Comparison of processing between Pixel 2XL and Sony A6000

[PhotoFactor]

I thought everybody would find this interesting.

It's why so many people ask "why does my phone's output look better than my good camera's output?".

I just ran a test on my A6000 versus my Pixel. The summary is that I shot a high contrast scene on auto metering on the Pixel, SOOC, default settings. I then ran my Sony with DRO settings 2, 4, 5, and Auto. I then ran the in-camera HDR at settings 2 stops, 3 stops, 4 stops, 6 stops, and Auto.

The upshot is that the Pixel STILL did a better job than any of those settings, from a color and DR perspective. And of course if I were to be using the Sony in a range of situations, each situation would require different optimal settings, but the Pixel would choose and be superior in most of them. This refers to default DR out of camera as well as color. We all know that total resolution, ability to push/pull RAW, and of course change focal lengths are superior on the ILC. But still, you can see why the smartphones are so compelling to an average user. They take almost no work and another bonus is you can electronically share, instantly. I've printed photos from my Pixel up to 11x14 and I doubt you'd notice a sharpness difference (in fact, specifically with that horrid Sony kit lens 16-50, the Pixel would probably be sharper).

See a few of these shots below.

Note the detail outside the window and where things are blown out between the pictures. Also note the shadow detail between the different pictures. The Pixel does better than any of the Sony settings for HDR or DRO. The Pixel protects the outdoor portion of the scene from blowout better than any of these Sony built-in settings. You could of course edit a Sony RAW to get a similar result, but it would take a lot of time and you might need to do some masking to get a similar result.

I did this test because in another forum, there was a lot of snark about how you could dial in settings on your good camera and equal or better phone exposure and DR. Based upon my test, that appears not to be the case - at least with the A6000. People don't want to believe me when I say how amazing the default processing is on the Pixel phones


Pixel 2 XL, default SOOC




Sony A6000, standard jpg output




Sony A6000, DRO Auto




Sony A6000, DRO 5 (highest setting)




Sony A6000, Auto HDR (in-camera)




Sony A6000, HDR bracketing (6 stops/maximum), in-camera processing

 

[onlyfreeman]

I believe it.

The software on cameras is no match to that on phones.

 

[noisephotographer]

Fujifilm's DR200 or DR400 + exposure compensation might be a bit better. In a store I compared it with my Nexus 5x in HDR+ On mode and the Fujifilm wasn't bad (actually in this scene the Nexus dynamic range was only better when I touched the highlights) (Nexus had more natural colors by the way). Canon's Highlight Tone Priority mode + less contrast + exposure compensation was much, much worse than Fujifilm, Nikon's dynamic range was much worse, too. Sony's Dro + exposure compensation was a bit worse than Fujifilm, if I remember correctly.

Though I think that Fuji's DR400 might look a bit unnatural sometimes. You can boost the shadows even more (there is an additional shadows setting), but then it often seems to look very unnatural, only seems to work for specific scenes, see the DR400%, Shadows -2, Highlights -2 example at https://www.jmpeltier.com/travelogues/wp-content/uploads/DR4.jpg from http://www.jmpeltier.com/2018/07/01/fujifilm-dynamic-range-settings/

Fujifilm's new color chrome effect (can be combined with DR400) can also capture the highlights better, see https://mirrorlesscomparison.com/wp-content/uploads/2018/10/Fuji-xt2-vs-xt3-Chrome-Effect-crop-2.jpg from https://mirrorlesscomparison.com/fujifilm-vs-fujifilm/fuji-xt2-vs-xt3/

or see the color chrome passage at https://www.dpreview.com/reviews/fujifilm-gfx-50s/7

 

[Gal Root]

Yeah, the A6000 is no match for the Pixel's algorithms. That's true ever since Google introduces HDR+.

 

[tonyz1]

Have you tried running your A6000 raw image through lightroom? It should be able to clean it up fast even if you just press auto.

I thought everybody would find this interesting.

It's why so many people ask "why does my phone's output look better than my good camera's output?".

I just ran a test on my A6000 versus my Pixel. The summary is that I shot a high contrast scene on auto metering on the Pixel, SOOC, default settings. I then ran my Sony with DRO settings 2, 4, 5, and Auto. I then ran the in-camera HDR at settings 2 stops, 3 stops, 4 stops, 6 stops, and Auto.

