Comparing G9ii vs G9 in low-light and dynamic range

[dp_oli]

I’ve been using the Panasonic G9 for about four years and so far I’ve really had great time shooting this camera. I mostly shoot travelling, landscapes (also night-scapes) and occasionally macro. The G9 IQ is in my opinion outstanding, even by today’s measures. The latitude one has with processing the files is not something I would have expected from a MFT camera when I bought into the system. However, with turning towards more video recently, but also with the hope of getting some of the benefits of the 7-years younger successor (autofocus! live composite! VLOG!), I decided to order a G9II.

Given the many positive reviews, I was convinced that the G9II would likely outperform the original G9 in pretty much all fields. I also came across the saga of the Dual-gain sensor that uses some half-documented magic sauce to boost DR and lower noise levels in the new 25MP sensor. I didn’t find conclusive information on how that would affect IQ under which circumstances. Many tests compare it to the OM-1, which I don’t own, but never to the original G9. So I decided to do a few test right after the arrival of the G9II.

The G9II reads out the sensor through two different gain channels to recover more shadow detail. The details are explained in the manual and in many other places here on the forum. Despite finding all that very interesting, at the end of the day I just want the thing to work when I’m out there. :-D

The G9II automatically activates DR boots for shutter speeds faster than 1/15s. However, as many have found out, my tests confirm that at least for ISO 200 and below, DR boost is active for everything faster than 1s. But doing long time exposures quite often, I still found this a severe limitation going into the purchase and was curious to see how that would affect IQ. To be clear: I didn't expect miracles, but my premise was that I didn't want to have situations with worse IQ vs. the camera I already own. If DR boost can lift IQ in the many scenarios with faster shutter speeds, that would be great enough!

Tldr: I will send the G9II back, because even if the peak IQ is better with the right exposure parameters for DR-boost to fully kick in, there is just not enough consistency to justify the (expensive) switch. More often, the results of the G9II are disappointing compared to the 7 years older G9. This is of course less valid for other people. If you have a different shooting style, you might never have a problem with the G9II’s shortcomings and only see the numerous advantages!

About the comparisons

All tests were performed with an Olympus 12-40 f2.8 lens. All the shots were taken in manual mode with the exposure settings being equal between both cameras unless stated otherwise.

I (always) shoot in RAW and process in ACR. So I didn’t test JPG quality at all. All photos were edited to push the darks and reveal as much DR as possible. Not that this would always be needed in real life, but this is what this test is about.

Forgive me, the subjects are in no way great arrangements. It’s a basement and a guest room full of clutter. I would much rather shoot actual stuff, but with the logistics of a comparison and running back to the PC, I found these to be the easiest way to set up consistent scenarios for what I wanted to test. I think one can still draw conclusions on things like DR or low-light performance. If you think that I am basing my verdicts on an imprecise or unfair test setup, please let me know.

TEST 1: DR-Boost and its effect

The first comparison shows a scene with high DR and compares the G9II to itself. Both shots are taken at ISO 100. At 1s, the first shot is out of the DR boost territory, the second shot with 1/8s is in.


G9II - DR boost off


G9II - DR boost on

Let’s start with the good: The shot with DR boost appears very clean in the shadows and shows a lot of detail. I can push the darks very far and still recover detail.

The bad: The picture without DR boost is pretty sad. It lacks a lot of detail in the shadows but also in the brights outside the window. The insect web is not resolved at all. I will let the details speak for itself.


G9II - DR boost off


G9II - DR boost on


G9II - DR boost off


G9II - DR boost on

How does it clean up with Adobe AI denoise? Better than expected, but still notably worse. The blown out highlights are of course beyond recovery, so no change there.


G9II - DR boost off, AI denoise


G9II - DR boost on, AI denoise

 

[dp_oli]

TEST 2: G9 vs. G9II

Sure, we found out the obvious: DR boost brings an advantage. I should add that the test above should be seen for what it is: going to f/16 for the 1s shot definitely drives the Olympus 12-40 way beyond it’s sweet spot for diffraction, so don’t judge the resolution. Judge the noise and the ability to resolve the dynamics of the scene. To me, it was clear to see that IQ suffered severely without DR Boost.

The following test brings the old G9 in the mix. Now I set the G9II also to ISO 200, which is the base ISO of the G9. Same scene, slightly different exposure because I put on an ND filter to get within a more normal aperture range. Development settings were identical, besides the G9 files needing a tad more highlight reduction for ACR to show texture in the window. But it got it nicely restored. Since you know the fancy “test scene” by now, I will focus on the highlights and the shadow details.

