Samsung's next trick in fighting large sensors - Motorola phone with 200MP camera leak

[mr moose]

@LAwLz

 

I'll say what I think, the top left picture to me looks like it has better detail, better contrast and better color, the top right looks a bit washed out.  the bottom image looks  closer to the top left than the top right.

 

Let me know later or in a PM which phone I should buy if pictures are a priority.

[Imbadatnames]

12 minutes ago, mr moose said:

 

No of course not.  You can't see them.

To be fair with Fuji if you’re using the app to transfer them to a phone it sends them at 2MP by default but this seems to be cropped even further or compressed through the forums uploading procedure  

[LAwLz]

1 hour ago, Imbadatnames said:

Are you linking the other photos or not? Also that image is only 1000x480 can I have the original image please 

Not sure what you mean. I have already posted the other photos.

I can give you the original pictures if you want but I am not sure why you want them so much before determining which one looks the best. If the Samsung sensor struggles so much and takes bad pictures, shouldn't it be easy to pick out which one is the Samsung and which one is the iPhone just from those images I linked? No need to have access to things like the metadata to determine that, right?

 

 

43 minutes ago, Imbadatnames said:

There’s no point comparing images that are 33x smaller than the original 

Ohh I get it. You think I downscaled the image. Don't worry, I cropped the image. What you see is a 1:1 pixel perfect representation of what came out of the cameras.

I did not take a 12MP image and then shrink it to 1/33 the size. I took the full image and then cut out everything but 1/33 of it. I almost feel insulted that you think I would downscale the image when trying to do a comparison. I even explained what I did here:

19 hours ago, LAwLz said:

I took the original photos, cropped out the same section and then put them in a single photo.

 

 

 

  

33 minutes ago, Imbadatnames said:

To be fair with Fuji if you’re using the app to transfer them to a phone it sends them at 2MP by default but this seems to be cropped even further or compressed through the forums uploading procedure  

Don't worry, the Fuji picture was not compressed to 2MP. 

I got the full image on my computer and it was that I used to crop it.

 

  

35 minutes ago, Imbadatnames said:

but this seems to be cropped even further

Yes it was cropped, but cropped does not mean lossily compressed. Every single pixel in the image I posted is exactly the same as it was in the original picture. I am simply not posting all of the pixels from the photo.

 

  

36 minutes ago, Imbadatnames said:

or compressed through the forums uploading procedure  

Don't worry, it has not been further compressed by the forum. You do have to click on the image to open it in full resolution though. What you see inline is further compressed but you do see the original when you click on it. Here are direct links if you want:

Picture 1

Picture 2

[Sauron]

On 1/26/2022 at 9:22 PM, LAwLz said:

Pictures from the 108 megapixel Galaxy S21 Ultra are around 20MB. 

I doubt these will be more than "just" 30MB or something around there. Still massive, but not that impractical.

Wait until people start taking RAW pictures with it

[Imbadatnames]

50 minutes ago, LAwLz said:

Not sure what you mean. I have already posted the other photos.

I can give you the original pictures if you want but I am not sure why you want them so much before determining which one looks the best. If the Samsung sensor struggles so much and takes bad pictures, shouldn't it be easy to pick out which one is the Samsung and which one is the iPhone just from those images I linked? No need to have access to things like the metadata to determine that, right?

You’re just not understanding my point. 

50 minutes ago, LAwLz said:

Ohh I get it. You think I downscaled the image. Don't worry, I cropped the image. What you see is a 1:1 pixel perfect representation of what came out of the cameras.

Why did you crop the fucking image? That’s not 1:1

50 minutes ago, LAwLz said:

I did not take a 12MP image and then shrink it to 1/33 the size. I took the full image and then cut out everything but 1/33 of it. I almost feel insulted that you think I would downscale the image when trying to do a comparison. I even explained what I did here:

you’re editing the image it’s not the original image anymore 

50 minutes ago, LAwLz said:

Don't worry, the Fuji picture was not compressed to 2MP. 

I got the full image on my computer and it was that I used to crop it.

