You're going to have to explain what you mean when you say "combined a shifted and an unshifted HDR image." Stitching generally uses AI-assisted warping and alignment. Incidentally, what you've posted is a JPEG, not an HDR image, so most of the tonality is coming from tone mapping in postprocessing -- in other words, a tone-mapping AI created it. ;-) In fact, the tones shown are synthesized and do not reflect the true appearance of the scene.
I think HDR-like techniques can be used specifically to approximate the appearance of a place. When we walk into this kind of lighting, our visual impression isn't formed at any one instant. The way our eyes dart around from dark to light areas, irises opening and closing constantly, is a lot like making multiple exposures. When the visual cortex combines them into what becomes our impression of the space, we're essentially doing HDR.
When I photographed spaces with this kind of lighting, my internal debate was often between fidelity to the experience and showing more interesting detail. I worked in spaces darker than this, and worried that being too true to the experience would lead to a very gloomy body of work. Most often I chose a compromise: shadows elevated enough to show all the interesting things, but not so much that the scene looked completely uncanny. This was sometimes easier said than done. When people make comments like, "that's cool, I thought it was a painting!" I worry that I went too far.
FWIW, my favorite exposure blending technique didn't use any AI, or high-bit depth intermediate image, or tone mapping. It just chose the highest contrast pixel groups from each image, and created a 16-bit TIF that you could then adjust as needed. I liked it because it wasn't really capable of wild effects. The difference between its results and what you could do with a single exposure were subtle: no blocked highlights, and no noise in the shadows.
I wish I can count every 200 mil pixels to make sure they not cheating. lol 