Shadows:


G9II - DR boost on


G9II - DR boost off

The G9 with equivalent settings:


G9 - equivalent settings to "DR boost on"


G9 - equivalent settings to "DR boost off"

I think we can all agree that the G9 without DR boost is way behind and very noisy. Besides that, I don't see a tremendous difference between the G9 and the G9II with DR boost. There's a little color shift, and I tend to say that the G9II with DR boost resolves the color a bit better/cleaner than the G9.

Highlights:


G9II - DR boost off


G9II - DR boost on

The G9 with equivalent settings:


G9 - equivalent settings to "DR boost off"


G9 - equivalent settings to "DR boost on"

While all settings yield an acceptable result and the difference in the highlights is not as severe, I find the G9II with DR boost off to be the worst again. If you take a look at the sticker on the window shades, it becomes quite obvious. The G9 also resolves the insect web much better, which might stem from a more fortunate ratio between the lower pixel count and the grid size of the web.

 

[jalywol]

Unfortunately, if you are looking for across the board improvements in deep shadows and high DR shooting, without electronic gymnastics to get you there, you will have to consider a larger sensored system.

I shoot both M43 and FF. I briefly had a G9II, and it definitely did have more information in the shadows than the 20MP Panasonic bodies, but, as you discovered, it still has its limits.

For situations where I know I am going to be pushing those limits with M43, the FF really steps up. The shooting envelope is just a lot wider, overall, and the noise floor is quite different than with M43.

I have both and S5 and an S9, and maybe you would want to consider adding in an S9 for your video work and some of your more demanding lighting situations for photography? I am quite pleased with mine (and I use it only for photography, as, even though it's aimed at the video content creator market, I don't do video, but I do like smaller bodied cameras, so... )

I still use my M43 for a lot of things, but having the FF for difficultly lit environments is really a bonus....

-J

 

[dp_oli]

TEST 3: Low light

DR is one thing, but not all scenes require maximum ability to resolve a high DR. However, low light is challenging, especially for MFT. That's why I wanted to see how the cameras compare at higher ISO settings and went into the basement. There, things got even weirder.

I took a nice scene of some basement clutter with identical exposure values across a range of ISO settings. Probably all steps of this ladder were beyond the cutoff curve for the G9II's DR boost which we only know of that it runs somewhere between 0.8s at ISO 200 and 1/15s at high ISOs. So the basement scene was not in favor of this curve, but neither is real life at sunset! ;-)

Scene:



Side note: it seems that the G9II doesn't bake the lens correction profile into the raw file the same way the G9 did. I can select it manually in ACR. Maybe there's also a camera setting I forgot.

ISO 200


G9II


G9

ISO 1600:


G9II


G9

ISO 6400


G9II


G9

This was when I deciced that there would be no way I'd keep the G9II. Sure, the G9's image is noisy, but it comes out pretty clean and the textures are not wishy-washy. Especially IDO 6400 is shocking to me. The G9II seems to invent detail where there is none. Look at the smeary stuff above the text on the carboard box or the area between the two vents on the humidifier. Even at ISO 200, there appears to be more detail in the G9 image, for example in the silver labels.

TEST 4: High-resolution in "good light"

What I mean by good light is that the exposure values allow DR boost to be enabled (as long as the cutoff is the same in the high-res mode). Let's go back to the first scene and look at some details.


G9II - HR


G9 - HR

What the...? Despite having the same exposure values as in test 1, we now have drastically clipped highlights in the G9II. Otherwise, the details are fine, but the G9II appears pretty jagged around the edges. I should clarify that the G9II was set to handheld HR mode. I don't know wheter that matters, but handheld mode was an upgrade feature I was excited about so I think it's fair to test that.


G9II - HR


G9 - HR

This might be the only result where the G9II really has an edge over the G9. While the G9 exposes the typical purple shadow noise, the G9II stays pretty neutral. The detail on the label looks also better.

TEST 5: High-resolution in "bad light"

I also did a high-res comparison with the basement scene. The exposure values indicate that DR boost must be off this time, but who knows - maybe averaging the noise levels over multiple frames can help out? Nope, the results were underwhelming again. :-|


G9II - HR


G9 - HR

While the result is not as devastating as the single-shot performance with the same settings (see test 3 for ISO 1600), it's still not able to catch up to the G9. In the brighter areas, I'd be willing to say that both are fine, but the shadows are really underwhelming. Here's another detail of a better-lit area.


