Skip to content
WorkflowListedExternal
Open source

Qwen-Image refiner — v1.0

With this model you can create highly refined images using, QWEN, in this workflow you will be able to use Loras that are known to ruin the original picture's quality, because the picture will be refined, this...

Sourcecivitai.com/models/2218152
FamilyQwen-ImageVersionRelease 2497272 — QWEN-Refined.jsonEstimated VRAMNot yet characterizedBgrade

Runtime profile

Access
External workflow
Status
Listed
Version
QWEN-Refined.json
Source
Civitai
Updated
Oct 18, 2026

Source description

With this model you can create highly refined images using, QWEN, in this workflow you will be able to use Loras that are known to ruin the original picture's quality, because the picture will be refined, this workflow will be able to create the most complicated pictures and it can make unusable LORAS, usable again.

The different detailers require separate models that should go in your ComfyUI /models/ultralytics/bbox folder. You can find these files here and here .

Inspired by GBRX

What you need

Model files
15.3 GiB known subtotal

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

Requirements

Models

12

Node packs

2

AI-generated commentary

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

Purpose

  • Qwen-Image refiner — v1.0 is a Qwen workflow for creating and refining images with LoRAs, Qwen model files, and separate detailer models; it states that it produces highly refined images and can make otherwise unusable LoRAs usable again.

Flow

  • The workflow requires the ComfyUI-Impact-Pack and ComfyUI-Impact-Subpack node packs.

Inputs and outputs

  • Inputs include object detectors, a segmentation model, a VAE, a text encoder, a UNet, LoRAs, and a checkpoint.

  • The stated output is highly refined images.

Setup

  • Place the separate detailer models in ComfyUI /models/ultralytics/bbox.

Dependencies

  • Named Qwen files are qwen_2.5_vl_7b_fp8_scaled.safetensors, qwen_image_fp8_e4m3fn.safetensors, qwen_image_vae.safetensors, and Qwen-Image-Lightning-4steps-V1.0.safetensors.

  • Other named model files are bbox/Eyeful_v2-Paired.pt, bbox/face_yolov8m.pt, bbox/female_breast-v4.2.pt, bbox/nipples_yolov8s.pt, bbox/vagina-v3.2.pt, sam_vit_b_01ec64.pth, and xl-innie-biglove.safetensors.

  • It also lists gonzalomoXLFluxPony_v60PhotoXLDMD.safetensors, comfyui-impact-pack, and comfyui-impact-subpack.

Configuration

  • The listed node-pack pins are cd34cfdd638a996fd011be1b490c70a7e0fb505f for comfyui-impact-pack and 1.3.2 for comfyui-impact-subpack.

Cautions

  • Before use, verify these exact filenames are available: bbox/female_breast-v4.2.pt, bbox/Eyeful_v2-Paired.pt, bbox/nipples_yolov8s.pt, bbox/vagina-v3.2.pt, and sam_vit_b_01ec64.pth.

    Suggestion · not verified

  • Also verify these exact filenames are available: Qwen-Image-Lightning-4steps-V1.0.safetensors, xl-innie-biglove.safetensors, and gonzalomoXLFluxPony_v60PhotoXLDMD.safetensors.

    Suggestion · not verified

Does this need to be edited?

Sign in to send an edit request.

Sources

1 source
CivitaiPrimaryVersion 2497272
civitai.com/models/2218152

Verified Oct 18

Source excerpts

3 excerpts
NoteOct 18civitai.com/models/2218152

Source context: 285 downloads · Type Workflows · Base model Qwen

NoteAug 6civitai.com/models/2218152

Source context: 247 downloads · Type Workflows · Base model Qwen

NoteJul 19civitai.com/models/2218152

Source context: 238 downloads · Type Workflows · Base model Qwen