kye

I own and use both. The Oly 12-100mm F4 is a great lens, but it's much bigger and heavier than the Pana 14-140mm (560g versus 265g) - both on an OM-1:

On an E-M1 ii/iii or OM-1 the 12-100mm supports Sync-IS which gives sublime video IS performance, but even with the relatively light (for that kind of camera) OM-1, the combo is 1.2 kg and somewhat front-heavy if you're using it handheld. A GH7 + 12-100mm would be nearly 1.4 kg.
As a 'travel' lens, IMHO the combination of low weight and focal length range makes the 14-140mm almost perfect (other than in really low light, of course).

@PannySVHS I've now tested the Crop Zoom (CrZ) mode in 1080p.
This is the first test, and I exposed for the sky (which it thought was the right thing to do) which meant that the plants were a bit low, so I ended up bringing them up a little in post.  The Prores HQ is great at retaining noise and so there's quite a bit visible despite me having shot this at base ISO 500.  I've found that ETTR is definitely recommended if you want a more modern looking cleaner image.
I also used the 12-35mm lens at F4.0 for all images as that's where it's the sharpest.
First is comparing the C4K Prores HQ vs 1080p Prores HQ (on a 1080p timeline):


Next we compare the CrZ vs zooming with the lens.
I have prepared these images in sets of three.  The first is the CrZ image, the second is zooming with the lens, the third is the CrZ image again but with sharpening added.  This allows you to compare both CrZ images directly with the 'proper' one, as the more zoomed CrZ images did look a little soft in comparison when viewed at 300%.
Around 14mm (1.16x):



Around 18mm (1.5x):



Around 25mm (2.08x):



Once I got those images into Resolve and looked at them I decided to re-shoot it with a better exposure.  So I chose a different framing that meant the sky wasn't influencing anything.  However, I didn't realise that where I was standing was going in and out of the sun, so some shots were washed out and I had to compensate for it in post, adjusting contrast/sat/exposure/WB to match.  Tests are never perfect but are enough to give a good idea of what's going on, and in real use where there is no A/B comparing going on no-one would ever spot it anyway.
There's also a slight difference in exposure between the C4K and 1080p modes too, which is a bit odd.  I imagine it's due to changing the sensor mode.  I compensated for that in all these tests too.
C4K Prores HQ vs 1080p Prores HQ (on a 1080p timeline):


Around 14mm (1.16x):



Around 18mm (1.5x):



Around 25mm (2.08x):



I am actually rather encouraged by these results, as my previous test was in low-light and I did on something with much sharper edges and that showed differences I'm not really seeing here.
However, it's not really surprising that the GH7 did this well, as even with a CrZ of 2.08x it's still reading an area of the sensor around 2776 pixels wide.  I say "around" that wide because there is a slight crop when you compare the native 5.8K mode with the native C4K, 4K, and 1080p modes, but I think the 2.08x crop will still be oversampled from the sensor by a good amount.
The other thing I noticed was that I couldn't adjust the CrZ function while I was recording, the button just didn't do anything.  I'm not sure if that's because I have it assigned to a button and that there might be some other way to engage it while recording.  Maybe through the controls that are used to control powered zoom lenses, not sure.
Anyway, it looks pretty darn good to me, and the grain actually reminds me of the OG BM cameras which are quote noisy at native ISOs too (and also lots of seriously high-end cinema cameras too).

The 9mm I tested is the Panasonic Leica 9mm F1.7, I'm not aware of a 9mm F1.4 - maybe you're thinking of the Leica 12mm F1.4?
Let me see if I can further tempt you!!
I have done some tests (images below) but found the following:
You can use the Crop-Zoom function (CrZ) to go up to 1.3x in C4K and up to 1.4x in 4K resolutions There is no 4K option in Prores, only C4K If the sensor was cropped to be a 1:1 readout, it would be a 1.4x in C4K and a 1.5x in 4K, but the CrZ mode stops just short of these amounts.  I suspect that they have limited it so that it is always downscaling, even if just slightly.
Test shots.
First set are with the S-16 Cosmicar 12.5mm F1.9 C-mount lens.  These are all on a 4K timeline, so you can really pixel-pee if you want to.  I didn't have quite enough vND to have it wide open on all the shots, so some are wide open but some are stopped down to F2.8.