G9II - HR


G9 - HR

SUMMARY

So what's the conclusion?

• With my photography interests, the G9II appears to be a half-baked product that - under perfect conditions - might have an edge, but is overall somewhat quirky and unreliable in it's consistency across the board of exposures.
• When dropping out of the DR boost advantage, things go really south compared to the G9. I assume the OM-1 which got a lot of praise will also fare better there and bring the other advantages I'm looking for.
• I don't dispute the overall awesomeness of the camera. I quickly noticed that a lot has improved since the G9 regarding handling, AF and feature richness. It's a camera I would love to love. But the IQ shortcomings are a dealbreaker. Now, one could argue that I'm in the wrong system if I am looking for IQ, but the old G9 shows that a lot is possible and sensor development didn't stop there. Panasonic just made a tradeoff that isn't for me, but presumably more for video focused creators.
• I'm sure one can work around some of these noise issues by exposing a bit differently, or stacking multiple shots. But I'm honestly not interested in doing all that mental work when the original G9 just delivered consistent and expectable results under all conditions.

Did I forget something? Can you reproduce the results? Let me know!

Oliver

 

[dp_oli]

Unfortunately, if you are looking for across the board improvements in deep shadows and high DR shooting, without electronic gymnastics to get you there, you will have to consider a larger sensored system.

I shoot both M43 and FF. I briefly had a G9II, and it definitely did have more information in the shadows than the 20MP Panasonic bodies, but, as you discovered, it still has its limits.

For situations where I know I am going to be pushing those limits with M43, the FF really steps up. The shooting envelope is just a lot wider, overall, and the noise floor is quite different than with M43.

I have both and S5 and an S9, and maybe you would want to consider adding in an S9 for your video work and some of your more demanding lighting situations for photography? I am quite pleased with mine (and I use it only for photography, as, even though it's aimed at the video content creator market, I don't do video, but I do like smaller bodied cameras, so... )

I still use my M43 for a lot of things, but having the FF for difficultly lit environments is really a bonus....

-J

Thanks for the reply. I don't feel like having a FF vs MFT discussion now, because there are millions of those out there and it's misleading for me. The facts are:

• I don't want to afford and carry two bodies (even less so two mount systems) so one has to be enough.
• I do believe that MFT is suitable for me because lenses are way smaller, even if the body is not. I have big hands and that won't change.
• What I wanted to have is a G9II that has all the updated features (which I find hard to get in a better package than in current flagship MFT bodies) while maintaining same or better photo IQ than the old G9. That would be surely possible, but isn't what the real G9II turned out to be.

 

[kalisti]

Unfortunately, if you are looking for across the board improvements in deep shadows and high DR shooting, without electronic gymnastics to get you there, you will have to consider a larger sensored system.

I shoot both M43 and FF. I briefly had a G9II, and it definitely did have more information in the shadows than the 20MP Panasonic bodies, but, as you discovered, it still has its limits.

For situations where I know I am going to be pushing those limits with M43, the FF really steps up. The shooting envelope is just a lot wider, overall, and the noise floor is quite different than with M43.

I have both and S5 and an S9, and maybe you would want to consider adding in an S9 for your video work and some of your more demanding lighting situations for photography? I am quite pleased with mine (and I use it only for photography, as, even though it's aimed at the video content creator market, I don't do video, but I do like smaller bodied cameras, so... )

I still use my M43 for a lot of things, but having the FF for difficultly lit environments is really a bonus....

-J

Thanks for the reply. I don't feel like having a FF vs MFT discussion now, because there are millions of those out there and it's misleading for me. The facts are:

• I don't want to afford and carry two bodies (even less so two mount systems) so one has to be enough.
• I do believe that MFT is suitable for me because lenses are way smaller, even if the body is not. I have big hands and that won't change.
• What I wanted to have is a G9II that has all the updated features (which I find hard to get in a better package than in current flagship MFT bodies) while maintaining same or better photo IQ than the old G9. That would be surely possible, but isn't what the real G9II turned out to be.

For video the sigma f/1.4 trio on the G9 dont do the wobbly AF thing, not much use if you need a zoom, could extend your G9's life perhaps?

