Detail rendering in the G9ii (vs G9) - image heavy

[Aoyagi]

Intro

When it comes to the G9ii, most of the talk about it since the release (in 2023, to my shock) was related to dynamic range. Or the pixel-shift high res mode. A few months ago, there was a lively discussion about the G9ii and how the end result it gives isn't as detailed as what the predecessor can do. It's been bugging me since as I intend to use the G9ii for making my work and hobby slightly easier once its price drops like the S5ii's price did. That work (and/or hobby) is mostly stills with occasional video. Genres I do vary from corporate events, weddings, to kids/sports (indoor and outdoor) with some travel/landscape/nature mixed in as well, as per usual.

I saw the topic pop up a few times elsewhere as well and sometimes it me wonder if it isn't time to transition to the L Mount completely (I already use an S5 for work).

I do enjoy the advantages that MFT still has though. Sensor speed, features for the £, access to cheap and compact lenses. etc. I am sure most in this forum are aware of these. Naturally, I wanted to see the G9ii in action myself and I decide to rent it whilst visiting a mate and do some Super Scientific Laboratory™ tests myself, with the RAW developer of my choice.

===

Context

I should preface this by saying that (spoiler alert) whilst I did find some evident loss of detail and some other problems, I still do intend to get the camera. I don't notice any image quality difference outside of charts and lab (or "lab") tests. Most of my work gets posted on social media or printed in very small sizes. The improved autofocus alone is worth it in my eyes, not to mention what I consider a more typical Lumix layout (that on-off lever and front dial, for example)

Also, I had ended up having very little time (minutes), so I just got some household items, but it should be enough for a rough idea of the difference between the two cameras.

===

Test

Anyway, to the pixel peeping.

The images are processed using DxO PhotoLab 6 with absolute bare minimum of processing, i.e. everything turned off, or with PL's suggested corrections. Please disregard colours and such.

First I wanted to compare the cameras at their base ISOs, by shooting an electric swatter mesh at medium distance with the PL 25mm f1.4 at f2.8.

G9ii:



G9:



And here is the focus point comparison. Feel free to guess which is which, but you can tell by the EXIF data.



To my eyes the G9ii is a bit more "bloomy". However, if there is a lack of detail, it looks to me it comes down just to in-RAW sharpening.

...

Next I wanted to see how the cameras resolve detail at the same ISO (200). I couldn't find anything better than a supposedly English cheddar cheese of questionable taste:

G9ii:



G9:



And again, a comparison here:



I think this shows two things - the G9 is slightly better in resolving the tiniest detail because of the G9ii's blooming, and at the same time the G9ii is much more prone to moiré.

In case anyone is curious, here is a comparison with what I believe are PhotoLab's default corrections (and the image viewer's smoothing turned off):



...

Of course, I was also interested in high ISO behaviour. I forgot to step the lens down after someone else played with the camera, but it's enough to give me an idea of what happens at ISO 12800 regardless.

G9ii:



G9:



And the side-to-side:



Seems to me the G9 gives slightly more definition, especially looking at the Power O.I.S. text where there is some obvious fuzziness. Again, I blame the bloom. If there is baked-in noise reduction, I don't really see a significant impact even when engaging in some extreme pixel peeping.

At ISO 6400 and at around 300% zoom, I think the amount of detail is still very similar:



Unfortunately I was doing all this before I read about DR Boost working only until specific shutter speeds, so I didn't take that into consideration. I don't see how DR Boost could negatively affect resolvable detail, however.

I did take a test shot of our late queen's tea the next day at more respectable shutter speeds though. It was going against the G90 (which is my EDC), not the G9, but it does the job:



Again, mild fuzziness, mild blooming on the G9ii.

It was shot with the plastic fantastic 25mm f1.7 at f2.8 so I hope there was no focus shift at play.

===

Conclusions

To sum this whole exercise up, perhaps my standards aren't as high as many other photographers, perhaps I'm just blind, but as far as general resolving detail is concerned, the difference I see is within "statistical error". It is there, though. I honestly can't tell if this is because of what I keep calling "bloom" that the G9ii or because the G9ii bakes in NR into its RAWs, which is a practice and actively dislike and just don't understand at all. Either way, it seems to me the camera is going to be an upgrade over the G9 for basically everything I do. I suppose I will have to keep an eye out for any artefacts, blotchy shadows or undesirable effects, including in the pixel-shift high res mode.

