I often thought it would be useful to generate a lot of similar images with specific elements present or missing from them. However I could not find suitable ComfyUI nodes to achieve this purpose. Therefore I have written some ComfyUI nodes which allow for all combinations of a set of features to be easily added to an image. This allows you to easily create a set of images with certain elements present or absent from each image.
You can find these nodes on my Github account or download them via the ComfyUI interface by searching for ComfyUI JMNodes. The basic idea is that you can set a number generator to cycle through all the numbers up to a specified value. Another node then converts these numbers into their binary equivalents. These binary outputs can then be fed to a switchable string input node, This only contributes a string if it receives a true value. Thus you can submit all combinations of a set of prompts by running through the appropriate number range.
To reinforce unusual combinations for instance Santa but no snow, you can use the Boolean not node. This inverts the Boolean outputs and then you feed the resulting prompts as a negative prompt to your image generator.
You can find a JSON workflow demonstrating this as part of the library. It is also embedded in the images in this article. If you are interested in creating your own custom nodes for ComfyUI, I can recommend this video tutorial and these articles, one on Reddit, and one on GitHub. The Github tutorial is due to be moved to the ComfyUI documentation website I believe.
I hope you find these new nodes useful. You might also find my other ComfyUI articles about image generation workflows, or upscaling and style transfer useful. Merry Christmas.
