Pixel 6 Pro vs 1" Camera - Low light torture test

Edgar_in_Indy

Senior Member
Messages
1,059
Solutions
1
Reaction score
872
Location
Indianapolis, US
I've been running my Pixel 6 Pro through its paces, and last night I took it around to some sites that I like to use to test a camera's low light abilities. These scenes represent a challenge, since they contain a lot of dynamic range, from deep shadows to bright light sources, and they feature a variety of detailed textures.

I am comparing my Pixel 6 Pro to my Panasonic LX-10. The LX-10 is notable excellent manual controls and for being the only 1" compact camera with an f/1.4 zoom lens (at wide angle), so for its size it's extremely capable in low light situations, and with its fast telephoto lens it's a very versatile camera.

Here's my LX-10, with an aftermarket grip:



A head-to-head comparison is complicated by the many various shooting modes and adjustments available on each camera. I therefore shot and/or processed each scene different ways multiple times on both cameras.

First, here is the same scene shot in RAW and processed to taste on both cameras, with the Panasonic's 20MP image down-sampled to 12MP to match the output of the Pixel 6 Pro. No noise reduction applied. First the Panasonic image, and then the Pixel:





And here are the SOOC JPG images from each camera, again with Panasonic on top:





I didn't care for the SOOC result from either camera. With the Panasonic, the blacks were crushed too much, resulting in a bleaker scene than I viewed in person, and with the Pixel the shadows were boosted too much, resulting in a flat and artificial image.

To be fair to the Panasonic, I almost never shoot my real cameras in JPG, so I have never bothered fiddling with the JPG presets in the camera. It's very possible that I could have set the camera up to capture a better JPG image.

As for the Pixel, there's not much in the way of global adjustments to JPG output, but when capturing an image you do have sliders to adjust color temp, brightness, and shadow levels. Here's the same scene shot on the Pixel again, but this time with the shadow slider turned down from the default setting. I much prefer that image to the default capture.



In the first set of RAW images posted above, I did not apply any noise reduction since I wanted the sensor performance on display. However, I typically apply noise reduction if I think an image can benefit from it. I usually use Topaz Labs Denoise AI, which uses AI to selectively apply noise reduction to different parts of the image where it thinks it needs it.

Here are full resolution RAW images from each camera, with noise reduction applied to taste:





And finally, here is the full-resolution 20MP RAW image from the LX10:



Summary:

The Pixel 6 Pro's main sensor compares favorably to a 1" sensor in low light. The 1" camera can capture slightly more detail in the in-focus areas, but the Pixel 6 Pro's smaller sensor has more depth of field, resulting in more areas in focus.

While noise levels for the Pixel 6 Pro compare favorably to the 1" camera, I did not observe the expected dynamic range advantages that computational photography is supposed to provide. Dynamic range seemed comparable between the two cameras, but perhaps that can still be considered a win for the Pixel 6, since it is using a much smaller sensor.

RAW files were easier to work with from the 1" camera. For the Pixel 6 I had to go deeper into the sliders to match the image from the 1" camera. And for some reason getting correct/pleasing white balance and tint from the Pixel files took more finessing.

The bad news about the Pixel 6 Pro is that it does seem to provide significantly better image quality than my last two mid-range Android phones that were using gcam...the Motorola Moto G Power 2020 and then the TCL 10 Pro.

However, it is much easier to use and shoot with the Pixel 6 Pro than with my old phones since the Pixel captures the image almost instantly, while my old phones would require a few moments to capture and merge the multiple exposures.

I apologize for the long post, but I wanted to make sure I gave both cameras a fair comparison. I have more scenes I can post if there's interest, including telephoto comparisons, but it's getting late for now and I need to go to bed.
 

