Skip to content
WorkflowVerifiedExternal
Download

Nvidia PiD 4K Image Upscaler Workflow (Z-Image, Flux.1, Flux.2, SD3 compatible) by Smittie — v1

initial release

Sourcecivitai.com/models/2677911
TaskMultiCapabilitiesImage to image · UpscalingFamilyZ-ImageVersionRelease 3006857 — Smittie_Nvidia_PiD_4K_Image_Upscaler.jsonEstimated VRAMNot yet characterizedUnratedgrade

Runtime profile

Access
External workflow
Status
Verified
Version
Smittie_Nvidia_PiD_4K_Image_Upscaler.json
Source
Civitai
Updated
Nov 3, 2026

What you need

Model files
0.9 GiB known subtotal

Remaining hardware or download-size estimates are not available yet.

Requirements

Models

4

Node packs

6

  • ComfyMath

    Possible

    Node pack · Registry

  • comfyui-easy-use

    Found

    Node pack · Registry

  • comfyui-kjnodes

    Found

    Node pack · Registry

  • GetNode

    Not resolved

    Node pack · Unknown

  • rgthree-comfy

    Found

    Node pack · Registry

  • SetNode

    Not resolved

    Node pack · Unknown

AI-generated commentary

AI-generated explanation based on source and configuration details. Suggestions are clearly labeled.

Purpose

  • Use this image-to-image workflow to upscale a low-resolution 512 or 1K image to 2K or 4K with Nvidia’s PiD (Pixel Diffusion); the input is an image and the output is a higher-resolution image.

Flow

  • The workflow includes nodes for loading and resizing images, PiD conditioning, and sampling.

Inputs and outputs

  • Its graph contains PreviewImage and SaveImage nodes for previewing and saving results.

Setup

  • Update ComfyUI to the latest build with PiD support, install missing custom nodes through ComfyUI Manager, and download the needed models.

Dependencies

  • The instructions name pid_flux1_1024_to_4096_4step_bf16.safetensors, gemma_2_2b_it_elm_bf16.safetensors, and ae.safetensors; a VAELoader selection in the workflow is not identified.

  • Custom-node names include comfyui-easy-use, comfyui-kjnodes, rgthree-comfy, ComfyMath, GetNode, and SetNode; check ComfyUI Manager to identify which packages provide ComfyMath, GetNode, and SetNode.

    Suggestion

Configuration

  • For Z-Image or Flux.1-like images at 1K, the instructions specify pid_flux1_1024_to_4096_4step_bf16.safetensors as the diffusion model.

  • The workflow notes suggest euler_ancestral for a crisper, sharper look and lcm for a softer look; to keep the image’s current resolution, they say to disable auto_resolution and set width and height to 0.

  • The title describes the workflow as compatible with Z-Image, Flux.1, Flux.2, and SD3.

Cautions

  • The workflow description does not specify a model-use license; check each model file’s usage terms before using it.

    Suggestion

Does this need to be edited?

Sign in to send an edit request.

Sources

1 source
CivitaiPrimaryVersion 3006857
civitai.com/models/2677911

Verified Nov 3

Workflow instructions

Author notes from Smittie_Nvidia_PiD_4K_Image_Upscaler.json · selected release

### The best samplers are: ### - euler_ancestral for more crisp, sharp images. ### - lcm for a softer look.

Author notes from Smittie_Nvidia_PiD_4K_Image_Upscaler.json · selected release

### Here you can load your image and crop or resize it. The longest side should be around 512 or 1K (depending on the model you use), which start_resolution and auto_resolution does for you, if enabled. ### If you don't care for the optimal resolution and want to keep the resolution as is, disable auto_resolution and set width and height to 0.

Author notes from Smittie_Nvidia_PiD_4K_Image_Upscaler.json · selected release

# Smittie's PiD Upscaler Workflow ## Description With this workflow you can upscale low-resolution images (512, 1K) to much higher resolutions (2K, 4K) using [Nvidia's PiD (Pixel Diffusion)](https://research.nvidia.com/labs/sil/projects/pid/). It is a simple image to image (I2I) workflow. ### ## Install Instructions ### 1. Update ComfyUI to latest build with PiD support ### 2. Install all missing custom nodes via the ComfyUI Manager ### 3. Download all needed models **Diffusion model** - [All available](https://huggingface.co/Comfy-Org/PixelDiT/tree/main/diffusion_models) - For Z-Image or Flux.1 like images in 1K resolution use [pid_flux1_1024_to_4096_4step_bf16.safetensors](https://huggingface.co/Comfy-Org/PixelDiT/blob/main/diffusion_models/pid_flux1_1024_to_4096_4step_bf16.safetensors) **Text encoder** - [gemma_2_2b_it_elm_bf16.safetensors](https://huggingface.co/Comfy-Org/PixelDiT/tree/main/text_encoders) **VAE** - [ae.safetensors](https://huggingface.co/Comfy-Org/z_image_turbo/tree/main/split_files/vae) **Model Storage Location** ``` 📂 ComfyUI/ ├── 📂 models/ │ ├── 📂 diffusion_models/ │ │ └── │ ├── 📂 text_encoders/ │ │ └── │ └── 📂 vae/ │ └── ``` ### ## Video Instructions If you don't know what to do, you can watch the [Nvidia PiD Tutorial by AI Search](https://www.youtube.com/watch?v=DlvhlQOBHBw). He works with a different workflow, which I have built upon. ### ## Acknowledgement Most credits goes to the Nvidia, ComfyUI and CivitAI team, as well as kijai and AI Search. Thank you guys!

Download files

Model artifacts

2 artifacts

nvidiaPid4KImageUpscalerWorkflowZ_v1.zip

zip · 9.19 KB · SHA-256 60500fc3ca5c…302e · Civitai

Download file

Smittie_Nvidia_PiD_4K_Image_Upscaler.json

json · 64.8 KB · SHA-256 3ef0a4091997…efc1

Download file

Source excerpts

3 excerpts
NoteNov 3civitai.com/models/2677911

Source context: 685 downloads · Type Workflows · Base model ZImageTurbo

NoteSep 29civitai.com/models/2677911

Source context: 647 downloads · Type Workflows · Base model ZImageTurbo

NoteSep 28civitai.com/models/2677911

Source context: 643 downloads · Type Workflows · Base model ZImageTurbo