@ for source-weight distribution.
   `https://ai.meta.com/resources/models-and-libraries/dinov3-license/`
 - Main use: background removal / alpha matting from RGB images
 
-v10 matting-only pipeline:
+v10 pipeline:
 
 - Depth: DepthAnythingV2 `vits` (`dav2s`)
 - Matting: ConvNeXtFusedMattingUNet — DINOv3 **ConvNeXt** `base` fused into a
   U-Net on 4-channel RGB+depth at **448²**
-- **Not included:** ISNet, DINOv3 ViT fgseg, OSS refiner
 
 ## ONNX Contract
 
@@ -70,8 +69,8 @@ resize the alpha matte back to the original image size.
 
 Read the sidecar JSON first when present. It records the authoritative
 `canvas_size` (448), `matting_input_size`, `depth_variant`, `convnext_size`,
-`refiner: false`, input and output names, precision, model version, SHA256, and
-optional benchmark/parity values.
+input and output names, precision, model version, SHA256, and optional
+benchmark/parity values.
 
 ## Export Command