This Slider LoRA is designed to control the camera zoom (zoom-in / zoom-out) in anime-style illustrations.
caution
### ⚠️ Important: How to Use
* **Keep the .json and .safetensors files together:** Please place the .json file in the same folder as the .safetensors file, and **make sure both files have the exact same filename**.
* **Tested Environment:** This model has been tested and confirmed working in **ComfyUI** I do not know if it works with other tools or UIs, and I cannot guarantee compatibility outside of ComfyUI.
* **Required Base Model:** Trained on and only guaranteed to work with [MOSS Image SFT Deband](https://civarchive.com/models/3007711/moss-image-sft-deband-fix-color-banding-or-int8-or-fast-and-quality-versions?modelVersionId=3412867). Compatibility with other models cannot be guaranteed due to the specifications of the training tool.
### ⚙️ Recommended Settings & Behavior
* Weight / Strength: Designed for positive values only. Negative values are not recommended, though they might still work decently.
* Stacking with Other LoRAs (Rank 32): This slider was trained at Rank 32. If you use it alongside other high-rank LoRAs or stack many LoRAs at once, the slider's effect will weaken. Please adjust the strength of each LoRA to find the right balance—this is expected behavior (by design).
### 📝 Author's Note
Just my personal opinion, but conventional LoRA wisdom doesn't seem to apply much to MOSS Image models. Training LoRAs for it is quite tricky, and I'm still struggling with it as we speak lol.
Please note: I don't speak English and I don't have much free time, so please understand that I cannot answer questions or take requests.
---
Trained using [Comfy-moss-slider-node](https://github.com/Shiba-2-shiba/Comfy-moss-slider-node). Many thanks to the creator, Shiba*2!