Now, I switched from the 4K to the C4K, which meant I had slightly less crop available and you can just start to see the edges of the image circle.  I suspect your mileage would vary depending on what lens you were using.  The Cosmicar is pretty wide, so if you were using a long focal length you'd probably get no vignetting at all.
This should also give a comparison between the 4K H.264 and the C4K Prores HQ.

Now we switch lenses to the 12-35mm and stopped down to F5.6 so we can compare the CrZ crop to a non-cropped image.
This is cropped to 1.3x using the CrZ function in C4K Prores HQ:

and this is without any CrZ and using the 12-35mm to zoom in to match the FOV:

I didn't shoot any clips this morning comparing the CrZ mode in 1080p, but I can also shoot a test for this if you're curious.
I had a closer look and discovered you can't change the zoom amount, which seems to be stuck at 3x zoom.  I'd say that it is resolving enough for focusing and I used it with the Cosmicar in the above test.  It's the normal story of using peaking and rocking the focus back and forth to find the sharpest spot.
At least I'd say that if you can't use it to manually focus then the problem isn't the punch-in feature but some other issue!

I'd argue that this kind of testing is actually necessary to understand how things behave.  Over the years I have tested a lot of things and it's amazing how many things that "everyone knows" do not stand up in even the most basic tests, but continue to be myths because no-one bothers to even look.
Aristotle claimed that women have fewer teeth than men, which is not true, but he obviously never actually looked to see if he was right - despite being married multiple times where he could easily have tested his claim at any time.
No, not mixed up, but the 12-35mm has a shallower DOF and so you have to know where in the image to look to compare sharp details in the focal plane.
This is the unsharpened cropped image:

This is the 12-35mm image:

This is the sharpened cropped image:

The sharpening is perhaps a little over-correcting, but the thin edges are still slightly blurred in comparison to the proper image from the 12-35mm.
This is where it is important to know how to read the results of a test.  This comparison of the zoom to the crop matched FOV but not DOF, and while I probably could have zoomed in using the 12-35mm and also stopped down at the same time to keep DOF the same, the lens sharpness would have been reduced so it wouldn't have been a fair test.  To get around that I should have tested using a flat surface like a resolution chart or a brick wall.
The problem with going that route is that now we're no longer testing anything close to real-life, and no longer answering questions about what will and won't work in real shooting.
The test wasn't "what percentage of resolving power is lost using the CrZ function?"...  it was "is the CrZ function usable for shooting with cropped lenses?".  Realistically I shouldn't have included the 12-35mm optical zooms at all, I should have just cropped in using the CrZ function and left the images to be judged on their own merits in isolation, the same way that any project shot using the CrZ function would be.
This is the danger of pixel-peeing - it distracts from the only thing that actually matters - the image.
The cosmicar really is a gem!
There's a reason that cinematographers have relentlessly driven up the price of vintage lenses over the last decades, and why modern lens manufacturers are designing and releasing brand new lenses with vintage looks, and manufacturers are even creating new mechanisms to control the amount and type of vintage looks with custom de-tuning functions.

The elephant in the room is Resolve.
As I have discussed and demonstrated in my "New travel film-making setup and pipeline - I feel like the tech has finally come of age" thread, over the last decade Resolve has gotten more feature-rich, but more importantly, it's made it HUGELY easier to use and get good images.
People now have a lot more knowledge about colour grading tools and techniques, that's for sure, but things like the Film Look Creator enable you to use a single node, you set your input and output colour spaces, and then you can adjust exposure / WB / saturation / contrast and all sorts of other things in the same tool.  You don't even need to apply a film look at all...  just select the "Blank Slate" preset, which sets it to have no look at all, and you can still use all the tools to adjust the image without having to worry about colour management at all.
Any improvement in your post-processes is a retroactive upgrade to your camera, your lenses, and all the footage you have already shot.
Colour grading is such a deep art that I think the average GH5 user back in the day was probably extracting a third of the potential of the images they'd shot, if that, simply because they didn't know how to colour grade properly.  I'm not being nostalgic about the GH5 either, the same applies for any camera you can think of.
There are reasons to upgrade your camera, for sure, but most of the reasons people use aren't the right reasons, and they'd be better spent taking the several thousand dollars it would take for a camera upgrade and taking unpaid leave from their job and improving their colour grading skills instead.