However, from just a quick test I think the G9ii does a better job than the G9. Might be just more in-camera sharpening:



Hopefully someone will find this useful.

AA

--
Confrontation beats confirmation.

 

[realspeed]

I have both the G9 and the G9ii and added the firmware update in the G9ii.

Personally I prefer G9ii apart from better picture quality using same lens, I find negotiating the menu far easier although there is a lot more to wade through.

You will always find "pixel peepers" who find fault in anything instead of just enjoying photography.

A lot also depends on the lens used as well. I have the 12-60mm ELmarit leica lens which does most of what I want. Al;so the 100-300 mm panasonic lens for more reach.

Or put another way the G9ii is now my "go to" camera and the G9 is the back up.

 

[j tokarz]

I have the G9 and am more than happy with the outcome. MORE than happy. If i havn't anything to compare its outcome with, theres absolutely no reason to upgrade. Enjoy your new friend. G9ii.

 

[Aoyagi]

Time to update the firmware again I guess, haha.

The problem with pixel peepers is that they can blow some shortcomings out of proportion. I really just wanted to see if that is the case with the G9ii.

What makes the G9ii's menu easier for you? The additional layer of foldering/categories?

 

[Lepewhi]

I agree. I have a g9 and am very happy with it. Maybe the G9ii is better. But is it 3 times better, comparing used prices. And with pp, images can often be cleaned up.

 

[mmartel]

Thanks for sharing.

My conclusion based on what I say in your test images is the same as yours — difference too small to make a practical difference except for all but the finest of fine details, and even then you have to be squinting way in.

I guess different sensors do different things, and higher megapixel sensors within the same format may not necessarily give you what you might expect — possibly due to other image processing stack differences (e.g., noise reduction as you speculated, microlens differences, CFA differences, DxO RAW processing software defaults and fine tunings, etc.).

Anyway, I'd certainly be in the same camp as you as far as conclusions go: If you find enough value in the added features of the G9ii, it's a clear winner — assuming price is not that dear to you.

That said, as a longtime G95 owner who added a G9 in June 2024, I do really like the long-in-the-tooth OG G9 quite a bit and I think it's value for money is amazing on the secondary market. As a hybrid video and stills shooter, I have every plan of keeping it as my B cam of choice behind my new favorite A cam (EM1X).

As a side note, I also find the "legendary G9 ergonomics" overblown, favoring the G95/G9ii control setup (better front dial, power switch, and video record button placement) and I can easily see going back and forth between the G95 and G9ii to be more intuitive than the G9 and G95.

Best wishes with your new gear!

 

[Jens_ac]

Hi

There is no support for the idea that there is any noise reduction in RAW in the G9m2. Any noise reduction was applied by you or the software you used.

Since the G9m2 offers higher resolution than the g9 and you used same lens for both but find the Image quality of G9 to be higher, it is clear, that you made a mistake.

What did you do which may have caused your strange conclusion?

 

[Aoyagi]

As a side note, I also find the "legendary G9 ergonomics" overblown, favoring the G95/G9ii control setup

I am definitely in the same camp. the G90 (or... G95/G95D/G97/G91/G99/G99ii) specifically fits my hand so well that it feels like it was moulded by it, haha.

 

[Aoyagi]

What suggests NR being applied is the slight mushiness in high ISOs in all G9ii RAWs I've ever seen.

Noise reduction was disabled for all the test shots, as was any sort of correction (besides that one comparison). More importantly, the developing conditions were identical.

 

[Jens_ac]

What suggests NR being applied is the slight mushiness in high ISOs in all G9ii RAWs I've ever seen.

Noise reduction was disabled for all the test shots, as was any sort of correction (besides that one comparison). More importantly, the developing conditions were identical.

If you can see mushiness then what you see is not RAW anymore but an image after the de-bayer/demosaic process and also other processes probably have been applied. In that case the software produced such mushiness and you should try to change the denoise settings or use other denoise software until you like the image. The camera itself does not apply any noise reduction in RAW images according to experts who investigated that. NR in RAW is a rumor. Different software gives different results depending on camera model and parameters given.

 

[Aoyagi]

If you can see mushiness then what you see is not RAW anymore but an image after the de-bayer/demosaic process and also other processes probably have been applied. In that case the software produced such mushiness and you should try to change the denoise settings or use other denoise software until you like the image.