The upshot is that the Pixel STILL did a better job than any of those settings, from a color and DR perspective. And of course if I were to be using the Sony in a range of situations, each situation would require different optimal settings, but the Pixel would choose and be superior in most of them. This refers to default DR out of camera as well as color. We all know that total resolution, ability to push/pull RAW, and of course change focal lengths are superior on the ILC. But still, you can see why the smartphones are so compelling to an average user. They take almost no work and another bonus is you can electronically share, instantly. I've printed photos from my Pixel up to 11x14 and I doubt you'd notice a sharpness difference (in fact, specifically with that horrid Sony kit lens 16-50, the Pixel would probably be sharper).

See a few of these shots below.

Note the detail outside the window and where things are blown out between the pictures. Also note the shadow detail between the different pictures. The Pixel does better than any of the Sony settings for HDR or DRO. The Pixel protects the outdoor portion of the scene from blowout better than any of these Sony built-in settings. You could of course edit a Sony RAW to get a similar result, but it would take a lot of time and you might need to do some masking to get a similar result.

I did this test because in another forum, there was a lot of snark about how you could dial in settings on your good camera and equal or better phone exposure and DR. Based upon my test, that appears not to be the case - at least with the A6000. People don't want to believe me when I say how amazing the default processing is on the Pixel phones


Pixel 2 XL, default SOOC


Sony A6000, standard jpg output


Sony A6000, DRO Auto


Sony A6000, DRO 5 (highest setting)


Sony A6000, Auto HDR (in-camera)


Sony A6000, HDR bracketing (6 stops/maximum), in-camera processing

 

[Panino Manino]

But look at the shadows.

Can't you get equal/better results lifting the shadows in post?

 

[PhotoFactor]

Have you tried running your A6000 raw image through lightroom? It should be able to clean it up fast even if you just press auto.

Short answer is no. I have an old version of LR (4) and don't tend to use it. I doubt that an Auto on a RAW would do it, but I could be mistaken.

Still, the point is that from another thread, people feeling threatened by phones are snarking about how poor the phones are, and how if they just "dialed in" HDR or DRO they could replicate phone output. While clearly the ILC has more versatility, high ISO capability, focal length capability, etc, the processing at least on the Pixels (and I suspect the latest iPhone) is superior to anything you can get SOOC from at least the Sony A6000. And I suspect that would be true even for FF systems from the big 3.

I may also test my Oly EM10ii if I get a chance. But in general, it's DR is worse than my Sony, so I doubt it would be any better than the Sony.

Sure you can get great output from the ILC - you just need to manually process which takes a lot of time and effort. And that's before you even worry about online sharing which would of course be a much larger hassle with an ILC than with a phone.

Different tools each have different strengths.

 

[PhotoFactor]

But look at the shadows.

Can't you get equal/better results lifting the shadows in post?

See my other comment. The test was to see if I could get as good a photo as my phone by changing settings and not PP.

 

[tonyz1]

Have you tried running your A6000 raw image through lightroom? It should be able to clean it up fast even if you just press auto.

Short answer is no. I have an old version of LR (4) and don't tend to use it. I doubt that an Auto on a RAW would do it, but I could be mistaken.

Still, the point is that from another thread, people feeling threatened by phones are snarking about how poor the phones are, and how if they just "dialed in" HDR or DRO they could replicate phone output. While clearly the ILC has more versatility, high ISO capability, focal length capability, etc, the processing at least on the Pixels (and I suspect the latest iPhone) is superior to anything you can get SOOC from at least the Sony A6000. And I suspect that would be true even for FF systems from the big 3.

I may also test my Oly EM10ii if I get a chance. But in general, it's DR is worse than my Sony, so I doubt it would be any better than the Sony.

Sure you can get great output from the ILC - you just need to manually process which takes a lot of time and effort. And that's before you even worry about online sharing which would of course be a much larger hassle with an ILC than with a phone.

Different tools each have different strengths.

Have you tried adjusting the metering mode on your Sony A6000? In your jpeg file from the pixel 2 XL, it says that it's metering is running in Center Weighted Average mode. Try putting your A6000 in a similar mode - it might adjust the light levels on your camera. I typically leave my metering mode on Multi or Highlights (it purposely makes it darker to prevent bright spots from being blown out - useful for raw editing).

Average tends to make the whole screen brighter.

I also ran the same test you did just now with my Sony A7R III in APS-C mode, Pixel 3, and iPhone X. iPhone quality is pretty poor and not worth discussing. Pixel 3's image isn't bad but it's over sharpened by default and you can easily get a better looking image from a Sony raw file with 10 seconds of Lightroom work. I'm happy to share my samples via PM if you're interested.

 

[PhotoFactor]

Have you tried running your A6000 raw image through lightroom? It should be able to clean it up fast even if you just press auto.