The elephant in the room is Resolve.
As I have discussed and demonstrated in my "New travel film-making setup and pipeline - I feel like the tech has finally come of age" thread, over the last decade Resolve has gotten more feature-rich, but more importantly, it's made it HUGELY easier to use and get good images.
People now have a lot more knowledge about colour grading tools and techniques, that's for sure, but things like the Film Look Creator enable you to use a single node, you set your input and output colour spaces, and then you can adjust exposure / WB / saturation / contrast and all sorts of other things in the same tool.  You don't even need to apply a film look at all...  just select the "Blank Slate" preset, which sets it to have no look at all, and you can still use all the tools to adjust the image without having to worry about colour management at all.
Any improvement in your post-processes is a retroactive upgrade to your camera, your lenses, and all the footage you have already shot.
Colour grading is such a deep art that I think the average GH5 user back in the day was probably extracting a third of the potential of the images they'd shot, if that, simply because they didn't know how to colour grade properly.  I'm not being nostalgic about the GH5 either, the same applies for any camera you can think of.
There are reasons to upgrade your camera, for sure, but most of the reasons people use aren't the right reasons, and they'd be better spent taking the several thousand dollars it would take for a camera upgrade and taking unpaid leave from their job and improving their colour grading skills instead.

This seems like a simple question, but the more I think about it, the less simple it gets.
Let's start out with the seemingly obvious answer - it looks like Super-16 because the sensor is literally a S16 sized sensor.  End of thread, thanks for coming, byeeeee!
Here are some thoughts suggesting it looks more like S35, or at least more than S16.  Some are good arguments, some aren't, but summed up I think they're hard to dismiss.
It appears sharper than S16, a lot sharper.
Without getting overly technical, S35 has around 4K resolution, but the level of contrast on the fine details is quite low, and it's well known that by the time you print and distribute a 35mm film it really only looks like about 2K once it's projected in cinemas.  This is perhaps the biggest argument for me - the P2K just looks like cinema did in the 90s.  I know this isn't comparing a 35mm neg scan with the P2K files, but virtually all the memories of 35mm film that most people would have are from movies shot and distributed on film, not from viewing modern film scans.
Lenses are much sharper now too, adding to it.  S16 lenses were often very vintage!
We have speed boosters, much faster lenses, and much wider lenses now.
One of the looks of S16 was longer focal lengths and deep DOF, but if we were to use the P2K like we would use any other camera, it would be with speed boosters and faster lenses which would have much shallower DOF.  The wider lenses we have now would be much sharper and faster too.  So the lens FOV, lens DOF, and sharpness combinations would all be much more like S35 was, and perhaps even exceed it.
How it's used would be much more modern.
The framing, movement, lighting, locations and subjects also play a role in 'placing' a medium.  This has probably changed less than the above arguments, and the things that any of us might shoot are more likely to still resemble things that I would associate with S16 (like FNW and TV and low-budget projects).
I'm curious to hear thoughts from others.
I've been reviewing my equipment and got to the P2K and thought "oh, it's a pocketable S16 camera" but my brain immediately added "that looks like 90s movies" and then I realised that these two things don't align!

The elephant in the room is Resolve.
As I have discussed and demonstrated in my "New travel film-making setup and pipeline - I feel like the tech has finally come of age" thread, over the last decade Resolve has gotten more feature-rich, but more importantly, it's made it HUGELY easier to use and get good images.
People now have a lot more knowledge about colour grading tools and techniques, that's for sure, but things like the Film Look Creator enable you to use a single node, you set your input and output colour spaces, and then you can adjust exposure / WB / saturation / contrast and all sorts of other things in the same tool.  You don't even need to apply a film look at all...  just select the "Blank Slate" preset, which sets it to have no look at all, and you can still use all the tools to adjust the image without having to worry about colour management at all.
Any improvement in your post-processes is a retroactive upgrade to your camera, your lenses, and all the footage you have already shot.
Colour grading is such a deep art that I think the average GH5 user back in the day was probably extracting a third of the potential of the images they'd shot, if that, simply because they didn't know how to colour grade properly.  I'm not being nostalgic about the GH5 either, the same applies for any camera you can think of.
There are reasons to upgrade your camera, for sure, but most of the reasons people use aren't the right reasons, and they'd be better spent taking the several thousand dollars it would take for a camera upgrade and taking unpaid leave from their job and improving their colour grading skills instead.