Attachments

  • 4264310.jpg
    4264310.jpg
    1.6 MB · Views: 1
  • 4264296.jpg
    4264296.jpg
    5.8 MB · Views: 1
  • 4264293.jpg
    4264293.jpg
    3.1 MB · Views: 1
  • 4264298.jpg
    4264298.jpg
    3.5 MB · Views: 1
  • 4264291.jpg
    4264291.jpg
    2.8 MB · Views: 1
  • 4264290.jpg
    4264290.jpg
    4.2 MB · Views: 1
  • 4264289.jpg
    4264289.jpg
    6.6 MB · Views: 1
  • 4264292.jpg
    4264292.jpg
    4.2 MB · Views: 1
  • 4264297.jpg
    4264297.jpg
    3.8 MB · Views: 1
Anybody know why my images don't have the "View: original size" link below them like I hoped they would? Did I do something wrong? Argh! That's annoying!

FYI, you can still get the full size image, but it requires clicking on the image to open it in the gallery, and then clicking the "JPEG" hotlink in the gallery sidebar.

EDIT: I did some experimenting, and it looks like the problem is because I used the DPR gallery, instead of uploading the pictures directly from my computer. Doesn't look like there's any way to change it now.
 
Last edited:
Thanks for this great comparasion :) I really like these kind of posts :)
Summary:

The Pixel 6 Pro's main sensor compares favorably to a 1" sensor in low light. The 1" camera can capture slightly more detail in the in-focus areas, but the Pixel 6 Pro's smaller sensor has more depth of field, resulting in more areas in focus.
For me these photos are great, but Pixel potential is proably even better than You think. Xiaomi MI10 Ultra have smaller sensor than Pixel 6 PRO and is able to capture good 48 megapixel night images

5709832c0e074d3fa7cdf0a0d4f2a61d.jpg
While noise levels for the Pixel 6 Pro compare favorably to the 1" camera, I did not observe the expected dynamic range advantages that computational photography is supposed to provide. Dynamic range seemed comparable between the two cameras, but perhaps that can still be considered a win for the Pixel 6, since it is using a much smaller sensor.

RAW files were easier to work with from the 1" camera. For the Pixel 6 I had to go deeper into the sliders to match the image from the 1" camera. And for some reason getting correct/pleasing white balance and tint from the Pixel files took more finessing.
Proably thanks to conventional RGB sensor. Phone have quad bayer filter, did You try RAW Therapee? It's very good to edit RAW from modern phones.
The bad news about the Pixel 6 Pro is that it does seem to provide significantly better image quality than my last two mid-range Android phones that were using gcam...the Motorola Moto G Power 2020 and then the TCL 10 Pro.
I think You are limited by 12 megapixel resolution in RAW. MI10 Ultra allow to install 48 megapixel mod via magisk, which unlock full resolution RAWs and effects are great. I don't know is that mod working with Pixel, but You can try.
I apologize for the long post, but I wanted to make sure I gave both cameras a fair comparison. I have more scenes I can post if there's interest, including telephoto comparisons, but it's getting late for now and I need to go to bed.
I will always happy to see that :) Overall auto mode processing in this phone not looksing good for me compared to RAW - this sensor is beast. Did You try install MTSL V5 GCAM version with MI11 Ultra config - output from this app is fantastic - almost good as photos developed from RAW. You can install from here https://www.celsoazevedo.com/files/android/google-camera/dev-mtsl/f/dl4/
 
Last edited:
Summary:

