WorkflowListedExternal

ComfyUI Utility Workflows — V3.0

(This is not an exhaustive changelog)

TaskImage to imageVersionEstimated VRAMNot yet characterizedUnratedgrade

Runtime profile

Source description

Update to V3

(This is not an exhaustive changelog)

V3:

Added WD14 Tagger

Added Batch WD14 Tagger

Added Detection Model Testing

Added Checkpoint Merger

Hardware and download-size estimates are not available yet.

Requirements

AI-generated commentary

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

ComfyUI Utility Workflows — V3.0 is a set of ComfyUI workflows for captioning or tagging, refining images or videos, generating maps, applying signatures, removing backgrounds, testing detection models, and merging checkpoints. Across the set, inputs include images, videos, and/’

The V3.0 download contains 14 workflow files, including ApplySignature.json, ImageRefiner.json, and VideoRefiner.json.

Captioning and tagging choices include JoyCaption, Batch JoyCaption, Florence-2 Captioning, QwenVL Captioning, WD14 Tagger, and Batch WD14 Tagger.
Image Refiner covers image upscaling and morphing; Video Refiner covers video upscaling and interpolation. Preprocessors generate OpenPose, Canny, LineArt, and Depth maps from an input image.
Other options apply a signature to an image, remove an image background, test detection models, or merge checkpoints.

Before use, install the listed custom nodes through ComfyUI Manager, as the listing instructs.
Suggestion · not verified

The listing names ComfyUI-JoyCaption, ComfyUI-Florence2, ComfyUI-QwenVL, and ComfyUI-WD14-Tagger for captioning workflows.
Other named custom nodes are ComfyUI-Frame-Interpolation, ComfyUI-SeedVR2_VideoUpscaler, ComfyUI ControlNet Aux, FBCNN, Rgthree, Pyosssss Custom Scripts, ComfyUI-Easy-Use, and ComfyUI Impact Pack.
The Image Refiner and Video Refiner files name 4x_foolhardy_Remacri.pth, ema_vae_fp16.safetensors, rife49.pth, seedvr2_ema_3b_fp16.safetensors, and seedvr2_ema_7b_sharp_fp16.safetensors.

The Checkpoint Merger note says to select primary and secondary XL models; connect a VAE to the nearby reroute only if you want to load one.
For Video Refiner, its note advises enabling torch.compile for long videos rather than single images.
Some workflow files specify node versions, including ComfyUI-JoyCaption 2.0.2, ComfyUI-Frame-Interpolation 1.0.7, and ComfyUI-FBCNN 1.0.1.

The listing does not specify a license.
QwenVLCaption.json lists AILab_QwenVL_Advanced; confirm that node is available in your ComfyUI setup before choosing that workflow.
Suggestion · not verified

Does this need to be edited?

Sources

1 source

CivitaiPrimaryVersion 2618109

Verified Nov 3