Indeed!
Actually, the killer combo for the GF3, if we think of it like a tiny vintage film camera, is when it's paired with the Olympus 15mm F8 body cap lens.
It is truly tiny....


In a sense it's an incredibly synergistic pairing, because it gives a 30mm FOV, which is wide enough to make any micro-jitters pretty minimal (especially if you add gate weave in post) and it's sharp, so the softness is just limited by the GF3, and it's deep DOF which fits with the 8mm look.  Without an ND you're also using the shutter to expose, which I understand is also how 8mm cameras worked?
However, perhaps the killer aspect of it is the way you would use it.  You'd never use this as your main setup, so this would be a carry-everywhere low-stakes camera for having fun with.  It would be what you pull out when being silly with friends, or filming random things that aren't so formal.  In a way, that's how people might have used an 8mm camera back in the day, because they weren't inundated with video and didn't have the media savvy we all have now, so they would have just pointed their home movie cameras at whatever was happening.
It's even got a lever that closes it for use in pockets, but it also works as a manual focus adjustment and close focus is something like 30cm / 12 inches which might even get a little bit of background blur (I can't recall) so it's quite versatile.
The challenge is that the F8 aperture means it's basically no good after sunset or indoors, so that's the weakness.  Apart from that, this is perhaps the most likely setup I would use this with.  There's an F5.6 version from a different manufacturer that is tempting, but re-buying it for only one extra stop is a bit hard to swallow.
Anyway, here's a video I shot with this combo quite some time ago....
I can't remember how I graded that, but I think I used a film emulation plugin that added a lot of softening in post, so don't take that as the limits of its resolving power.  It also shows a lot of rolling shutter, so maybe the strategy would be to have it on a strap around your neck and pull that tight when shooting to stabilise the camera a little.
There is something about the extreme lack of technical performance that makes my brain think "well, this isn't going to win awards for literally anything, so ignore all the rules and just shoot and have fun!"

Tokina 28-85 F4 is a heavy baddy, if that's what you are looking for.:) I have never used mine on an actual shoot.
GF3, with that 12MP sensor from the GF1 days is a harsh higlights specialist camera. Respect for giving love to this novelty.:) @kye
Yes, the CZJ 40 has a nice rendering. I would love to try it on a S16 sensor size photo camera like the Nikon 1 series since it's a 16mm lens. But then I like the FOV of the 40 on mft. So I will keep enjoying it that way. Using and enjoying a lens is what it's about after all.:)
 

The comparison was great. I was thinking you focused on the leave behind the first one. Anyway, 2x crop in 4K mode looks like a great, even downsampled FHD image.
I have the Fuji 12.5 F1.4. Your Cosmicar looks like it's outresolving my lens quiet a bit. The Fuji is still a nice lens though with a very solid build and delivering a beautiful image.

Thanks! and yes, I also like those particular shots too.
Some time ago I invested in a M42-MFT speed booster and since then looked almost exclusively for M42 lenses, except for telephoto lenses where a speed booster isn't required.  Vintage FF lenses don't normally get wider than 28mm, at least the ones that don't cost much, and at 28mm the difference between an M42 lens with my SB and an FD lens (for example) without one is a 40mm FOV vs a 56mm FOV.  
I know I shot the above images without a SB, but mine really steers my buying habits towards that system.
I have now fully converted my setup to native AF MFT lenses (14-140mm, 12-35mm, 9mm, 14mm) so I now need to work out what I will use my MF and vintage lenses for.  When I shot those images the IBIS stabilised the image but not the flares, so the video files aren't really usable.  This means that if I want to shoot with very vintage lenses I need to shoot without IBIS and physically stabilise the camera, either going for a shaky image and embracing the aesthetic, or going for a more stable image and using a heavier setup / tripod / both.
Thinking about turning off IBIS and going for a more vintage look, my thoughts turned back to my GF3.  So I compared the softness of the image from my GF3 to the softness and grain of film, and depending on the amount of movement and detail in the image it's somewhere between being a Super-8 camera and a Super-16 camera.
I am still pondering this information, as I'm not really sure what I would shoot with a S16-like camera and vintage lenses, but I definitely feel some attraction to this concept.  
Also, there is something super-cool looking about this setup!