The Pixel 6 Pro's main sensor compares favorably to a 1" sensor in low light. The 1" camera can capture slightly more detail in the in-focus areas, but the Pixel 6 Pro's smaller sensor has more depth of field, resulting in more areas in focus.
For me these photos are great, but Pixel potential is proably even better than You think. Xiaomi MI10 Ultra have smaller sensor than Pixel 6 PRO and is able to capture good 48 megapixel night images
Perhaps. I'm still getting familiar with it.
RAW files were easier to work with from the 1" camera. For the Pixel 6 I had to go deeper into the sliders to match the image from the 1" camera. And for some reason getting correct/pleasing white balance and tint from the Pixel files took more finessing.
Proably thanks to conventional RGB sensor. Phone have quad bayer filter, did You try RAW Therapee? It's very good to edit RAW from modern phones.
No, I've only used Adobe Photoshop.
The bad news about the Pixel 6 Pro is that it does seem to provide significantly better image quality than my last two mid-range Android phones that were using gcam...the Motorola Moto G Power 2020 and then the TCL 10 Pro.
I think You are limited by 12 megapixel resolution in RAW. MI10 Ultra allow to install 48 megapixel mod via magisk, which unlock full resolution RAWs and effects are great. I don't know is that mod working with Pixel, but You can try.
Oops, I meant to say "the Pixel 6 Pro does not seem to provide significantly better image quality than my last two mid-range Android phone", but it sounds like you knew what I meant. And you may be right about the 12MP being a limiting factor.
I apologize for the long post, but I wanted to make sure I gave both cameras a fair comparison. I have more scenes I can post if there's interest, including telephoto comparisons, but it's getting late for now and I need to go to bed.
I will always happy to see that :) Overall auto mode processing in this phone not looksing good for me compared to RAW -
Yeah, I definitely get better results from RAW, but I wasn't expecting otherwise since that has always been the case with every camera I've owned.
this sensor is beast. Did You try install MTSL V5 GCAM version with MI11 Ultra config - output from this app is fantastic - almost good as photos developed from RAW. You can install from here https://www.celsoazevedo.com/files/android/google-camera/dev-mtsl/f/dl4/
No, one of my primary reasons for buying the Pixel was that it essentially had Gcam built in. I've installed it on my last three phones, but it was always a bit of a hack since there were certain aspects that were wonky or did not work perfectly with the various different camera modules since it was being shoe-horned into a 3rd party phone.

Using the native app on the Pixel has been a much more pleasant experience, but I did not realize there would still be advantages to installing Gcam. I'm honestly a bit disappointed if I need to install Gcam to unlock the phone's true potential, but maybe it's something I'll try.
 
Yeah, I definitely get better results from RAW, but I wasn't expecting otherwise since that has always been the case with every camera I've owned.
this sensor is beast. Did You try install MTSL V5 GCAM version with MI11 Ultra config - output from this app is fantastic - almost good as photos developed from RAW. You can install from here https://www.celsoazevedo.com/files/android/google-camera/dev-mtsl/f/dl4/
No, one of my primary reasons for buying the Pixel was that it essentially had Gcam built in. I've installed it on my last three phones, but it was always a bit of a hack since there were certain aspects that were wonky or did not work perfectly with the various different camera modules since it was being shoe-horned into a 3rd party phone.

Using the native app on the Pixel has been a much more pleasant experience, but I did not realize there would still be advantages to installing Gcam. I'm honestly a bit disappointed if I need to install Gcam to unlock the phone's true potential, but maybe it's something I'll try.
Yes, this is a bit funny 😅 I understand reasons of buying pixel - i also like when phone works and i don't care about gcam settings and bugs :) I don't know why, but processing in pixel is not very natural comparing to MTSL output. Maybe it's for people, who don't want natural SOOC, but good looking photos on instagram, but it's easy to get proper one with MTSL V5 and MI11 Ultra config(also GN sensor). It is 12 megapixel photo straight from camera, i think You will be happy with this :) Sorry for off-topic.
9a0bf26b555d4a139754168ac04f334a
 

Attachments

  • 9a0bf26b555d4a139754168ac04f334a.jpg
    9a0bf26b555d4a139754168ac04f334a.jpg
    7.9 MB · Views: 1
Last edited:
Yeah, those high resolution photos you've been sharing are nice. I really wish the Pixel would allow full output like that. In low-light 12MP is probably fine, but with good lighting the big sensor in the Pixel 6 should be capable of much more.
 
Just curious if you shot your Pixel low-lights using the regular HDR+ or Nightsight (presumably on a tripod)? I notice the ISO higher on the Pixel shots, and my experience has been additional exposures in HDR+ "Enhanced" or "Nightsight" lower the ISO on the computational shot.

I've used GCam up through version 7.3, but the Pixel 6 has now moved to more camera implementation on its proprietary Tensor SoC, and I haven't seen much on how that will affect GCam mods going forward. (I was hoping the smaller sized Pixel models would have optical telephoto, the main attraction for me but not wanting a big slab in my pocket, so am still more inclined to get a GCam compatible alternative.)
 
