22

static/footer.md

@@ -1,6 +1,25 @@
 
-## Usage
-- **Input Images**: Upload image(s) to train LoRA.
+
+## Step by ste instruction
+- Upload a minimum 1 image to "Input Images".
+- Press "Generate LoRA".
+- After a moment, your LoRA will be ready to download.
+- Press "Generate image" to use LoRA above.
+The generated image will have similar style as the images you uploaded.
+
+If you're using ComfyUI, you can try plugging the downloaded LoRA into: <https://huggingface.co/Qwen/Qwen-Image>.
+(I haven't tried this yet, so please let me know if you have)
+
+If you're using Python, you can use the LoRA above to generate an image using instructions from DiffSynth-Studio (link below).
+
+## Features
+- Use a single image to generate LoRA (though more images are better).
+- You can download the LoRA you generate.
+- There's also an option to generate an image using the LoRA you created.
+
+## Speed
+- LoRA generation takes about 20 seconds (H200 ZeroGPU).
+- Image generation using LoRA takes about 50 seconds (maybe something wrong here).
 
 ## References
 - **DiffSynth-Studio**: <https://huggingface.co/DiffSynth-Studio/Qwen-Image-i2L>