I've shot 4k footage since 2014, and when mastering in 1080p, I often use digital zoom in post. There are multiple factors I would look at.
- Cameras that use a single sensor will display debeyering artefacts when examining the direct pixel readout (mushier detail and reduced colour information). This issue is nicely improved when downsampling 4k to 1080p, but is revealed when digitally zooming. The Z190, however, uses a 3-chip design. If each chip is full 4k (which the marketing copy certainly implies), the quality difference between downsampled and a direct pixel crop should be significantly less noticeable than on single-chip competitors.
- Lens resolution can be a contributing factor as well. If the lens barely keeps up with the 4k resolution of the sensors, or the three-chip design introduces optical issues, then downsampling will likely hide minor softness. Digital zoom will reveal these optical shortcomings simply by magnifying what's being projected onto the sensor. On the upside, lens performance is typically best near the centre, so a centre crop is at least likely to be optically stronger than the edges.
- While exposure and colour should look identical because nothing is changing (aperture, shutter speed, gain, pixel pitch), you're right that the exposure performance will change, as measured by the amount or type of visible grain. From user reports I've read, this is best demonstrated in the Sony A7 series of cameras where similar light sensitivity (based on the amount of visual noise) can be seen in direct pixel readouts from the A7s (giant pixels, lower resolution) and downsampled readouts from the A7r (smaller pixels, higher resolution, but the subsequent higher noise is reduced during downsampling). Digital zoom will reveal the per-pixel noise that downsampling would have averaged out, which can impact perceptual sharpness.
- If you decide to record in 4k and downsample to 1080p afterward, zooming as needed in post, compression may come into play as well. While the Z190 records 4:2:2 10-bit, digital zoom can reveal minor compression softness in colour detail that would have been reduced during downsampling. However, I would expect this to apply only to downsampling/zooming 4k footage in post. If the camera is recording a 1080p 4:2:2 image directly, switching between a downsampled full sensor readout and a crop of the centre pixels should have little to no impact on compression results.
- Some cameras may simply process things slightly differently when performing downsampling versus direct pixel readout.
When tracking down the source of the softness, I'd recommend starting by recording 4k footage and cropping it in post, then comparing it to footage shot using the 1080p zoom mode. Though highly unlikely, if poor camera processing of 1080p footage is to blame you'll see the difference here.
It's more likely a combination of lens resolution, optical performance of the three-chip design, and perceptual sharpness improvements from reduced noise when downsampling. But by all means, try some locked down resolution tests first! After tracking down the source(s) of the softness, then it's a question of how to combat it.
My approach has always been to shoot in 4k and rely on "digital zoom" in post while never reaching a full 2x zoom (direct pixel crop). This ensures the footage is always downsampled to some extent, maintaining the higher relative quality we're used to seeing now.