Short answer is no. I have an old version of LR (4) and don't tend to use it. I doubt that an Auto on a RAW would do it, but I could be mistaken.

Still, the point is that from another thread, people feeling threatened by phones are snarking about how poor the phones are, and how if they just "dialed in" HDR or DRO they could replicate phone output. While clearly the ILC has more versatility, high ISO capability, focal length capability, etc, the processing at least on the Pixels (and I suspect the latest iPhone) is superior to anything you can get SOOC from at least the Sony A6000. And I suspect that would be true even for FF systems from the big 3.

I may also test my Oly EM10ii if I get a chance. But in general, it's DR is worse than my Sony, so I doubt it would be any better than the Sony.

Sure you can get great output from the ILC - you just need to manually process which takes a lot of time and effort. And that's before you even worry about online sharing which would of course be a much larger hassle with an ILC than with a phone.

Different tools each have different strengths.

Have you tried adjusting the metering mode on your Sony A6000? In your jpeg file from the pixel 2 XL, it says that it's metering is running in Center Weighted Average mode. Try putting your A6000 in a similar mode - it might adjust the light levels on your camera. I typically leave my metering mode on Multi or Highlights (it purposely makes it darker to prevent bright spots from being blown out - useful for raw editing).

It was set on center (as opposed to spot or multi). Yet the Sony still blew out the outside, which was in the center.

Average tends to make the whole screen brighter.

Yes, probably it would have made the room brighter but blown out the outside even further.

I also ran the same test you did just now with my Sony A7R III in APS-C mode, Pixel 3, and iPhone X. iPhone quality is pretty poor and not worth discussing. Pixel 3's image isn't bad but it's over sharpened by default and you can easily get a better looking image from a Sony raw file with 10 seconds of Lightroom work. I'm happy to share my samples via PM if you're interested.

Thanks for running an additional test and comparing to the iPhone. I agree the Pixel does a lot of sharpening. And I'm sure the Sony would be better with the correct editing. The test wasn't to say you can't take a better picture with the higher-end camera (surprise!), it was to compare the default phone processing and see how close you could get to that by chaning the camera's settings. And the answer seems to be not very close to as good as the Pixels. That's why I and many others are so amazed by the Pixel's "camera".

 

[Panino Manino]

But look at the shadows.

Can't you get equal/better results lifting the shadows in post?

See my other comment. The test was to see if I could get as good a photo as my phone by changing settings and not PP.

I don't agree that Pixel's algorithms are so "magical" as some people sees it (the crushed blacks is something that I can't accept). Having the opportunity to capture and combine a dozen of photos almost instantly will give it a few advantages of course, but all that super dynamic range isn't in practice mostly just pushing the shadows super hard? Remembers me the DRO option from apps like OpenCamera, it does practically the same on any garbage sensor+soc. But cameras don't do this because the manufactures want to avoid the downsizes to the image quality, leaving the decision for the user to do this manually. It's a choice, maybe eventually the camera manufactures will add this choice, but for now they don't want even if they can.

 

[Jostian]

the Pixel still over-processes though, look at the detail from these 2 crops, the detail from the Sony is much better and natural looking, though the Pixel 2 does an admirable job the processed look to fine details from the Pixel 2 and phones in general means that my dedicated camera's are still safe (for now)...




Pixel 2




Sony A6000



--
Jostian

https://www.flickr.com/photos/jostian/

 

[Panino Manino]

The Pixel (or GCam) isn't "over-processing". It's a side effect from "computational multi-frame/exposure".

Look at the photos from phones from some years ago and compare with photos from phones now that use this new holy grail. All the new phones have this "pixelated" look to some degree.

It's not just the Sony camera that will look more natural, any camera that don't combine a gazillion of exposures will look more natural and I have to say, if the Pixel/GCam over-sharpens than it does consistently much better than almost any other camera out there from the past 3 years (this is very visible on grass.

 

[Jostian]

The Pixel (or GCam) isn't "over-processing". It's a side effect from "computational multi-frame/exposure".

the look is still over-processed, i.e.: less natural looking.

Look at the photos from phones from some years ago and compare with photos from phones now that use this new holy grail. All the new phones have this "pixelated" look to some degree.

yip, but as you say they are getting better and better.

It's not just the Sony camera that will look more natural, any camera that don't combine a gazillion of exposures will look more natural and I have to say, if the Pixel/GCam over-sharpens than it does consistently much better than almost any other camera out there from the past 3 years (this is very visible on grass.

camera phone, yes the Pixels are about as good as you can get, the older Nokia's N8, 808, etc had the most natural processing but are now defunct as an OS.

