Update README.md

README.md

@@ -1,74 +1,77 @@
+
 ---
 tags:
 - flux
+- inpainting
 - image-to-image
 - lora
 - fal
 - diffusers
 - ilkerzgi
 language: en
-base_model: black-forest-labs/FLUX.1-Kontext-dev
-instance_prompt: Make it metallic with a black background and a 3D perspective
+base_model: black-forest-labs/FLUX.1-Kontext-dev-inpainting
+instance_prompt: place it
 license_name: flux1-dev-non-commercial-license
-license_link: https://huggingface.co/black-forest-labs/FLUX.1-Kontext-dev/blob/main/LICENSE.md
+license_link: https://huggingface.co/black-forest-labs/FLUX.1-Kontext-dev-inpainting/blob/main/LICENSE.md
 license: other
 widget:
-  - src: images/input3.jpeg
-    prompt: Make it metallic with a black background and a 3D perspective
-    output:
-      url: images/output3.jpeg
-  - src: images/input2.jpg
-    prompt: Make it metallic with a black background and a 3D perspective
+  - src: in1.png
+    prompt: place it
     output:
-      url: images/output2.jpeg
-  - src: images/input5.jpg
-    prompt: Make it metallic with a black background and a 3D perspective
+      url: out1.png
+  - src: in2.png
+    prompt: place it
     output:
-      url: images/output5.jpeg
-  - src: images/input7.jpg
-    prompt: Make it metallic with a black background and a 3D perspective
+      url: out2.png
+  - src: in3.png
+    prompt: place it
     output:
-      url: images/output7.jpeg
+      url: out3.png
 ---
 
-# Metallic Objects Kontext Dev LoRA
+# Contextual Paste – Kontext‑Dev‑Inpainting LoRA
 
 <Gallery />
 
-This is a LoRA (Low-Rank Adaptation) model that has been trained to transform input images into a metallic style, featuring a black background and a 3D perspective.
+This LoRA fine‑tunes **black‑forest‑labs/FLUX.1‑Kontext‑dev‑inpainting** to **paste** (“place it”) a given asset into a masked region so that it blends **realistically** with the surrounding scene.  
+**Square (1:1) inputs work best. LoRA scale = 1.0 or 0.9 recommended.**
+
+---
 
 ## Table of Contents
 - [Model Details](#model-details)
-- [Uses](#uses)
-- [Training](#training)
+- [Usage](#usage)
+- [Training Summary](#training-summary)
+
+---
 
 ## Model Details
 
-**Model Description:** This model is a LoRA fine-tune of `black-forest-labs/FLUX.1-Kontext-dev`. It's designed to be used in an image-to-image pipeline to apply a specific artistic style. When prompted with "Make it metallic with a black background and a 3D perspective", it converts regular images into 3D-looking metallic objects on a dark background.
+| Field            | Value                                                                 |
+|------------------|-----------------------------------------------------------------------|
+| **Parent model** | `black‑forest‑labs/FLUX.1‑Kontext‑dev‑inpainting`                     |
+| **Type**         | LoRA for masked *inpainting*                                          |
+| **Trigger**      | `place it`                                                           |
+| **Author**       | [ilkerzgi](https://huggingface.co/ilkerzgi)                           |
+| **License**      | flux1‑dev‑non‑commercial                                              |
+| **Weights**      | [Download adapter (.safetensors)](https://huggingface.co/ilkerzgi/Kontext-Dev-Inpainting/resolve/main/WVVtJFD90b8SsU6EzeGkO_adapter_model_comfy_converted.safetensors) |
+| **Demo**         | [fal.ai playground](https://infinite-kanvas.vercel.app/) |
 
-* **Trained by:** [ilkerzgi](https://huggingface.co/ilkerzgi)
-* **Model Type:** LoRA for Image-to-Image generation
-* **Language(s):** English
-* **License:** [flux1-dev-non-commercial-license](https://huggingface.co/black-forest-labs/FLUX.1-Kontext-dev/blob/main/LICENSE.md)
-* **Parent Model:** [black-forest-labs/FLUX.1-Kontext-dev](https://huggingface.co/black-forest-labs/FLUX.1-Kontext-dev)
-* **Resources for more information:**
-  * [Download the weights](https://huggingface.co/ilkerzgi/metallic-objects-kontext-dev-lora/resolve/main/metallic-objects-kontext-dev-lora.safetensors)
-  * [Try the model directly on fal.ai](https://fal.ai/models/fal-ai/flux-kontext-lora?fromTraining=fe1a8832-be01-4df6-9b26-5dcbcde83d42)
+This adapter makes sure that the input image fragment you want to insert is rendered with consistent lighting, perspective, and texture, matching the target area seamlessly.
 
-## Uses
+---
+ 
 
-#### Direct Use
-This model is intended to be used for artistic image generation. You can use it to apply a "metallic object" style to your own images. The model works best when using the trigger phrase: "**Make it metallic with a black background and a 3D perspective**".
+### Tips
 
-## Training
+* **Square crops** (512×512, 768×768) yield the most consistent blend.
+* Keep the mask tight around the insertion region.
+* Use a two‑stage upscaling workflow for high‑resolution outputs.
 
-#### Training Data
-I trained the model on a custom dataset of 20 paired (before and after) images of various objects to learn the "metallic" style transformation.
+---
+ 
 
-#### Training Procedure
-I performed the training using the [FLUX Kontext Trainer on fal.ai](https://fal.ai/models/fal-ai/flux-kontext-trainer/playground).
-The base model for training was `black-forest-labs/FLUX.1-Kontext-dev`.
+This LoRA was trained to copy the masked input fragment and render it in‑context, preserving scene coherence in terms of lighting, shading, and perspective.
 
-**Hyperparameters:**
-*   **Learning Rate:** 0.0001
-*   **Steps:** 1300
+---
+