I've been playing a lot with DeNoise lately too. However, I have yet to find an advantage over ACR/LR that justifies the additional workflow steps, time, and cost.
I loaded your orange box images into Photoshop, and would disagree that the DeNoise version is better. If I blow it up 300-400% and look at the flat orange area on the top I see "artifacts" that look like jpeg compression artifacts. They are not jpeg artifacts, just look like them. Dark splotches, 3 to 5 pixels wide.
I imagine these are from the sharpening in DeNoise (detail recovery, blur reduction). Your Lightroom version does not look much better, but if I take your original and process it in ACR for just noise reduction and use maybe a little stronger setting than you apparently used, I get a version that looks a little better than the DeNoise.
If I also sharpen the original in ACR, I get similar artifacts. But if I use a strong mask, the artifacts are suppressed while the fine detail is still enhanced (the white lens outline top left, the printing, etc.).
I like that DeNoise has separate settings for Overall, Shadows, and Highlights. I wish ACR/LR could do that. However, the DeNoise Shadows adjustment reaches way too far into the midtones. The Highlight adjustment is pretty good and restricted to just highlights. Also, the Detail Enhancement and Blur recovery seem to generate far too many artifacts. And DeNoise does not offer a mask on these adjustments like ACR/LR does. In your sample, the flat orange area gets hit when it does not need to, and the results are not good.
But overall, you have to pixel peep at huge magnifications to really see the differences, and even then they are minor. That's the reason I see little value in DeNoise over ACR/LR. In reduced size web images and even prints, I doubt if the differences are visible.