Even if there were no gap, the filter edges might leak some light into the image by conducting light that enters at the filter’s cut like in a (lousy) optical fibre. To prevent this, the edges of the filter should be painted black, something that e.g. Leica does in (some of) their lenses.
I assume the blur in the grass is due to the excessively long exposure time, and perhaps a small breeze? It does now look infraredish though. I think you paid a lot less for that filter than I paid for my filters.
Now you need to find some spot to put the infrared capability to use, with camera on tripod, and set up perfectly. I don’t know how you will be able to focus it though - for me, I had to experiment, as the focusing marks on the lens no longer apply. I went out to my balcony, then shot one image at infinity, then more shots for each of the markings on the lens depth of field scale. As I recall, if I put the infinity mark on the focusing ring right over the f/11 or f/8 mark on the camera, I got the best results - I put a red dot there between those two settings, which is what I now use for “infinity”.
I suspect you did focus, but your focus information is all for white light, not infrared.
I can take a photo of my 35mm 7Artisans lens, to show you where the new “infinity” mark is. It’s much more of a correction than I expected, but I think the 7Artisans lens just turns more than I expected. With my old cameras and lenses in the 1970’s, there was a tiny “R” marked on the lens, to show the infinity marking for infrared. I can send you a photo of that too, if it would be helpful.
Option to ‘turn off’ (disable) the ‘Face detection’ in the ‘Smart Lighting → Spot Weighted’.
Face detection, its okay, but i like to add manually with the tool if needed. I also use the tool some percent (like 5-7% on photos) when no face at all - and its scan thru for nothing.
Smaller selection in ‘Face detection’ auto finded ‘selected box’ → i found (so, my personal point) usually its select a bit too large (the selected box), its more select the ‘head’ then the face. May with some percentage settings for that to shrink a bit. Example: if photo is 5000x4000px and the detected face is 1000x1000px, the selection box is can be smaller - like 75% (25% reduction of the original box), may with some non linearization based on detected width/height, like: if the detected face is only 400x400px, its may can shrink to 90% (10% percent reduction).