Edit photos by instruction: today's results · Edit Photos by Instruction in ComfyUI with Qwen Image

Open Command Prompt inside the ComfyUI folder
Activate the virtual environment before running Python commands
Pull the latest source with git and install requirements

Action	Result
Type cmd in address bar	Command Prompt opens in ComfyUI folder
Activate virtual environment	Environment name appears before prompt
Run git pull --ff-only	ComfyUI source updates to latest version
Run pip install -r requirements.txt	Only changed packages install

Do this in your own tools, step by step. Ticks are saved on this device.

Open your ComfyUI folder
Type cmd in the address bar and press Enter
Run the update commands in order

What does the Lecture 2 update route do?

Updates ComfyUI source and requirements inside its virtual environment

What workflow opens after updating?

The Qwen Image 2.1 text-to-image workflow from Lecture 3

00:00 Greetings everyone. Today I am going to show you how to edit images with instructions in ComfyUI. We will change 1 detail of a photo, and keep everything else. We will also replace words on a sign, dress this man in a jacket from a 2nd photo, and keep 1 explorer consistent across 3 storyboard scenes. Check out the tutorial description to see the chapters, and to download the workflow

00:35 and the photos we use. Let us get started. Before the new material, let us update ComfyUI the same way we did in Lecture 2. I am opening our ComfyUI folder, which holds the application, its virtual environment and the model path file we already prepared. In the address bar, type cmd and press Enter. A Command Prompt opens directly in this folder, so every command we run next works inside this ComfyUI installation. First, activate the virtual environment of this installation.

01:08 The environment name appears in front of the prompt, so the Python commands that follow use the packages of this ComfyUI. Now pull the latest ComfyUI source with git. The fast forward option moves our copy along the official history, and git lists every file that this update changes. Next, install the requirements into this environment. Pip compares each package with the version the new

01:39 source asks for and installs only what changed. Lecture 2 explains every step in detail if you need it. The update is installed, and ComfyUI is running again. Refresh the browser to load the new version. Then open the Qwen Image 2.1 workflow from Lecture 3; it is also in this lesson's files. This is text-to-image. The Empty Latent Image is where sampling starts: pure noise. Denoise is 1,