GF3 + SB + Tokina RMC 28-70mm F3.5-4.5 + vND...   giving a FF equivalent of 40-100mm F5-6.4.  This is quite similar to lots of S16 zooms back in the day too.  For example the S16 Meteor 5-1 17-69mm F1.9 lens is equivalent to 49-198mm F5.5, etc.
Part of my is very interested in finding a larger bulkier zoom and really leaning into the form-factor, but I couldn't find any around, and even if I did I'm not sure what they'd cost and if they'd be worth it to me (considering I don't even know what I'd use this for!)
This is just wonderful...  the trees have a painterly look that sort of makes them feel a bit hyper-real and a bit dream-like at the same time.  Great stuff!

I'd argue that this kind of testing is actually necessary to understand how things behave.  Over the years I have tested a lot of things and it's amazing how many things that "everyone knows" do not stand up in even the most basic tests, but continue to be myths because no-one bothers to even look.
Aristotle claimed that women have fewer teeth than men, which is not true, but he obviously never actually looked to see if he was right - despite being married multiple times where he could easily have tested his claim at any time.
No, not mixed up, but the 12-35mm has a shallower DOF and so you have to know where in the image to look to compare sharp details in the focal plane.
This is the unsharpened cropped image:

This is the 12-35mm image:

This is the sharpened cropped image:

The sharpening is perhaps a little over-correcting, but the thin edges are still slightly blurred in comparison to the proper image from the 12-35mm.
This is where it is important to know how to read the results of a test.  This comparison of the zoom to the crop matched FOV but not DOF, and while I probably could have zoomed in using the 12-35mm and also stopped down at the same time to keep DOF the same, the lens sharpness would have been reduced so it wouldn't have been a fair test.  To get around that I should have tested using a flat surface like a resolution chart or a brick wall.
The problem with going that route is that now we're no longer testing anything close to real-life, and no longer answering questions about what will and won't work in real shooting.
The test wasn't "what percentage of resolving power is lost using the CrZ function?"...  it was "is the CrZ function usable for shooting with cropped lenses?".  Realistically I shouldn't have included the 12-35mm optical zooms at all, I should have just cropped in using the CrZ function and left the images to be judged on their own merits in isolation, the same way that any project shot using the CrZ function would be.
This is the danger of pixel-peeing - it distracts from the only thing that actually matters - the image.
The cosmicar really is a gem!
There's a reason that cinematographers have relentlessly driven up the price of vintage lenses over the last decades, and why modern lens manufacturers are designing and releasing brand new lenses with vintage looks, and manufacturers are even creating new mechanisms to control the amount and type of vintage looks with custom de-tuning functions.

I love this

The elephant in the room is Resolve.
As I have discussed and demonstrated in my "New travel film-making setup and pipeline - I feel like the tech has finally come of age" thread, over the last decade Resolve has gotten more feature-rich, but more importantly, it's made it HUGELY easier to use and get good images.
People now have a lot more knowledge about colour grading tools and techniques, that's for sure, but things like the Film Look Creator enable you to use a single node, you set your input and output colour spaces, and then you can adjust exposure / WB / saturation / contrast and all sorts of other things in the same tool.  You don't even need to apply a film look at all...  just select the "Blank Slate" preset, which sets it to have no look at all, and you can still use all the tools to adjust the image without having to worry about colour management at all.
Any improvement in your post-processes is a retroactive upgrade to your camera, your lenses, and all the footage you have already shot.
Colour grading is such a deep art that I think the average GH5 user back in the day was probably extracting a third of the potential of the images they'd shot, if that, simply because they didn't know how to colour grade properly.  I'm not being nostalgic about the GH5 either, the same applies for any camera you can think of.
There are reasons to upgrade your camera, for sure, but most of the reasons people use aren't the right reasons, and they'd be better spent taking the several thousand dollars it would take for a camera upgrade and taking unpaid leave from their job and improving their colour grading skills instead.

You could do a similar testing like kye for banana leaves and textures in low light, when you enter your neighbor's garden at night.😊

On an unrelated note, aren't banana trees some of the prettiest trees there are? The shape of the leaves and the way they move in the wind is just an attractive thing. I have a friend who once sailed the South Pacific. He said when they would be at sea for days and would approach an island with banana groves that it was such a beautiful sight.
I'm in the only place in Canada warm enough to grow them ( and just barely too) so people plant them as decoration a lot. 

Went for a wander in the rain over the weekend with the GH7 and this lens:

I applied some filmic colours and a bit of grain, but the halation / bloom / softness / flares are all the lens.








