Update README.md

README.md

@@ -4,10 +4,6 @@ tags:
 - lora
 - diffusers
 - template:diffusion-lora
-widget:
-- output:
-    url: images/eeeeeeeeeee.png
-  text: '-'
 base_model: black-forest-labs/FLUX.1-Kontext-dev
 instance_prompt: >-
   [photo content], recreate the scene from a bottom-up perspective. Preserve
@@ -15,11 +11,70 @@ instance_prompt: >-
   background sky or floor elements adjust naturally to the new angle,
   maintaining authentic shadowing and perspective.
 license: apache-2.0
+language:
+- en
+pipeline_tag: image-to-image
+library_name: transformers
 ---
-# Kontext-Bottom-Up-View
+# **Kontext-Bottom-Up-View**
 
-<Gallery />
+The Kontext-Bottom-Up-View is an experimental adapter for black-forest-lab's FLUX.1-Kontext-dev, designed to transform scenes into a bottom-up perspective, preserving accurate depth, scale, and lighting direction to enhance overall realism. The model ensures that background elements such as sky or ground surfaces adjust naturally to the new viewing angle, maintaining coherent geometry, texture consistency, and visual balance. It was trained on 800 image pairs (400 start images and 400 end images) to achieve precise, geometry-consistent bottom-up scene generation.
 
+> [!note]
+[photo content], recreate the scene from a bottom-up perspective. Preserve accurate depth, scale, and lighting direction to enhance realism. Ensure the background sky or floor elements adjust naturally to the new angle, maintaining authentic shadowing and perspective.
+
+---
+
+## Parameter Settings
+
+| Setting                  | Value                    |
+| ------------------------ | ------------------------ |
+| Module Type              | Adapter                     |
+| Base Model               | FLUX.1 Kontext Dev - fp8 |
+| Trigger Words            | [photo content], upscale the low-quality image to 4K resolution, enhancing sharpness, clarity, and fine details while preserving the original texture, colors, lighting, and natural appearance. Remove noise, blur, and compression artifacts without over-smoothing or distorting facial or object features. Ensure realistic depth, balanced contrast, and accurate tones, achieving a high-definition, lifelike result that maintains the integrity of the original image. |
+| Image Processing Repeats | 45                       |
+| Epochs                   | 24                       |
+| Save Every N Epochs      | 1                        |
+
+    Labeling: DeepCaption-VLA-7B(natural language & English)
+    
+    Total Images Used for Training : 800 Image Pairs (400 Start, 400 End)
+
+## Training Parameters
+
+| Setting                     | Value     |
+| --------------------------- | --------- |
+| Seed                        | -         |
+| Clip Skip                   | -         |
+| Text Encoder LR             | 0.00001   |
+| UNet LR                     | 0.00005   |
+| LR Scheduler                | constant  |
+| Optimizer                   | AdamW8bit |
+| Network Dimension           | 64        |
+| Network Alpha               | 32        |
+| Gradient Accumulation Steps | -         |
+
+## Label Parameters
+
+| Setting         | Value |
+| --------------- | ----- |
+| Shuffle Caption | -     |
+| Keep N Tokens   | -     |
+
+## Advanced Parameters
+
+| Setting                   | Value |
+| ------------------------- | ----- |
+| Noise Offset              | 0.03  |
+| Multires Noise Discount   | 0.1   |
+| Multires Noise Iterations | 10    |
+| Conv Dimension            | -     |
+| Conv Alpha                | -     |
+| Batch Size                | -     |
+| Steps   | 3500 & 400(warm up)  |
+| Sampler | euler |
+
+---
 
 ## Trigger words
 
@@ -33,8 +88,6 @@ You should use `and lighting direction to enhance realism. Ensure the background
 
 You should use `maintaining authentic shadowing and perspective.` to trigger the image generation.
 
-
 ## Download model
 
-
-[Download](/prithivMLmods/Kontext-Bottom-Up-View/tree/main) them in the Files & versions tab.
+[Download](/prithivMLmods/Kontext-Bottom-Up-View/tree/main) them in the Files & versions tab.