One question about the camera model in ray-tracing program

In a simple ray-tracing application, we usually emit ray from camera to each pixel. In this condition the range of illumination information received by each pixel is limited to the solid angle that each pixel extent to camera’s position. As show in pictures below.

In first picture, pixel (b) can only receive the indirect illumination about object2. But in real life, pixel (b) should be able to receive direct illumination of object2. As shown in second picture. But this direct illumination is not accounted for in current ray-tracing methods.

Is there really something wrong with the usual calculations model or is there something wrong with my understanding?

The second figure shows light coming in from object 2 “directly” to the image plane. But, when you look in a direction, you only get light coming from that direction. I understand the confusion. A better mental model is to think of a ray from the eye looking at a little square in a screen door - think of that as the “pixel”. What do we see through that “pixel”? Say we see a table. Now, light coming from all different directions can hit that table, and some (usually tiny fraction of) that light will bounce off the table toward our eye. So, that’s how light from light sources and from other objects (which is turn are lit by light sources) affects the surface we see through our “pixel” in the screen. The “camera plane” is just a convenience for how to form eye rays and assign their results to specific pixels.

That said, there are complications, e.g., if the surface is glass or some other semitransparent material. Or the atmosphere itself, such as fog, can affect how much light travels back to the eye. But light does not travel as shown in the second figure - it does not travel to this image plane, which is just a convenience and does not normally exist in real life, e.g., in our eyes it’s the retina behind the lens that captures the image. You could put a glass diffuser plate in place of the image plane, like how showers have, just for fun. This would give you a blurry image. In this case Figure 2 would be correct, in that the diffuse plate would be affected by light from many different directions.