Just remember, the less you pay for a lens, the more fun it is.... and this lens is a lot of fun.

The 9mm I tested is the Panasonic Leica 9mm F1.7, I'm not aware of a 9mm F1.4 - maybe you're thinking of the Leica 12mm F1.4?
Let me see if I can further tempt you!!
I have done some tests (images below) but found the following:
You can use the Crop-Zoom function (CrZ) to go up to 1.3x in C4K and up to 1.4x in 4K resolutions There is no 4K option in Prores, only C4K If the sensor was cropped to be a 1:1 readout, it would be a 1.4x in C4K and a 1.5x in 4K, but the CrZ mode stops just short of these amounts.  I suspect that they have limited it so that it is always downscaling, even if just slightly.
Test shots.
First set are with the S-16 Cosmicar 12.5mm F1.9 C-mount lens.  These are all on a 4K timeline, so you can really pixel-pee if you want to.  I didn't have quite enough vND to have it wide open on all the shots, so some are wide open but some are stopped down to F2.8.







Now, I switched from the 4K to the C4K, which meant I had slightly less crop available and you can just start to see the edges of the image circle.  I suspect your mileage would vary depending on what lens you were using.  The Cosmicar is pretty wide, so if you were using a long focal length you'd probably get no vignetting at all.
This should also give a comparison between the 4K H.264 and the C4K Prores HQ.

Now we switch lenses to the 12-35mm and stopped down to F5.6 so we can compare the CrZ crop to a non-cropped image.
This is cropped to 1.3x using the CrZ function in C4K Prores HQ:

and this is without any CrZ and using the 12-35mm to zoom in to match the FOV:

I didn't shoot any clips this morning comparing the CrZ mode in 1080p, but I can also shoot a test for this if you're curious.
I had a closer look and discovered you can't change the zoom amount, which seems to be stuck at 3x zoom.  I'd say that it is resolving enough for focusing and I used it with the Cosmicar in the above test.  It's the normal story of using peaking and rocking the focus back and forth to find the sharpest spot.
At least I'd say that if you can't use it to manually focus then the problem isn't the punch-in feature but some other issue!

Indeed it is, and indeed I do!
Haven't turned it on in.. well.. some time.  
I must admit I find it funny that my first video-first camera was the XC10 but moved on because I wanted shallower DoF, and now I'm back to shooting deep DoF with a 10x zoom lens.  This is why I never sell anything - I've lost count of the number of times I learn something new and then pull things out of the discard pile again, and although it's mostly lenses, you never really know.

I was just poking around in the menus and noticed there is an option where you can switch between Full and Pixel:Pixel, so that's the same as the ETC mode on the GH5.  It looks like you can use this with any resolution.
Also, you can record C4K in Prores RAW, which is a 1:1 sensor readout, so exactly a 1.41x, or a horizontal crop factor of 2.934 from FF (the GH7 horizontal crop factor is 36/17.3=2.0809).  The bitrates are a bit heavy though at either 1700Mbps or 1100Mbps and it's Prores RAW so you can't import it directly into Resolve and will need to transcode with a third party utility.
This thing has so many options, and the more I poke around in it, the more it feels like a cinema camera in the body of a MILC.

@PannySVHS I've now tested the Crop Zoom (CrZ) mode in 1080p.
This is the first test, and I exposed for the sky (which it thought was the right thing to do) which meant that the plants were a bit low, so I ended up bringing them up a little in post.  The Prores HQ is great at retaining noise and so there's quite a bit visible despite me having shot this at base ISO 500.  I've found that ETTR is definitely recommended if you want a more modern looking cleaner image.
I also used the 12-35mm lens at F4.0 for all images as that's where it's the sharpest.
First is comparing the C4K Prores HQ vs 1080p Prores HQ (on a 1080p timeline):


Next we compare the CrZ vs zooming with the lens.
I have prepared these images in sets of three.  The first is the CrZ image, the second is zooming with the lens, the third is the CrZ image again but with sharpening added.  This allows you to compare both CrZ images directly with the 'proper' one, as the more zoomed CrZ images did look a little soft in comparison when viewed at 300%.
Around 14mm (1.16x):



Around 18mm (1.5x):



Around 25mm (2.08x):



