Floyo
Floyo
Workflows
API
Pricing
Floyo
Floyo
Workflows
API
Pricing

Basic Upscaler Workflow

Using for basic upscale every single image

27

Generates in about 3 mins 17 secs

Nodes & Models

LoadImage
CheckpointLoaderSimple
UpscaleModelLoader
CLIPTextEncode
ImageUpscaleWithModel
ImageScaleBy
VAEEncode
KSampler
VAEDecode
SaveImage

This workflow is a straightforward Image-to-Image (Img2Img) Upscaler pipeline built in ComfyUI. It enhances resolution, refines fine details, and sharpens low-resolution input images using model-assisted upscaling combined with a secondary diffusion pass (Ultimate SD Upscale style logic).

---

Key Features & Pipeline Steps

1. Load Image & Checkpoint: Starts with a base image (e.g., 512x768) and loads your preferred Stable Diffusion checkpoint.

2. Model Upscaling: Utilizes an upscale model (such as RealESRGAN_x2.pth) paired with an Upscale Image By node (set to 2.0x scale) to increase raw image dimensions efficiently.

3. VAE Encoding: Converts the upscaled pixel image back into latent space via VAE Encode.

4. K-Sampler (Refinement Pass):

* Denoise Strength: Set to a low value (e.g., 0.35) to preserve original character features and pose while enhancing textures and sharpness.

* Sampler Settings: Uses euler sampler with normal scheduler across 22 steps.

5. Final Output: Decodes back to image space, delivering a crisp, high-resolution result (e.g., 2048x3072).

---
How to Use

1. Load your input image into the Load Image node.

2. Select your desired SD model in Load Checkpoint.

3. Choose your preferred upscaler model in Load Upscale Model (e.g., RealESRGAN, 4x-UltraSharp).

4. Adjust the Denoise parameter on the KSampler (0.3 - 0.4 works best for detail enhancement without changing the original art style).

5. Queue Prompt to generate!

Read more

N