But the output does look MUCH sharper and detailed than the preview.
That's the key problem!
It's an interesting observation to get to the bottom of, but it may be an inaccurate observation!
On-the-fly downsampling (as in screen previews smaller than at 100% zoom) can be done in several ways. Broadly speaking, all downsampling risks producing aliasing artifacts that may be mistaken for resolution (while they are actually artifacts).
So assuming ACR previews (either in LR or PS)
of a TIFF output file from Capture One, there is a good chance of visible differences below 100% zoom. DPP will also have its own demosaicing and offers two levels of on-screen display (a slower higher quality one, and a faster lower quality one), so I'll leave that out of the comparison since there are multiple variables at play.
The question then becomes, which application does a better job of mimicking the full-size 100% display, after resampling to output size and post-resampling output sharpening, when viewed at sub-100% zoom levels (or simulated actual output size)?
The only way to really/reliably predict how the final image will look is by viewing it at a 100% zoom level (because that avoids downsampling), and mentally adopting output resolution at a given viewing distance. But even then there remains a mismatch in the case of printed output, which is of much higher resolution than the display we preview on is capable of showing. That obviously makes it harder to judge the full image from only a scaled down preview, but such is the challenge of displaying/previewing downsampled images. Therefore, resampling artifacts and viewing distances/resolutions keep screwing with our visual acuity (which also varies by individual).
Capture One (version 10) is pretty unique at providing an Output Sharpened Preview (the tool is called 'Recipe Proofing'), which is only really possible at 100% zoom level AFTER profiling for output and sharpening AFTER resampling for output size (with the obvious caveat of having adequate display gamut and resolution, which is also often not sufficient for a 100% predictable preview).
So not an easy task to declare which is better, but interesting enough to explore.
Cheers,
Bart