As I keep repeating, there was no NR applied by the RAW developer. None. It wasn't just set to zero, it was completely disabled.

I am getting pretty much the same result with other RAW developers as well.

The camera itself does not apply any noise reduction in RAW images according to experts who investigated that. NR in RAW is a rumor. Different software gives different results depending on camera model and parameters given.

Would you mind sharing this investigation? I hope it also explains why the G9ii resolves ever so slightly less detail than the G9 under the same conditions as well, despite having a higher resolution, albeit marginally.

 

[Palombasso]

Hi, thanks for the effort and comparisions. As a G9 owner I'm interested in the results (although if I upgrade to G9ii it would be mostly because of the PDAF).

As the cameras are of different resolutions, shouldn't the 1:1 side by side comparisons show different "sizes" on the objects?

Could the blooming is an effect of some kind of resizing on the viewer software?

 

[Jens_ac]

If you can see mushiness then what you see is not RAW anymore but an image after the de-bayer/demosaic process and also other processes probably have been applied. In that case the software produced such mushiness and you should try to change the denoise settings or use other denoise software until you like the image.

As I keep repeating, there was no NR applied by the RAW developer. None. It wasn't just set to zero, it was completely disabled.

I am getting pretty much the same result with other RAW developers as well.

So if you are sure, that it is not the software, then you should take a look if something else coused the error - wrong focusing, shaking, vibrations, any kind of error on your side...

The camera itself does not apply any noise reduction in RAW images according to experts who investigated that. NR in RAW is a rumor. Different software gives different results depending on camera model and parameters given.

Would you mind sharing this investigation? I hope it also explains why the G9ii resolves ever so slightly less detail than the G9 under the same conditions as well, despite having a higher resolution, albeit marginally.

Usually the person who claims, that a camera does noise reduction or sharpening in its internal RAW pipeline should back up such claim with reproducible observations. It is a bold claim.

Here it (the noise in the RAW) was investigated and there was no noise reduction effect found in G9M2:

https://www.dpreview.com/forums/post/67856708

Thus: Your observation should be checked again and/or needs other explanation. Do I see some strange kind of ca (chromatic) in your G9M2 images? Those clear red and blue dots not present in the G9 image? Also other things look a bit strange in your images. So most likely you just did something wrong with those images. This is totaly ok since the camera is new for you and it may take some time to get the workflow adapted to your needs - just like with any new camera. Anyway such is your normal learning curve and nothing to blame the camera about. Such blaming may confuse other readers here and that would not be nice for them.

Best regards,

Jens

 

[Aoyagi]

As the cameras are of different resolutions, shouldn't the 1:1 side by side comparisons show different "sizes" on the objects?

You're right, I also should have mentioned that I did a 1:1 comparison. Looking at 100% with cameras of different resolution whilst comparing them doesn't make a lot of sense to me. Differences are a lot less obvious then, if there are any. So I did a 100% on the G9 and roughly matched the G9ii.

Could the blooming is an effect of some kind of resizing on the viewer software?

These are JPGs, so viewer software (it's Faststone - highly recommended!) is extremely unlikely to cause anything like that. You can download the source photos yourself and try in any software of your choice!

 

[Aoyagi]

So if you are sure, that it is not the software, then you should take a look if something else coused the error - wrong focusing, shaking, vibrations, any kind of error on your side...

The camera was firmly planted on a solid glass desk, I did 10 shots with AF-S refocusing on the same spot each time.

Usually the person who claims, that a camera does noise reduction or sharpening in its internal RAW pipeline should back up such claim with reproducible observations. It is a bold claim.

Well, I am quite sure I did that.

Here it (the noise in the RAW) was investigated and there was no noise reduction effect found in G9M2:

https://www.dpreview.com/forums/post/67856708

I saw Serguei's thread. His own graph suggests that NR or something with the same/similar effect is taking place from ISO 3200 including. To my knowledge, nobody said that it starts happening after ISO 3200. The word is that it happens at ISO 3200, and after. Serguei's graphs support this. Comparing ISO 3200 to 4000 doesn't make a lot of sense then.

I should add that this is just one expert, and his data doesn't support the conclusions. You have multiple other experts saying that the G9ii does apply slight NR. I've been laughing at Sony for doing this for years so I am not particularly happy about it, but it's subtle enough.

