Update README.md

README.md

@@ -14,18 +14,36 @@ instance_prompt: null
 ---
 # Wan-LoRA-Arcane-Jinx-v1
 <Gallery />
-<b>Please note that this is just a quick test I made. I didn't intend to share it, but eventually uploaded the LoRA model upon request. 
-Because of this, there isn't much information shared about it and most probably I'll remove the repo in the next 24 hours. 
-In the future, I'll dig deeper into it and upload the final models.</b>
 
-This LoRA was trained on the 1.3B Wan t2v model.
+## Trainig details:
+I tried a method where I first trained the LoRA on images only and then after a certain number of steps continued training only on videos.
 
-<b>TRIGGER WORDS</b>: csetiarcane, J1nx
+- LR: 2e-5
+- Optimizer: adamw_optimi
+- epochs: 15 (image) + 7 (video)
+- steps: 2550 (image) + 2520 (video)
+- dataset: 135 videos and 135 images
+- repeats: 10
+- rank: 128
+- batch size: 2
+- gradient accumulation steps: 1
 
-<b>Strength</b>: try lowering it during inference. It is possible that the model is overtrained a bit. I didn't had a chance to test it properly.
+For training I used the [Diffusion-Pipe](https://github.com/tdrussell/diffusion-pipe/tree/main) repo. 
 
-## Download model
+## Inference
 
-Weights for this model are available in Safetensors format.
+For inference you can use 
+- [ComfyUI native nodes](https://blog.comfy.org/p/wan21-video-model-native-support)
+- [Kijai's wrapper nodes](https://github.com/kijai/ComfyUI-WanVideoWrapper)
 
-[Download](/Cseti/Wan-LoRA-Arcane-Jinx-v1/tree/main) them in the Files & versions tab.
+<b>Trigger words</b>: csetiarcane, Nfj1nx
+
+(If it isn't enough so you don't get the style/character, I recommend adding "animation style" to you prompt. It can help providing the style but in some cases the result will be too cartoony)
+
+<b>Strength</b>: 0.9-1.0
+
+## Acknowledgement
+
+Special thanks to Kijai for the tireless and excelent work done for the community to enable us to use these solutions as soon as possible,
+as well, as to Comfyanonymous for implementing them so quickly with such a good quality natively into ComfyUI,
+and also to TDRussel for making LoRA training available to the community so quickly.