The technical details involve using an image processing apparatus comprising of a pixel region with numerous pixels. Each pixel having its own microlens and numerous photoelectric conversion units. This also requires an amplification unit to amplify a signal output from the pixel region, and a scanning to read a the signal as well as the signal output from the conversion unit. Still with us. The rabbit hole gets a little deeper, but it helps if you follow along with Figure 1. I recommend Google Translate for the patent images.
After passing through the amplification and conversion units, the image passes through the control unit. Now the magic can happen. The processing unit is then able to expand a dynamic range, or it can focus faster.
To improve dynamic range, the amplification unit amplifies part of the signal at different gain levels. To improve focusing response, the amplifying unit amplifies the other part of the signal not amplified for dynamic range. And lastly, the processing unit does not expand the dynamic range, but the focus detecting unit performs the focus detection. Clear as mud? It helps to look over the patents and read the above a few times to make sense of it. That's what we did. But in essence, the patent explains how Canon is able to expand dynamic range in their cameras while also improving the focusing performance.
Canon Patent JP 2023-138838A Complete Blueprints