🔍

Advanced API implementation of ComfyUI in Touchdesigner [ UPDATED ]

BASIC INFORMATION

The author olegchomp provides us a simple interface to ComfyUI – props on that! – The workflow needs to be implemented in the TouchDesigner project itself. Check out for updates and current download at his github page: TDComfyUI.

To make it work, just create a workflow in the default comfyUI environment and export it in the API version as *.json. Just throw this file into your TouchDesigner project and connect it to your ComfyUIAPI node. Keep in mind, that the output node in ComfyUI needs to be a WebSocket node to send data from Comfy to TD.

UPDATED WORKFLOWS!

Due to updates in both ComfyUI and Touchdesigner – we crash into horrible incompatibilites! Therefore: keep in mind to pick the right versions of both Plugin and Touchdesigner to have it running.

#0 – Install local AI engine / ComfyUI

Dowload and install ComfyUI Portable – this is an advanced emvironment to design with AI in a node based system. Download the PORTABLE, hit the update.bat and run.bat afterward.
https://github.com/comfyanonymous/ComfyUI?tab=readme-ov-file

#1 – Download TouchDesigner

Use Touchdesigner 2023.12 ! ( other work maybe )


#2 – Download the API Plugin

Download the version v.1.0.6 of the TD2Comfy Plugin by OlegChomp
https://github.com/olegchomp/TDComfyUI/releases/tag/v.1.0.6

or load the full example here



TOUCHDESIGNER WORKFLOW: td_2_comfy_controlnet_img2img


#3 – Download the workflow json for Comfy


DOWNLOAD COMFYUI WORKFLOW: comfy_workflow_depth_controlnet_preprocessor_base64

BACKGROUND INFOS

Basic img2img interfacing ComfyUI in TouchDesigner



DOWNLOAD: basic_img2img_comfy_td_workflow.json


after inserting the workflow json, just connect your source image to the comfy node


DOWNLOAd TD WORKFLOW: TD_basic_img2img_workflow

image feed 2 DepthNET to image out




DOWNLOAD td_2_img_depth_wf.json

DOWNLOAD TD WORKFLOW: TD_img_2_depth_imgout_comfy_workflow