Here's another scene from the same night. This is is particularly challenging since it features a large, very bright sign in the center of a dark area. It also represents a white balance challenge with the two different light sources...the sign, and the building entryway. And the bricks, stonework, and concrete provide plenty of tests for texture and detail retention.

First, here's the RAW shot from each camera, with the Panasonic's image scaled down to 12MP to match the Pixel:

27b063928f8d45e987f3d2a9b5ba5e2b.jpg

f7b7b5c2afc948b694ce5de374261d42.jpg

And here are the two RAW files, with noise reduction applied to taste (but without down-scaling the 20MP image):

43b57d7c110849739745303c3accc479.jpg

755979f2a0414683bdb1632f9f4cba16.jpg

And here are the JPEG's SOOC:

c54eeea855e64dcca68ad3960b12e43f.jpg

7eb836395cc4434d8e53f5598abe1689.jpg

And here is the Pixel 6 Pro's JPEG with the shadows slider boost reduced:

945e15fd1a09476fa5ba9313d375c832.jpg

And finally, for the sake of thoroughness, the full 20MP Panasonic RAW file with no noise reduction:

f93755c880dd41afafb6e065883ccc5e.jpg

My thoughts: The Pixel 6 did very well in terms of noise and detail, producing a less noisy image than the 1" camera, but I was left disappointed in terms of dynamic range. This scene is a torture test for dynamic range, and I was really expecting the computational photography to make a bigger difference in this regard.

I'm sure it did help, as I had to boost the shadows quite a bit for the 1" photo, which is probably why the Pixel ended up with a less-noisy image, but the sign is blown out just as much in both pictures.

I will make an observation that it was easier to capture a sharp, in-focus images in low light with the Pixel than it was with the camera. When shooting at slow shutter speeds with the camera I had to be sure to use very good form, or else I would end up with evidence of camera shake. This did not seem to be an issue at all with the phone.
 
Last edited:
Just curious if you shot your Pixel low-lights using the regular HDR+ or Nightsight (presumably on a tripod)? I notice the ISO higher on the Pixel shots, and my experience has been additional exposures in HDR+ "Enhanced" or "Nightsight" lower the ISO on the computational shot.
No, I did not use Nightsight. I've never cared for the feature on any phone where I've seen it, since it usually gives results similar to a long nighttime exposure. That can be a cool creative effect at times, but you usually end up with a scene that looks much brighter than it was in person. It also seems like the nightsight shots usually end up softer, with more obvious noise reduction.

I think on another night I did experiment with the same scene using nightsight, so I'll take a look later and see if I can dig those pictures up.
 
Last edited:
One other way you can "adjust" the jpg output on a pixel is to tap the screen for the spot you want exposed. So if you tap on a dark spot, you'll get a brighter image, and if you tapped on the bowling pin, then you would get a darker image and the pin would be more properly exposed. It works and is a fast way to adjust exposure. I use this technique on sunsets, for example. By tapping near the sun/bright clouds, I darken the rest of the image and it improves color saturation.
 
One other way you can "adjust" the jpg output on a pixel is to tap the screen for the spot you want exposed. So if you tap on a dark spot, you'll get a brighter image, and if you tapped on the bowling pin, then you would get a darker image and the pin would be more properly exposed. It works and is a fast way to adjust exposure. I use this technique on sunsets, for example. By tapping near the sun/bright clouds, I darken the rest of the image and it improves color saturation.
Yes, that is a handy way to adjust exposure. In this case though, adjusting for the pin would cause the rest of this scene to be extremely underexposed. I'm not sure there is a perfect exposure for this scene... It's a tough one for sure.
 
Just curious if you shot your Pixel low-lights using the regular HDR+ or Nightsight (presumably on a tripod)? I notice the ISO higher on the Pixel shots, and my experience has been additional exposures in HDR+ "Enhanced" or "Nightsight" lower the ISO on the computational shot.
No, I did not use Nightsight. I've never cared for the feature on any phone where I've seen it, since it usually gives results similar to a long nighttime exposure. That can be a cool creative effect at times, but you usually end up with a scene that looks much brighter than it was in person. It also seems like the nightsight shots usually end up softer, with more obvious noise reduction.