Thus: Your observation should be checked again and/or needs other explanation. Do I see some strange kind of ca (chromatic) in your G9M2 images? Those clear red and blue dots not present in the G9 image? Also other things look a bit strange in your images. So most likely you just did something wrong with those images. This is totaly ok since the camera is new for you and it may take some time to get the workflow adapted to your needs - just like with any new camera. Anyway such is your normal learning curve and nothing to blame the camera about. Such blaming may confuse other readers here and that would not be nice for them.

Right.

I took test shots with each camera from the same position on a very solid glass table. Ten shots for each scene using AF-S (refocused every time, obviously) to get the best focus, manual mode, mechanical shutter, IBIS on, highres off except for that one specific high res test, and a non-log or HLG picture profile. The shutter speeds weren't slow enough to activate long exposure NR. There are no other RAW-relevant settings that I can think of in either camera. The lighting was the same, the lens was the same, the exposure triangle values were the same.

I then took the RAWs from both cameras, transferred them to a PC using a card reader, not some app which might alter them.

In that computer I opened them all (not imported, opened) in PhotoLab (which fully supports both cameras), disabled all corrections, colour rendering, sharpening, noise reduction, or exposure adjustments. PL has a very easy way to see what's enabled and nothing but the colour working space (which naturally cannot be disabled) was enabled.

(By the way, I've also tried Lightroom - again. Same results, only with much more pain disabling everything.)

Then I exported the RAWs to JPGs at the full quality, no cropping or any other processing, with the "as shot" ICC profile (default).

Picked the best out of each series.

I did the exact same thing for both cameras.

Feel free to tell me where in this process I made a mistake. It's really quite straightforward.

 

[Jens_ac]

So if you are sure, that it is not the software, then you should take a look if something else coused the error - wrong focusing, shaking, vibrations, any kind of error on your side...

The camera was firmly planted on a solid glass desk, I did 10 shots with AF-S refocusing on the same spot each time.

Usually the person who claims, that a camera does noise reduction or sharpening in its internal RAW pipeline should back up such claim with reproducible observations. It is a bold claim.

Well, I am quite sure I did that.

Here it (the noise in the RAW) was investigated and there was no noise reduction effect found in G9M2:

https://www.dpreview.com/forums/post/67856708

I saw Serguei's thread. His own graph suggests that NR or something with the same/similar effect is taking place from ISO 3200 including. To my knowledge, nobody said that it starts happening after ISO 3200. The word is that it happens at ISO 3200, and after. Serguei's graphs support this. Comparing ISO 3200 to 4000 doesn't make a lot of sense then.

I should add that this is just one expert, and his data doesn't support the conclusions. You have multiple other experts saying that the G9ii does apply slight NR. I've been laughing at Sony for doing this for years so I am not particularly happy about it, but it's subtle enough.

Thus: Your observation should be checked again and/or needs other explanation. Do I see some strange kind of ca (chromatic) in your G9M2 images? Those clear red and blue dots not present in the G9 image? Also other things look a bit strange in your images. So most likely you just did something wrong with those images. This is totaly ok since the camera is new for you and it may take some time to get the workflow adapted to your needs - just like with any new camera. Anyway such is your normal learning curve and nothing to blame the camera about. Such blaming may confuse other readers here and that would not be nice for them.

Right.

I took test shots with each camera from the same position on a very solid glass table. Ten shots for each scene using AF-S (refocused every time, obviously) to get the best focus, manual mode, mechanical shutter, IBIS on, highres off except for that one specific high res test, and a non-log or HLG picture profile. The shutter speeds weren't slow enough to activate long exposure NR. There are no other RAW-relevant settings that I can think of in either camera. The lighting was the same, the lens was the same, the exposure triangle values were the same.

I then took the RAWs from both cameras, transferred them to a PC using a card reader, not some app which might alter them.

In that computer I opened them all (not imported, opened) in PhotoLab (which fully supports both cameras), disabled all corrections, colour rendering, sharpening, noise reduction, or exposure adjustments. PL has a very easy way to see what's enabled and nothing but the colour working space (which naturally cannot be disabled) was enabled.

(By the way, I've also tried Lightroom - again. Same results, only with much more pain disabling everything.)

Then I exported the RAWs to JPGs at the full quality, no cropping or any other processing, with the "as shot" ICC profile (default).

Picked the best out of each series.

I did the exact same thing for both cameras.

