WorkflowListedExternal

[Flux,SD] Segment multi comparison workflow(easy) — v1.0

Sourcecivitai.com/models/1041990

TaskImage to imageFamilyFLUXEstimated VRAMNot yet characterizedBgrade

Runtime profile

Access: External workflow
Status: Listed
Source: Civitai
Updated: Oct 2, 2026

Source description

Segment Performance Comparison Workflow

This workflow compares the segmentation performance of Florence2, segment anything, and ultralytics. It's compatible with SD1.5, SDXL, and FLUX, and allows easy integration of YOLO models. The workflow includes mask splitting and selection features, enabling various experiments. Designed for easy use in ComfyUI, it effectively analyzes the performance of segmentation models.

세그먼트 성능 비교 워크플로우

이 워크플로우는 Florence2, segment anything, ultralytics의 세그먼트 구분 성능을 비교합니다. SD1.5, SDXL, FLUX와 호환되며, YOLO 모델을 쉽게 적용할 수 있습니다. 마스크 분할 및 선택 기능도 포함되어 있어 다양한 실험이 가능합니다. ComfyUI에서 사용하기 쉽게 설계되었으며, 세그멘테이션 모델의 성능을 효과적으로 분석할 수 있습니다.

===========================================================

AI-generated commentary

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

Purpose

Use this ComfyUI workflow to compare image segmentation with Florence2, Segment Anything, and Ultralytics: load an image, compare the segmenters, then split and select masks.

Flow

Mask nodes include SegsToCombinedMask, MaskToImage, MaskPreview+, and PreviewImage.

Inputs and outputs

LoadImage is the named image input; the description says masks can be split and selected.

Setup

The workflow requires these ComfyUI node packs: comfyui_segment_anything, ComfyUI_essentials, florence_dw, ComfyUI-Custom-Scripts, ComfyUI-Impact-Pack, and ComfyUI-Impact-Subpack.
The listed model files are microsoft/Florence-2-base, sam_vit_b_01ec64.pth, and segm/skin_yolov8m-seg_400.pt.

Dependencies

Named Florence2 nodes: DownloadAndLoadFlorence2Model, Florence2, Florence2Postprocess, Florence2Run, and LoadFlorence2Model.
Named Segment Anything nodes: GroundingDinoModelLoader (segment anything), GroundingDinoSAMSegment (segment anything), SAMLoader, and SAMModelLoader (segment anything).
Other named nodes: ImpactSEGSToMaskBatch, ImpactSimpleDetectorSEGS, LoadImage, MaskPreview+, MaskToImage, PreviewImage, SegmDetectorSEGS, SegsToCombinedMask, ShowText|pysssss, UltralyticsDetectorProvider, and VHS_SelectMasks.

Configuration

The description lists compatibility with SD1.5, SDXL, and FLUX, and says YOLO models can be integrated.
The listed base model is Flux.1 D.

Cautions

Before setup, check that segm/skin_yolov8m-seg_400.pt is the YOLO model you intend to use.
Suggestion

Does this need to be edited?

Sources

1 source

CivitaiPrimary

civitai.com/models/1041990

Verified Jul 20

Source excerpts

1 excerpt

NoteJul 20civitai.com/models/1041990

Source context: 290 downloads · Type Workflows · Base model Flux.1 D