What parameters should I change to make this image better ? Something is off with the samples but changing the number doesn’t do anything …

And I’ve noticed some square artefacts on the shadows. This appear when the image is not in it’s native scale.
I have a low performance PC that’s why i’m using this primary+secondary GI methods.
Also, I can’t export a render in BMP or JPG format in any other depth colour other than 8 bit/channel. Is this an issue concerning the CPU/GPU ? I thought JPG was compliant with depths up to 64 bit/channel. Some times I get colour bands because of that. TIFF works fine.
