Images

Load Image Node

Loads an image into ComfyUI.

Besides searching for the node and adding it, there are a few handy ways:

  • a. Click choose file to upload to select a file.
  • b. Drag and drop an image onto the node.
  • c. With an image in your clipboard, press Ctrl + V on the canvas to paste it as a Load Image node.

Uploaded images are copied to the ComfyUI\input folder and loaded from there. The original file is not referenced directly.

These images stay there unless you delete them from the input folder, so once you've loaded an image, you can reuse it as many times as you like.

Preview Image Node

Lets you check generated images right there. It doesn't save anything, so use it when you want a quick look at intermediate steps or results.

  • As with other image-related nodes, you can right-click and select Save Image or Copy Image to save or copy the image.
  • It's handy because you don't have to dig through the output folder for the image.

Save Image Node

Saves the generated image.

  • Default save location: ComfyUI\output
  • Sequential numbers are automatically added to the filename. (e.g., ComfyUI_00001_.png)

You can control the save directory and filename format by entering values in filename_prefix.

Commonly used formats are as follows:

Use Case Format Example Description
Insert Date %date:yyyy-MM-dd% Inserts the date (e.g., 2025-11-23)
Create Subfolder Include / Including / treats it as a folder hierarchy 
%date:yyyy-MM-dd%/cat_project   #Example
ComfyUI/output/2025-11-23/cat_project_00001_.png    #Output Example

If you want to change the default save location, set it using Command Line Arguments.

main.py --output_dir --output-directory [path]

Load Image (from Outputs) Node

The latest image saved by the Save Image node is loaded here. It's sometimes used to fake a loop.

Image Comparer (rgthree) Node

You can compare two images with a slider.

image_comparer_rgthree.json

Videos

Load Video Node

Loads a video.

In ComfyUI, many nodes treat video as a sequence of images. This node, however, outputs the VIDEO type, so you can't use it as is. You need to split the video into images, audio, and fps with the Get Video Components node.

That's why I recommend the Video Helper Suite, covered below, for working with video.

Videos are normally compressed, but ComfyUI expands them into an uncompressed sequence of images when it loads them. Loading something like a high-fps 4K video can use up all your PC's RAM and crash it, even if the clip is only a few seconds long.

The Video Helper Suite nodes let you resize frames and limit how many are loaded, so they're also the better choice for keeping memory use in check.

Save Video Node

Saves the generated video.

This time it's the other way around: the image sequence has to be turned back into VIDEO, so use the Create Video node to combine the images and audio.

🎥Video Helper Suite

Kosinkadink/ComfyUI-VideoHelperSuite

A set of handy nodes for working with video. It's been around longer than the core nodes, and since it's feature-rich and easy to use, it's often used instead.

Load Video (Upload/Path) Node

You can lower the FPS or shrink the video as you load it. If you load it at just the settings today's video generation AI needs, you avoid putting extra load on your system.

Like the Load Image node, you can also load videos by drag and drop.

Get the fps, frame count, and so on of the loaded video from the Video Info node.

Parameter Name Description
force_rate Specify frame rate (0 to disable)
custom_width Width when resizing (0 to disable)
custom_height Height when resizing (0 to disable)
frame_load_cap Maximum number of frames to load (0 for unlimited)
skip_first_frames Number of frames to skip at the beginning
select_every_nth Get 1 frame every N frames (decimation)
format Output format (presets according to the model can be selected)

Video Combine Node

Creates a video directly from an image sequence and audio.

Parameter Name Description
images Input images (sequence)
audio Input audio
frame_rate Output frame rate
loop_count Loop the input video in the output
filename_prefix Same as Save Image node
format Output format (mp4, gif, webp, etc.)
pingpong Ping-pong playback (adds reverse playback after playback)
save_output If off, it will not be saved and only previewed

Audio

Load Audio Node

Loads an audio file.

Preview Audio Node

Lets you listen to the audio right there.

Save Audio Node

Saves the audio.

There are three versions, one for each save format:

  • Save Audio (FLAC)
  • Save Audio (MP3)
  • Save Audio (Opus)

Webcam

You can capture webcam input as an image.

For details, please see Webcam Input.