tulut_comfyui_Gadget Introduction
tulut_comfyui_Gadget is a set of custom ComfyUI nodes for restoring small details in full-body anime images made with Anima models. In a wide composition, a face, hand, or foot may occupy only a few pixels. Even when the rest of the image looks good, those areas can appear blurry, distorted, or incomplete.
The extension helps you work on those problem areas individually. It detects faces, hands, and feet, enlarges the relevant regions for refinement, then places the refined details back into the image. Its aim is to make full-body artwork clearer while retaining the sharp lines and clean shading of the original anime style. The author describes the extension as currently intended only for Anima models.
How tulut_comfyui_Gadget Works
Think of a full-body image as a painting viewed from across the room: the whole pose is visible, but the smallest features are hard to make out. The detailer temporarily moves closer to a face, hand, or foot so that area has more room to be refined.
The process has four main parts:
- Find the area. A detection model identifies a face, hand, or foot in the image.
- Focus on it. The extension crops that region so it can be processed separately from the full image.
- Enlarge and refine it. Upscaling and resampling give the small region more working resolution. The detailer then restores detail using the selected workflow settings.
- Return it to the composition. The refined region is incorporated into the full image.
You can use the face, hand, and foot detailers for their respective problem areas. For example, if a portrait has a clear face but poorly defined hands, you can focus on hand refinement rather than treating every area as equally in need of repair.
The workflow example shows the nodes used with an upscale model. AnimaLLLite support is built into the detailer nodes: you select an AnimaLLLite model in a node’s parameters rather than connecting separate AnimaLLLite nodes.
Tulut Detailer workflow showing the detailer nodes and an upscale model
tulut_comfyui_Gadget Features
Face refinement. The Tulut Face Detailer uses face_yolov8m.pt to locate anime faces for localized refinement. This is useful when a character’s face looks soft or loses definition in a full-body shot.
| Before | After |
|---|---|
| Face before refinement | Face after refinement |
Hand refinement. The Tulut Hand Detailer uses hand_yolov8s.pt to find hands. Use it when small hands appear unclear or distorted in an otherwise usable composition.
| Before | After |
|---|---|
| Hand before refinement | Hand after refinement |
Foot refinement. The Tulut Foot Detailer uses FootYolov8x_v20.pt to find feet. It targets missing or indistinct foot details that can occur when feet take up little space in the original image.
| Before | After |
|---|---|
| Foot before refinement | Foot after refinement |
Combined refinement. You can use all three detailers in a full-body workflow when the face, hands, and feet each need attention. The examples below show a combined result and a closer breakdown of the affected areas.
| Before | After |
|---|---|
| Full-body image before combined refinement | Full-body image after combined refinement |
Breakdown comparing face, hand, and foot details
Embedded AnimaLLLite selection. The detailer nodes let you select an AnimaLLLite model within their parameters for localized use. This keeps that choice inside the detailer rather than requiring an external node connection.
Upscaling and resampling. The workflow can use an upscale model, such as CUGAN, while dynamic resampling works with the target guide size. These steps are intended to provide enough resolution for refinement without losing the crisp appearance of anime line art.
tulut_comfyui_Gadget Models
The extension names three detection models. Their job is to find the regions to refine; they are not separate image-generation models.
| Model | Used by | What it finds |
|---|---|---|
face_yolov8m.pt | Tulut Face Detailer | Faces |
hand_yolov8s.pt | Tulut Hand Detailer | Hands |
FootYolov8x_v20.pt | Tulut Foot Detailer | Feet |
Choose the detailer that matches the feature you want to improve, or use all three for a full-body image with several problem areas. The extension also supports selecting an AnimaLLLite model inside a detailer node and using an upscale model such as CUGAN in the workflow. The provided information does not identify particular AnimaLLLite or upscale model files, so it does not establish a complete list of choices for those settings.
Learn More about tulut_comfyui_Gadget
Start with the full workflow image to see how the detailer nodes fit into an image-processing workflow. Then compare the face, hand, and foot examples with their after images above to see what each detailer is meant to address. The detail breakdown matrix provides a closer view of the three refined features.