Once I got those images into Resolve and looked at them I decided to re-shoot it with a better exposure.  So I chose a different framing that meant the sky wasn't influencing anything.  However, I didn't realise that where I was standing was going in and out of the sun, so some shots were washed out and I had to compensate for it in post, adjusting contrast/sat/exposure/WB to match.  Tests are never perfect but are enough to give a good idea of what's going on, and in real use where there is no A/B comparing going on no-one would ever spot it anyway.
There's also a slight difference in exposure between the C4K and 1080p modes too, which is a bit odd.  I imagine it's due to changing the sensor mode.  I compensated for that in all these tests too.
C4K Prores HQ vs 1080p Prores HQ (on a 1080p timeline):


Around 14mm (1.16x):



Around 18mm (1.5x):



Around 25mm (2.08x):



I am actually rather encouraged by these results, as my previous test was in low-light and I did on something with much sharper edges and that showed differences I'm not really seeing here.
However, it's not really surprising that the GH7 did this well, as even with a CrZ of 2.08x it's still reading an area of the sensor around 2776 pixels wide.  I say "around" that wide because there is a slight crop when you compare the native 5.8K mode with the native C4K, 4K, and 1080p modes, but I think the 2.08x crop will still be oversampled from the sensor by a good amount.
The other thing I noticed was that I couldn't adjust the CrZ function while I was recording, the button just didn't do anything.  I'm not sure if that's because I have it assigned to a button and that there might be some other way to engage it while recording.  Maybe through the controls that are used to control powered zoom lenses, not sure.
Anyway, it looks pretty darn good to me, and the grain actually reminds me of the OG BM cameras which are quote noisy at native ISOs too (and also lots of seriously high-end cinema cameras too).

Simply fantastic images and I like the dark look. Though I'd change the 14-140 to M.Zuiko 12-100 f4 for that constant aperture, while losing tele but gaining on a wide end. Feels like 100mm on a M43 body should be enough though.

The 9mm I tested is the Panasonic Leica 9mm F1.7, I'm not aware of a 9mm F1.4 - maybe you're thinking of the Leica 12mm F1.4?
Let me see if I can further tempt you!!
I have done some tests (images below) but found the following:
You can use the Crop-Zoom function (CrZ) to go up to 1.3x in C4K and up to 1.4x in 4K resolutions There is no 4K option in Prores, only C4K If the sensor was cropped to be a 1:1 readout, it would be a 1.4x in C4K and a 1.5x in 4K, but the CrZ mode stops just short of these amounts.  I suspect that they have limited it so that it is always downscaling, even if just slightly.
Test shots.
First set are with the S-16 Cosmicar 12.5mm F1.9 C-mount lens.  These are all on a 4K timeline, so you can really pixel-pee if you want to.  I didn't have quite enough vND to have it wide open on all the shots, so some are wide open but some are stopped down to F2.8.







Now, I switched from the 4K to the C4K, which meant I had slightly less crop available and you can just start to see the edges of the image circle.  I suspect your mileage would vary depending on what lens you were using.  The Cosmicar is pretty wide, so if you were using a long focal length you'd probably get no vignetting at all.
This should also give a comparison between the 4K H.264 and the C4K Prores HQ.

Now we switch lenses to the 12-35mm and stopped down to F5.6 so we can compare the CrZ crop to a non-cropped image.
This is cropped to 1.3x using the CrZ function in C4K Prores HQ:

and this is without any CrZ and using the 12-35mm to zoom in to match the FOV:

I didn't shoot any clips this morning comparing the CrZ mode in 1080p, but I can also shoot a test for this if you're curious.
I had a closer look and discovered you can't change the zoom amount, which seems to be stuck at 3x zoom.  I'd say that it is resolving enough for focusing and I used it with the Cosmicar in the above test.  It's the normal story of using peaking and rocking the focus back and forth to find the sharpest spot.
At least I'd say that if you can't use it to manually focus then the problem isn't the punch-in feature but some other issue!