 

[noisephotographer]

By the way, even when you brighten the shadows of the Dro 5 A6000 image, it looks worse than the Pixel 2 due to possible vignetting of the Sony lens (and because the A6000 doesn't have the latest Aps-c sensor). Awful color noise.

I even think that my Nexus 5x might perform a little bit better at base Iso than the Pixel 2 due to the larger sensor, but when you use Night Sight, the Pixel 2 should be better at base Iso.

 

[liopleurodon]

Yup, I've done the test too, Pixel 1 vs 80D and more recently Pixel 3 vs a7 III. However I think the scene you chose is an outlier. Most scenes do not have such wild DR to capture. Such a scene is the Pixel's strength.

On an a6000 I've have shot that with highlight priority metering and +1EV of exposure compensation and then pulled the highlights down in post and pulled shadows and midtones up.



However I'd be willing to bet the default profile in DxO PhotoLab would produce an image from your raw that looked just like the Pixel 2's only with more detail. DxO's smart lighting algorithm is pretty excellent with no user input.



Here's an a7 III vs Pixel 3 Night Sight (basically HDR+ without any time constraints) shot in a scene that plays to the Pixel 3's strengths. The a7iii shot was just processed through the standard preset on DxO PhotoLab 2. If you look closely the Sony shot is clearly better, but at web sizes they look fairly interchangeable.









Also as far as getting that Pixel style tile&merge tech in camera, computer power needs to catch up first. The a6000 can't read images from its sensor fast enough do what the Pixel does (Pixel uses a stacked BSI sensor like the a9 and only has 12mp to deal with on a physically smaller sensor). The growth in processing time required going from 12mp to 24mp isn't a simple 2x linear, its more like 4x. You need both fast readout speeds from the sensor and an SoC that can sustain fast image processing of multiple 24mp frames without making the rest of the camera unusable.

 

[liopleurodon]

By the way, even when you brighten the shadows of the Dro 5 A6000 image, it looks worse than the Pixel 2 due to possible vignetting of the Sony lens (and because the A6000 doesn't have the latest Aps-c sensor). Awful color noise.

I even think that my Nexus 5x might perform a little bit better at base Iso than the Pixel 2 due to the larger sensor, but when you use Night Sight, the Pixel 2 should be better at base Iso.

Larger sensor but slower glass. You're getting a very similar amount of total light from the two devices which should result in similar SNR levels.

 

[noisephotographer]

By the way, even when you brighten the shadows of the Dro 5 A6000 image, it looks worse than the Pixel 2 due to possible vignetting of the Sony lens (and because the A6000 doesn't have the latest Aps-c sensor). Awful color noise.

I even think that my Nexus 5x might perform a little bit better at base Iso than the Pixel 2 due to the larger sensor, but when you use Night Sight, the Pixel 2 should be better at base Iso.

Larger sensor but slower glass. You're getting a very similar amount of total light from the two devices which should result in similar SNR levels.

No, that's not true for base Iso because a camera with a faster lens needs to choose a shorter exposure time in order to prevent overexposure.

 

[tonyz1]

By the way, even when you brighten the shadows of the Dro 5 A6000 image, it looks worse than the Pixel 2 due to possible vignetting of the Sony lens (and because the A6000 doesn't have the latest Aps-c sensor). Awful color noise.

I even think that my Nexus 5x might perform a little bit better at base Iso than the Pixel 2 due to the larger sensor, but when you use Night Sight, the Pixel 2 should be better at base Iso.

You can't brighten up the jpeg - you have to work in the raw file to do the editing correct. If you work in the raw files for a Sony sensor camera vs. in the Pixel 3 or any of these smartphone cameras, you will notice a huge difference in dynamic range.

 

[noisephotographer]

By the way, even when you brighten the shadows of the Dro 5 A6000 image, it looks worse than the Pixel 2 due to possible vignetting of the Sony lens (and because the A6000 doesn't have the latest Aps-c sensor). Awful color noise.

I even think that my Nexus 5x might perform a little bit better at base Iso than the Pixel 2 due to the larger sensor, but when you use Night Sight, the Pixel 2 should be better at base Iso.

You can't brighten up the jpeg - you have to work in the raw file to do the editing correct. If you work in the raw files for a Sony sensor camera vs. in the Pixel 3 or any of these smartphone cameras, you will notice a huge difference in dynamic range.

Of course one can brighten jpgs. I do this a lot and the difference is not always huge. When the shadows are not completely black/clipped, there is usually not a huge difference. The Sony A6000 HDR jpg has much, much less noise in the shadows.