I think on another night I did experiment with the same scene using nightsight, so I'll take a look later and see if I can dig those pictures up.
The basic HDR+ is very good for handheld along with touch metering or the exposure slider - some GCams have fairly sophisticated under-the-hood settings. My go to "keep it simple" spin has these:



13c0358dda0e4328ac3bd05df0f123fb.jpg.png

The "custom lib" and hilight/shadow "saturation adjustments" could be helpful shooting a lot of low light. Reducing the frames for Enhanced and Nightsight will speed them up, though not like regular HDR+. My phone works with a couple of the more tweak-intensive GCams that really go to town on custom settings and configuration files - but the sell for me is simplicity.
 
Yeah, the stock Google camera app doesn't have a whole ton of options for controlling how an image is captured, however, I do find the Google Photos editing app included on the phone to be pretty useful and full-featured for tweaking the final image. I frequently shoot a photo and then open it right up in the editor to adjust it to taste. I'll even use one of the filters on occasion.

Oh, and you asked if I was using a tripod. No, everything was shot handheld, so the shots reflect my typical real-world use scenario. But as I mentioned, the Pixel 6 seems to be very adept at capturing sharp photos in what would typically be a low shutter speed situation with a traditional camera.
 
Last edited:
Just curious if you shot your Pixel low-lights using the regular HDR+ or Nightsight (presumably on a tripod)? I notice the ISO higher on the Pixel shots, and my experience has been additional exposures in HDR+ "Enhanced" or "Nightsight" lower the ISO on the computational shot.
Okay, I went back through the photos on my phone and I found some shots from last month where I shot the same scene with "Night Sight" enabled. To me, Night Sight only made the photo worse. It made the scene look even flatter and more artificial than the regular JPG engine already does.

A couple disclaimers...these pictures were taken on different nights, so the composition is slightly different. It looks like I was stepped back another step or two, so the view of the building façade is slightly wider. Also, there were some lighted Christmas decorations, which may have warmed up the front of the building a bit.

And as the Night Sight photo was taken on January 9th, it's possible the camera app was an earlier version, so the output may not represent 100% the current Pixel 6's Night Sight output...I'm just not sure.

Anyway, here's the regular JPG on top compared to the Night Sight JPG on bottom:

088dc4ce5163415c85b56b91397b542d.jpg

a7a27bb34f3e4278ab34ea8985ae3635.jpg

And here's another regular JPG (from the original comparison set), but with the shadows slider turned down, which I think results in a more natural and better looking image:

12fb5e2b75dd4cf88e6b4816bf4001c8.jpg

I also compared the RAW files. First, the corresponding RAW file from the regular JPG (also from the original set of images):

37d515ee3f114581b7f3a04c2f31ae0e.jpg

And now the RAW file from the Night Sight image:

d931a976f6ff483ab1eb08eb8096b40f.jpg

Putting aside the slight difference in composition, I cannot really see a difference between the two images. Note that I used the same settings to develop the two RAW images in Adobe Camera RAW.
Perhaps the Night Sight image has a slightly warmer cast, but that could probably be attributed to the presence of Christmas lights in the scene. I'm also not sure what the skies were like, if one image had more moonlight than the other.

But it appears that whatever Night Sight does to a photo does not alter the RAW image.
 
Last edited:
And here's the bowling pin sign picture captured with Night Sight, with the same disclaimer as the post just above:

f28ef5134d10429e94e7e27819f49a27.jpg

And here's the Night Sight RAW image, processed to taste:

cc751daa03984925a6d097372a2d21eb.jpg

Same results as above...the Night Sight image looks fake and horrible to me. Noise, however does seem to be very well controlled, so maybe there is some advantage to shooting in Night Sight, assuming you plan to work with the RAW file later. But this is something I would have to experiment with more before I can say.
 
I like that result from toning down hilight/shadow.

Less noise should be a Nightsight advantage, understanding the different shoot dates the HDR+ ISO looks over 700 with the Nightsight around 140 - strange that the DNG and JPEG show different EXIF ISO levels (?). Again not familiar with current Pixel 6's Tensor changes, Nightsight takes a lot more frames and challenges handholding, more like an open shutter as you said, I keep a little tripod with when I expect to use it. On the older version GCam above I have set HDR+ Enhanced (now discontinued) to only 8 frames, usually pretty manageable handheld, with 24 for Nightsight.
 