 

[mmartel]

Just wanted to say thank you for the testing and sharing your observations. It can be tough to be objective and, especially, to be critical when you have high hopes for newly-acquired gear. But it's really valuable. So thanks.

As a hybrid video + stills shooter myself, and someone who very recently decided to consolidate my gear on two G9s (subtracting my G95 and EM1X, adding to my existing G9), I've gone around and around on whether to roll the dice on a G9ii.

Even as old as it is now, the OG G9 still offers a solid mix of quality, performance, and features for many users and use cases. It's not perfect and, yes, many cameras do many things better. But if it reliably does what you need it to do, and you don't mind the size of it (I sure don't), I think it's still a very good choice. Maybe, perhaps, it's still one of the best values in the m43 format — especially if you're a hybrid video + stills shooter.

 

[dp_oli]

Thanks! I wrote it up partially for myself to avoid jumping to a conclusion too quickly. A folder full of pictures can get quite messy so writing it down helps getting the thoughts in line. But if it helps others to make or confirm a buying decision, that's great!

I agree that the G9 is not perfect but close. I just wish they kept the G9II more photo-centric and avoided the quirks this thread is about. Because some of the new features are quite nice and I'd say the product progressed quite well in other ways. I will probably stick with the G9 for now. The OM-1 seems nice, but I think even just the position of the power switch is something that would drive me nuts

 

[dp_oli]

I thought I owe the G9II at least one outing before sending it back and I also wanted to end this thread with some more real-life photos so I went for a little walk at night.

I used the G9 with one of those fun Chinese 35mm f/0.95 lenses, stepped down to f/4. The Olympus 12-40 on the G9II was zoomed in to 35mm and stepped down to f/4 too. As you can guess, the Olympus produces a much flatter image, but at least in the center the results should be comparable.

As note at the beginning: I also tried to do HHHR shots with the G9II of the first motive in the respective mode. It failed 4 in 5 times to process the image, and the one successful example was blurry.


G9II, no NR


G9, no NR

Once again, the G9II produces underwhelming results with pretty noisy and washed-out details.


G9II, no NR


G9, no NR

The small letters on the vessel are almost legible in the G9 photo, but much less so in the G9II.


G9II, no NR


G9, no NR

It seems Panasonic saw the need to bake a lot of NR into the file, which softens the details even if the sky seems a bit less noisy.

After all this, the G9II surprised me on a different shot.


G9II, no NR


G9, no NR


G9II, no NR


G9, no NR

Now we have a ton more detail. The picture from the G9II looks almost perfectly clean.

Do I have an explanation for this? Absolutely not. It would require more tests. The first scene was shot with ISO 2000, the second one with ISO 3200. Maybe there's something about the G9II's algorithms that prefers typical full-stop ISO values? We don't know. The last example shows that there is a lot of potential to squeeze out of the new sensor, but maybe it's worth waiting for another iteration on the design, just as we've seen a big leap forward from the GH6.

 

[bobnn]

The OM-1 seems nice, but I think even just the position of the power switch is something that would drive me nuts

You can almost certainly re-assign the power switch function to to function lever that surrounds the AEL button.

 

[BryanHL]

I am not in the habit of trying to get good shots in dark conditions. Maybe if I acquire a FF camera I will, or a flash.

I am more interested in learning how the two G9's compare with AF, subject acquisition and tracking - in normal light conditions not edge case low light situations.

 

[Aoyagi]

In other words, if you have the G9 and do a lot of long exposure photography, the G9ii probably isn't for you, right?

 

[realspeed]

Don't you just hate nick pickers who try their best to prove something in circumstances a camera is not designed to do. M43 cameras are known to be not good in low light so why try and prove otherwise. THe OP has failed to mention the benefits of the G9ii over the G9 ( I have both). No mention of better image stablisation for one example.

 

[Jim Cox]

I was thinking about buying the G9ii but after reading about the problems of the dual gain turning off I decided not to get it. The problem is most pronounced at low ISO settings on long exposures. Here is a link to one of the forum sites but it is quite technical.

https://www.dpreview.com/forums/thread/4771660

At 200ASA without the dual gain, such as taking a long exposure to blur water with nd filters, it appears to lose quite a bit of dynamic range. I take a number of pictures like that. I also take Northern lights or moonlit shots. The moon lit shots I often use 200 ASA and the Northern lights 800 ASA with several second exposures. The G9ii seems to be similar to the G9 if you use 3200 ASA or higher but it appears to be very bad at the lower ASAs. It is too bad because there appears to be a big improvement with daylight shots at short exposures.

