Are there any best practices in terms of using either of these in a bot? How does a change of resolution, zoom, scaling etc affect the reliability? I’d assume the image recognition would be sensitive. I think I might be having an issue but don’t recall the resolution I used when capturing images.
You can find some OCR best practices in this topic. Most of them are applicable to all image actions.
If the resolution when you capture the image is different from resolution when the script is executed, image-based actions can fail because the bot will look for exact image on the screen. So, if the resolution or zoom change, you need to recapture the images.
We are now working to make image-based automation more flexible. Feel free to post examples of your image-based actions that fail here