Feel free to tell me where in this process I made a mistake. It's really quite straightforward.

 

[Aoyagi]

Maybe for your workflow and your image taste the G9 is better - for many others the G9m2 is better, which may be not relevant for you. But with the NR in RAW with the G9m2 you are wrong, same as you were wrong assuming in jpg settings NR=0 was minimum when it actually is NR=-5.

It is your current situation in the learning curve for the optimum workflow and not the camera to blame.

If you need any help please ask. If you just want to influence people I am out - boring since I am fine.

Maybe you are new to Lumix cameras, but the noise reduction setting in them, like any other setting for the photo style, applies directly only to out-of-camera JPGs. It has no effect on RAWs besides embedding the photo style information in it for when you want ot develop the raw with the settings used in the camera (if the RAW developer supports it).

When I was talking about setting NR to zero, I was, obviously, talking about the RAW developer, not the irrelevant in-camera setting.

If you need to learn more about what settings affect RAWs, please refer to my previous comment or make a new thread. I and many others will happily assist.

As you were unable to respond to a single thing from my previous comment, let alone point out any mistake I might have made that's not made up, I am kindly going to suggest to stop blaming the user when there is ample evidence it's the camera. Such blaming may confuse other readers here and that would not be nice for them.

That said, I'm not blaming the camera. Although there is some marginal difference in detail rendering (plus some other subtle differences), the image quality it provides is great and I am looking forward to working with it more.

Good day!

 

[Wboz88]

OP, i know your sole focus here was details, and no objections to that. But, one thing I started noticing as soon as I started glancing at G9ii photos after its release is that the images have a distinctive, for lack of a better word, "depth" or "three dimensionality" that earlier Lumix cameras lack.

I am not sure if this is down solely to Lumix choosing to output higher contrast images than before, or a more complex set of changes stemming from the new sensor. I realize that subjectively people can have different opinions, but the 9ii images pop in a pleasing way to me.

I hope to one day have Lumix put that sensor in something smaller, at least no larger than the G95, but if not then maybe quite a number of years in the future I will grab a well used G9ii for use when the size is not an issue. I don't really feel that way about anything Lumix made between the GX85 I have and the current flagship, which is to say, I do feel the quality of the G9ii images took a significant step forward.

Thank you for sharing your findings so we can see

 

[dukeblue91]

I just purchased a used G9 last November and coming from the G7 I'm still learning the in's and out's of the G9.

So far I'm in love with the camera and it has so many features and settings that I'm at times overwhelmed.

The image quality is outstanding especially compared to the G7 which my son is using now and really likes.

I'm in no way looking to upgrade anytime soon as far as the body is concerned.

As for glass well that's a different story, I'm looking to get the PL 100-400 next.

Just my .2 cent.

 

[Jens_ac]

Maybe for your workflow and your image taste the G9 is better - for many others the G9m2 is better, which may be not relevant for you. But with the NR in RAW with the G9m2 you are wrong, same as you were wrong assuming in jpg settings NR=0 was minimum when it actually is NR=-5.

It is your current situation in the learning curve for the optimum workflow and not the camera to blame.

If you need any help please ask. If you just want to influence people I am out - boring since I am fine.

Maybe you are new to Lumix cameras, but the noise reduction setting in them, like any other setting for the photo style, applies directly only to out-of-camera JPGs. It has no effect on RAWs besides embedding the photo style information in it for when you want ot develop the raw with the settings used in the camera (if the RAW developer supports it).

When I was talking about setting NR to zero, I was, obviously, talking about the RAW developer, not the irrelevant in-camera setting.

If you need to learn more about what settings affect RAWs, please refer to my previous comment or make a new thread. I and many others will happily assist.

As you were unable to respond to a single thing from my previous comment, let alone point out any mistake I might have made that's not made up, I am kindly going to suggest to stop blaming the user when there is ample evidence it's the camera. Such blaming may confuse other readers here and that would not be nice for them.

That said, I'm not blaming the camera. Although there is some marginal difference in detail rendering (plus some other subtle differences), the image quality it provides is great and I am looking forward to working with it more.

Good day!

So it was not you complaining about jpg noise reduction in the other thread? Then sorry for my error.

In RAW there is no noise reduction I can find - instead I can reduced noise a lot in post using RAW from the G9m2, which would not work so nicely in case noise was reduced already in camera.

Bye.