Last edited:
As for the Pixel, there's not much in the way of global adjustments to JPG output, but when capturing an image you do have sliders to adjust color temp, brightness, and shadow levels. Here's the same scene shot on the Pixel again, but this time with the shadow slider turned down from the default setting. I much prefer that image to the default capture.
While noise levels for the Pixel 6 Pro compare favorably to the 1" camera, I did not observe the expected dynamic range advantages that computational photography is supposed to provide. Dynamic range seemed comparable between the two cameras, but perhaps that can still be considered a win for the Pixel 6, since it is using a much smaller sensor.

RAW files were easier to work with from the 1" camera. For the Pixel 6 I had to go deeper into the sliders to match the image from the 1" camera. And for some reason getting correct/pleasing white balance and tint from the Pixel files took more finessing.
Not sure what you're smoking, but the SOOC image from the Pixel 6 is far nicer than the LX10 image,

AND far nicer than your processed RAW image, in terms of colour, dynamic range and overall detail (if you look at the corners and edges).
 
Not sure what you're smoking, but the SOOC image from the Pixel 6 is far nicer than the LX10 image,
Not to my eyes, but I realize there must be a lot of people who like that kind of heavily processed rendering, with highly-boosted shadows, since that's what pretty much every smartphone is programmed to output. Personally, I hate it.
AND far nicer than your processed RAW image, in terms of colour, dynamic range and overall detail (if you look at the corners and edges).
The lens in the LX-10 is a pretty extreme design, being an f/1.4-2.8 24-72mm equivalent collapsible zoom covering a fairly large sensor (for a compact camera). So it's admittedly not the sharpest lens at f/1.4, but even so, the sharpness looks comparable to me through most of the central area of the image.

Something else to keep in mind is the differences in depth of field. With its larger sensor and wider lens aperture, the Panasonic is going to have a much narrower DOF than the Pixel camera.

The influence of DOF on overall sharpness has even been noted when comparing newer phones like the Pixel 6 and iPhone 13 to older phones with smaller sensors. In certain situations images taken on the older phones can appear to capture more detail due to the narrower DOF on the newer phones.
 
Last edited:
Boss of Sony wrote:ng.

Not sure what you're smoking, but the SOOC image from the Pixel 6 is far nicer than the LX10 image,

AND far nicer than your processed RAW image, in terms of colour, dynamic range and overall detail (if you look at the corners and edges).
"The bad news about the Pixel 6 Pro is that it does seem to provide significantly better image quality than my last two mid-range Android phones that were using gcam...the Motorola Moto G Power 2020 and then the TCL 10 Pro."

"the bad news..." - this says what this whole thing is about
 
"The bad news about the Pixel 6 Pro is that it does seem to provide significantly better image quality than my last two mid-range Android phones that were using gcam...the Motorola Moto G Power 2020 and then the TCL 10 Pro."

"the bad news..." - this says what this whole thing is about
In case anybody missed it, my original quote was supposed to say that "the Pixel 6 Pro does not provide significantly better image quality than my last two mid-range Android phones that were using gcam..."

Here's the same scene photographed with my last phone, the TCL 10 Pro, developed from RAW:

8bc7b56392164a20a720a6022cd9296d.jpg

And here's the RAW conversion from the Pixel 6 Pro again:

ac30148cd31945ee859cb44624e3e824.jpg

You can probably ignore any brightness or color differences, since the photographs were shot and processed at different times, and those differences are from the post-processing and were done to taste. But, yeah, the Pixel 6's newer and larger sensor didn't seem to result in any better image quality. Perhaps if they didn't limit it to 12MP the Pixel 6 would be able to distance itself more?

Of course the TCL 10 Pro was not a slouch either, since it had a larger-than-average 1/1.7" main sensor and output 16MP image files.

If it weren't for the telphoto lens on the Pixel 6 Pro, and the much quicker capture time when taking photos, I probably would have stayed with my TCL.
 

Keyboard shortcuts

Back
Top