Some more test shots today. Here is 5K-30P (camera set to low sharpening):

Here is 4K-60P (camera set to low sharpening):

In terms of actual resolution, these look pretty close to me.

Here is 4K-60P (camera set to medium sharpening):

That's just a crazy amount of sharpening. For comparison, Here is my Hero 8, 2.7K-60P (camera set to medium sharpening):

2.7K seems to me to be the sweet spot for Hero 8 resolution, and in-camera medium sharpening is giving decent results. The cynic in me wonders if the Hero 10's resolution bump as reported in the reviews is really just an overdose of in-camera sharpening.

Using Vegas' Unsharp Mask Fx, I can't get close to what the Hero 10 is doing in that 4k-medium shot. Vegas' Sharpen does a better job with the wood grain in the corner regions -- thank you wjauch -- but you really have to dial it up. Photoshop's unsharp mask seems to give better control over the wood grain; I can't seem to get the radius right in Vegas Unsharp. PS's smart sharpen can't seem to find edges in the wood grain, so not useful there. Resolve's sharpen tool gave I thought ok but overall mediocre results.

Looking again at the Hero 10 shots, I'm wondering if I'm too close to the fence and there's a focus issue, esp in the corners. I guess I'll reshoot again tomorrow and see if a little distance helps. Got to say: this is not exactly thrilling picture quality. The Hero 8 is making a strong case for itself at 2.7K. But no horizon-leveling, which I think is one of the biggest reasons to upgrade.

Is this a new camera? Have you examined the lens for smudges and flaws?

It is new, and I thought about that possibility, but the GoPro's are kind of notorious for less than stellar image quality. I'll see if I can find some downloadable samples for another reference. I do wonder about those corner regions, esp compared to my H8.

There's convolution Kernel too, preset sharpen, you'd have to read up on it if you wanted to make adjustments, it's controls are more complex

Voukoder, in addition to having a customizable Unsharp filter, also has two top-tier denoise filters, hqdn3d and NLMeans. which are admittedly quite slow.

Congratulations on avoiding destructive high-pass sharpening filters.

Disregard the H10 images above for quality -- I think I may have had Hypersmooth on, which is destructive. Will reshoot and post again. Sorry!

Ok, here are some thoughts after reshooting my comparisons:

H10 in-camera "medium" sharpening is crazy (unusable) on 4K, not as aggressive on 5K. H8 optimizes at 4K/30, which has a slight but clear resolution advantage over 4K/60 and 2.7K/60. H8 in-camera "medium" sharpening is nice, at least for my test scene, and easily good enough that I would just go with it rather than taking on the extra step of sharpening in post.

As for the H10's 5K...I'm just not seeing it. The H8 has a more detailed picture at 4K/30 than the H10 at 5K/30. The H10's 5K footage is challenging to sharpen in post, and even with effort does not match the sharpness of the H8's 4K/30. Sigh.

I've played around with both cams in wide mode and tried correcting distortion using Vegas/Photoshop's tools. GoPro's in-camera correction ("linear") is better, at least compared to VP/PS's presets.

Looking at the vectorscope, the H8's yellows, greens, and cyans are very stunted--a definite weakness. The H10 looks better balanced in terms of color (this is comparing both on Native). I haven't tried grading H10 yet to see if that translates into real-world benefits.

All of these resolution tests were done with stabilization off and the camera sitting on a tripod. That is not how most people will use the camera. I'm moving on to hand-held comps to see if the H10's horizon-leveling mode at least matches the H8's resolution when Hypersmooth is on. That would be nice, as it is two generations downsteam.

H10 5K/30 Linear (in-camera sharpening--low):

H8 4K/30 Linear (in-camera sharpening--med):

Medium is too much for in-camera sharpening. There is piping in the bottom image.

And the sharpness wthout "Hypersmooth" is about what I expected -- I knew something was off.

There is piping in the bottom image.

I'm not familiar with that term for sharpening artifacts. What's "piping"?

Your neighborhood seamstress knows. It's probably better known as a "halo" artifact.

Got it. It's there on the H8 "medium" footage, for sure, but maybe not an unreasonable compromise given how soft the footage is to start.

But the H10's "medium" is just bonkers.

I've seen comments on other forums complaining about this in the Hero 9 (the start of the new generation), so I guess no fix forthcoming. :(

So original point remains: with the H10, looks like you can choose either wildly over-sharpened footage (and that's on Medium; who knows what High looks like??), or very soft "low" footage which is challenging to sharpen in post.

Your original intuition is correct -- stick with unsharp mask or similar in post, keep in-camera sharpening at minimum, and live with the rest. You said,

 As you'd expect, pretty big difference between claimed and actual resolution with GoPro.

Something interesting I notice in the low/medium comparisons is that GoPro's in-camera sharpening appears more evenly distributed from the center to the corners, while Vegas' sharpening FX seem to affect the center more.

I think what's happening is that the image itself is blurrier in the corners, and GoPro's algorithm compensates for that by adding more sharpness to the corners, while Vegas' just apply sharpness evenly across the image. One way to replicate this within Vegas is to use the Bezier Mask fx.

Something interesting I notice in the low/medium comparisons is that GoPro's in-camera sharpening appears more evenly distributed from the center to the corners, while Vegas' sharpening FX seem to affect the center more.

Both are apparently destructive high-pass filters. That is not a trivial consideration because blown pixels propagate over multiple fx layers.

To repeat my original advice -- if you feel you must sharpen minimally in-camera, don't use a high-pass filter in post. Nunca.

Unsharp Mask is a low-block filter. As suggested, sometimes what is needed is a light denoise filter. Personally, I wouldn't spend a lot of time fooling with the OOF corners in GoPro. One suggestion is to shoot at 4k or 8k and crop into the frame for a lower resolution output.

Not sure if it's clear but with GoPro there is not a no-sharpening option. Setting is either low, medium, or high. For the H8, I think "medium" is a very reasonable compromise esp given what most users want to do. For the H10, "medium" is significantly more destructive--unacceptably so, to my eyes.

Which means starting with the H10's extremely soft picture on the "low" setting and then building something pleasing in post.

When I'm facing a challenge like this, I like to deconstruct things so I can see how someone else solved the problem. In this case, that means learning how to replicate the H8's in-camera "medium" sharpness, so that I can add it to H10 footage, and modify it to taste.

I've been frustrated because, using Vegas' Unsharp Mask, I can't get close to matching whatever it is that GoPro is doing. With Vegas' Sharpen fx, I get a better match, though I always though Sharpen was just a simpler version of USM, and whatever you could do with Sharpen you could also do with USM.

Cropping in general not a great solution for GoPros. I don't have the tools to measure precisely, but my spidey sense says GoPro's true resolution is around FHD -- and maybe not even that.