Jim

 

[dukeblue91]

Good comparison dp_oli!

That's one of the reasons overall why I bought the G9 over the newer model G9II.

I love the G9 I have and I'm blown away by the quality of the pictures and the feature set of the OG G9.

Thanks for sharing,

Duke

 

[dp_oli]

Don't you just hate nick pickers who try their best to prove something in circumstances a camera is not designed to do. M43 cameras are known to be not good in low light so why try and prove otherwise. THe OP has failed to mention the benefits of the G9ii over the G9 ( I have both). No mention of better image stablisation for one example.

Would you talk to me like this in real life too?

I never claimed to make this a complete review and even mentioned that otherwise I see a lot of improvements in the new G9Ii

All I wanted to point out is that the G9Ii is a step back for the stuff that I (occasionally) look for. I don't understand why people have such a hard time to talk about shortcomings of a particular MFT camera without making the obvious general argument that other formats can do things better.

There seem to be 1 or 2 people out there that appreciated my comparisons and can make sense of the intention behind them. If you are not one of them, you can move on. Also, I find the difference to be quite severe and beyond the level of nitpicking.

 

[dp_oli]

I was thinking about buying the G9ii but after reading about the problems of the dual gain turning off I decided not to get it. The problem is most pronounced at low ISO settings on long exposures. Here is a link to one of the forum sites but it is quite technical.

https://www.dpreview.com/forums/thread/4771660

...

Jim

Thanks Jim! I did a lot of research but this thread never popped up in my results. I will have a look at it.

 

[kalisti]

Don't you just hate nick pickers who try their best to prove something in circumstances a camera is not designed to do. M43 cameras are known to be not good in low light so why try and prove otherwise. THe OP has failed to mention the benefits of the G9ii over the G9 ( I have both). No mention of better image stablisation for one example.

Would you talk to me like this in real life too?

I never claimed to make this a complete review and even mentioned that otherwise I see a lot of improvements in the new G9Ii

All I wanted to point out is that the G9Ii is a step back for the stuff that I (occasionally) look for. I don't understand why people have such a hard time to talk about shortcomings of a particular MFT camera without making the obvious general argument that other formats can do things better.

There seem to be 1 or 2 people out there that appreciated my comparisons and can make sense of the intention behind them. If you are not one of them, you can move on. Also, I find the difference to be quite severe and beyond the level of nitpicking.

They've reviewed themselves imo.

I think most of the people who've read you post appreciate the time you've taken to both explain the narrow comparison that you've explored, and explain why, seemed crystal clear to me.

 

[GrayCard]

One thing to keep in mind is that the original G9 has a pixel pitch of 3.34 microns, while the G9MII has a pixel pitch of 3.00 microns. That means that each pixel in the original G9 accepts ~11% more light per pixel, compared to the G9MII. Neither is stellar in low-light photography, but the original G9 will be the better choice of the two for the genre. In comparison, the original S1 at 24 megapixels has a pixel pitch of 5.93 microns, making it a far better choice for low-light work. accepting ~78% more light per pixel. All that said, the ergonomics of the original G9 leave the G9MII in the dust IMESHO.

 

[mmartel]

One thing to keep in mind is that the original G9 has a pixel pitch of 3.34 microns, while the G9MII has a pixel pitch of 3.00 microns. That means that each pixel in the original G9 accepts ~11% more light per pixel, compared to the G9MII. Neither is stellar in low-light photography, but the original G9 will be the better choice of the two for the genre. In comparison, the original S1 at 24 megapixels has a pixel pitch of 5.93 microns, making it a far better choice for low-light work. accepting ~78% more light per pixel. All that said, the ergonomics of the original G9 leave the G9MII in the dust IMESHO.

The pixel size thing is a red herring/myth* - assuming you resample to common viewing sizes, which the OP has done (IIRC).

If you don't resample to a common viewing size (i.e. you compare camera A vs B at different levels of scene magnification) then you're magnifying the noise in the image... so you're going to see it more in the higher megapixel camera.

When it comes to image quality, total light (without oversaturating the sensor) matters most. Then other factors such as sensor tech, read noise, and so forth.

* Number of pixels (which is inversely correlated with pixel area) affects read noise but it's generally very far down on the importance list