Implement a filter to translate the image to the detected head or eyes#77
Implement a filter to translate the image to the detected head or eyes#77schuellerf wants to merge 1 commit intoallo-:masterfrom
Conversation
|
Can you change the class Name (not the filter name) to CamelCase to match the other filters? I tested it by moving the background image. |
|
And maybe you should rebase it onto my branch to check for any conflicts. But I think there were no breaking changes (except that it will break when using mediapipe, but the other filters using part masks are currently broken with mediapipe as well). |
|
Which branch? Due to |
For a parallax effect, I would also include to enrich the image scaling to be able to scale larger than the webcam dimensions (for now there is only scaling with the resulting image being the webcam resolution again, I think) |
Rebase and change the class name on the branch you want to merge (currently
Indeed, the Image moved around over a black background, but having an image larger than the output frame is not that complicated. I am not sure how to fit it in the current pipeline which passes webcam size frames around, but for the effect alone maybe the filter could load the image itself and then re-use the head tracking filter. I'd need to read what other movements belong to a real parallax effect. It just looked already similar when I tested moving around the background image. |
I just wanted to play around with a mainly transparent image just containing some text which floats next to my head.
It seems to work and does what I intended to