Something I found interesting when testing Lumariver was how differences between it and Adobe's DNGPE are small when colors fit well within a small gamut like sRGB, but then grow huge when colors exceed that gamut. No suprise, I guess. Kind of like comparing two cars. At 50mph on cruise control over a smooth road, not much difference. Crank it up to 80mph and take a curve over some railroad tracks, now you feel the difference.
Like most people I suspect, I do intial tests using the CC24. Shots of that chart usually fit within sRGB, with only the Cyan patch pushing the limit, and then just barely. Differences between Lumariver and DNGPE are small, but easily visible. I found the Lumariver reds are more "red" (more magenta) than the DNGPE version. Lumariver is usually more saturated. A slight bump in sat or vibrance on a DNGPE version brings it closer to Lumariver.
But then I tested on real world shots with "challenging" colors, mostly reds and yellows in some Asian costumes. The Lumariver vs. DNGPE differences were huge. And more interesting, the Lumariver results varied widely depending on the "compression" method chosen under the "Look" options. I discovered that the "no compression" option was wiping out fine detail in strong reds. The "Adobe98" and "Adobe98 Strong" compression options retained detail, but produced significantly different color.
On more inspection I realized that the loss of detail in the Lumariver "no compression" image was due to my monitor. The detail was there and I could see it by viewing the RGB channels separately (in ProPhoto 16bit). But my "normal" gamut NEC monitor profile just could not display the detail in color.
But then I noticed something strange. When I converted that "no compression" image from ProPhoto to Adobe98, the detail was permanently lost. Inspecting the individual channels I saw that the Green channel was totally blocked. The RGB values were 0-0-0 across most of the strong red color. The Red and Blue channels looked OK. The lesson I learned was stay away from "no compression" and use the default "Adobe98 strong".
The question of which was more accurate remained. So I conducted a highly scientific test. A group of grandkids spent the weekend. As usual, they were dressed in bright, neon glow clothing. I made a lot of shots in bright sun, then picked three that challenged gamuts, mostly in reds, yellows, and oranges. I processed each with my best DNGPE profile and a Lumariver "Adobe98 Strong" profile, and displayed them side-by-side on my profiled NEC monitor. FWIW, these were Canon 5D4 shots.
I brought two of the kids in and asked them which colors on the screen looked closer to the real thing, which they were still wearing. My eyes are old, and my left eye was injured years ago, so I don't trust them. These kids are 12 and 15, so their eyes are as good as it gets. They both chose the DNGPE versions. In 2 of the 3 cases they said it was obvious, no doubt. These were reds. In the 3rd case, a more orange color, they were less certain, but both chose the DNGPE version.