Sharpness seems very natural to me, although I am not at the level of pixel peeing as others around!
What I can say is that the Prores feels like Prores from a cinema camera.  So the files edit like butter, the grain is well captured and not removed / crunched, etc.  I've done quite a lot of low light high-ISO testing in the last few weeks and even up to ISO 12,800 the footage cleans up in post using temporal NR, which wouldn't work if the compression killed all the noise.
Punch-in focussing is available during recording, and pops up automatically if you touch the focus ring on a native lens, and has a custom amount of zoom.  I'd assume it's the same as previous cameras where you have an option to give you a punch-in box in the middle of the monitor, or for the whole monitor to show the punched-in part.  The focus peaking was also active within and outside the punched-in part of the screen.
The in-camera digital zoom is changed from previous models, and significantly improved at that.  It's quite different now.
Let's say I have my 9mm lens fitted.  I hit the button I have mapped it to (it's called Crop Zoom "CrZ") and it activates the feature, showing me the current focal length (9mm) and there are a bunch of ways to get it to smoothly zoom in and out, displaying the current equivalent focal length as it goes (10mm, then 11mm, etc).  
The function is integrated into the zoom controls for the powered zoom lenses too, so I think you can zoom in and it will zoom the lens in as much as it can and then (if enabled) it will keep zooming in with the digital zoom.
I thought the idea was it will keep zooming in until it gets to a 1:1 sensor read-out and then won't go any further, but the manual just lists some rather arbitrary zoom amounts.  With my 9mm lens, if I shoot with the C4K mode it will go to 11mm, but on the 1080p mode it will zoom in to 24mm.
In my tests I've found that the in-camera cropped images are free from artefacts, and I'd even zoom in/out during recording using it if I felt the need to.  
I'd happily use it for S16 cropping, or any other cropping you wanted.  Perhaps the only caveat is that if you wanted to crop more than the 1.3x it will do in C4K, or 1.4x in UHD, then you have to use the 1080p mode, and that mode seems to have a slightly different look to the images, a bit more like the OG BM cameras in that it looks like a lower-resolution sensor readout.  It's got a bit of that lower-resolution more sharpening look to it, rather than a higher-res-downscaled look to it.  It's subtle, but it's there.  It's still high-quality, but just compared to the 4K modes it's noticeable.
I've been doing lots of tests for my next ballooning trip, and these include low-light testing.  I figured I'd take my 14-140mm zoom for when the light is sufficient, and I'll take my new 9mm F1.7 as my ultra-wide, but was wondering if the 9mm could be my low-light non-wide lens as well.  
I did two tests.
The first test was an ultra-low-light test.  
I tested:
- GX85 with TTartisans 17mm F1.4 manual prime at F1.4
- GX85 with TTartisans 17mm F1.4 manual prime at F2
- GH7 with 9mm F1.7 at F1.7 (shot in C4K and cropped to be 17mm FOV in-post)
- GH7 with 9mm F1.7 at F1.7 (shot in 1080p and cropped in-camera to be 17mm FOV)
- GH7 with 12-35mm F2.8 at F2.8 and 17mm
- GH7 with 14-140mm at 17mm
- GH7 with TTartisans 17mm F1.4 manual prime at F1.4
- GH7 with TTartisans 17mm F1.4 manual prime at F2.0
- GH7 with Voigtlander 17.5mm F0.95 manual prime at F0.95
- GH7 with Voigtlander 17.5mm F0.95 manual prime at F1.4
- GH7 with Voigtlander 17.5mm F0.95 manual prime at F2.0
I reviewed all of them with just a 709 conversion, with NR/sharpening, and with tonnes of NR/sharpening.
This is a test of lots of things being traded-off against each other, as the slower lenses all needed a higher ISO, and the 9mm was sharp wide-open and brighter but also pulling from a smaller sensor area, but I didn't upload to YT so it's not a full pipeline test.
The result was that the Voigtlander won, the TTartisans at F2.0 was good, the 12-35 was good, but the 9mm was still acceptable and waaaaaay better than the GX85 + TTartisans wide open (which was what I shot the previous outing with and I found to be disappointing - the combo of the TTartisans at F1.4 combined with the GX85 ISO6400 was just a killer combo).  
I also tested the 9mm F1.7 wide-open vs the 12-35mm F2.8 stopped down to F4.0 against each other in good lighting and native ISO and using the 1080p in-camera zoom to match focal lengths.
I reviewed all of them with just a 709 conversion, with NR/sharpening, and with NR/sharpening put through my FLC pipeline (which includes softening the image slightly and adding grain).  I didn't upload it to YT either, so it's not a full-pipeline test but was a good indicator of it.
I found that the 9mm zoomed to 12mm was equivalent to the 12-35mm, at 18m it was noticeably softer, and at 24mm it was really noticeable and getting into vintage territory.  
I can post some stills if you're really curious.