 

  

it definitely was because I’ve downloaded the image and it came out to 1000x480 or just under 0.5MP

50 minutes ago, LAwLz said:

Yes it was cropped, but cropped does not mean lossily compressed. Every single pixel in the image I posted is exactly the same as it was in the original picture. I am simply not posting all of the pixels from the photo.

it’s clearly not the Fuji image also looks deep fried so I have no idea what ISO you’re using 

50 minutes ago, LAwLz said:

Don't worry, it has not been further compressed by the forum. You do have to click on the image to open it in full resolution though. What you see inline is further compressed but you do see the original when you click on it. Here are direct links if you want:

Picture 1 3.24 MB · 0 downloads

Picture 2 4.47 MB · 0 downloads

You’ve grouped the three together again. SHARE THE ORIGINAL IMAGES FFS. Why are you trying to hide? Get the original image, don’t edit it and share it why is that so hard? 

[Spotty]

15 minutes ago, Imbadatnames said:

it definitely was because I’ve downloaded the image and it came out to 1000x480 or just under 0.5MP

The forum will display a preview image in the thread that is limited to maximum 1000 pixels width/height. If you right click and save you are only downloading the preview image. To view the full image you need to click on it first.

[LAwLz]

3 hours ago, Imbadatnames said:

You’re just not understanding my point. 

Then can you please elaborate?

 

3 hours ago, Imbadatnames said:

Why did you crop the fucking image? That’s not 1:1

Because if I didn't do that you could just look at the resolution and figure out which was which. I am starting to suspect that that's why you want me to post the original as well, so that you can "cheat" and pick the iPhone image without having to judge the quality of the images.

Also, it is 1:1, in the section I posted.

If I tear out a page of a book, would you say I have altered the text on the page? Of course not. The text is still the same. It's just not the entire book. I want you to evaluate the section I posted.

 

 

3 hours ago, Imbadatnames said:

you’re editing the image it’s not the original image anymore 

Why does it matter if it's not technically the original anymore if what I posted is a pixel perfect representation of the original images? It's just that it's a small section of it.

I wonder totally understand your frustration if I had resized the image, or applied more compression to it or some other destructive process, but in this case nothing like that has happened, and that was intentional.

 

 

3 hours ago, Imbadatnames said:

it’s clearly not the Fuji image also looks deep fried so I have no idea what ISO you’re using 

ISO 1600

1/4 sec shutter speed

f/2.8 aperture

 

3 hours ago, Imbadatnames said:

SHARE THE ORIGINAL IMAGES FFS. Why are you trying to hide?

The resolution of the images.

If I post the original images you can either just look at the EXIF and see "Camera maker: Apple" and "Camera maker: Samsung". Even if I scrub the EXIF you will still be able to just look at the resolution. The iPhone takes pictures with a resolution of 4032x3024. The Galaxy S21 Ultra takes pictures at 4000x3000.

If I post the full images you get several ways of figuring out which is the iPhone image and which is the Galaxy S21 Ultra image. I created this test in a way where you can't cheat.

 

 

Can we just be honest for a minute and admit that the difference in camera performance isn't that big? The Samsung phone certainly don't "struggle" with taking decent photos. It's in the like top 5 smartphone cameras that you can get. Some might prefer the iPhone's camera in some scenarios, and some might prefer the S21 Ultra's camera in some scenarios. I think there are certainly merits to these high megapixel sensors. Do I think we need 200 megapixels in a phone camera? Nope, I think that's beyond the practical limit of our current sensor tech. Do I think 108 megapixels are beyond that limit? Probably. 

 

Of course they can't compete with a DSLR, but do phone cameras even have to? They are different categories of devices and therefore don't compete with each other.

Speaking of DSLRs, Sony have started doing pixel binning in some scenarios on their mirrorless cameras. The Alpha 7S III and the Alpha 1 do pixel binning when recording video for example since it helps with HDR recording. If you want a serious sensor then look up the IMX661. It's a large format sensor from Sony that does pixel binning too. I wouldn't be surprised if we start seeing more cameras with pixel binning in the future. It provides great flexibility and the best of both worlds (high resolution and large pixels) without that big drawbacks.

[Zodiark1593]

40 minutes ago, LAwLz said:

Snippity

If I had to guess, the Top-Right image is most likely the iPhone. The mushy details, particularly in the grass is indicative oh aggressive noise reduction that is pretty characteristic of iPhones. Top left is probably the Samsung. The slightly warm color toning and gradients does indicate some form of processing, and smartphones are generally very good at this. The bottom one is probably the Fujifilm then. The details I cannot describe as either mushy, nor sharp, and there isn't a ton of subtlety of the color. Probably very minimal processing was done. It feels a bit more "raw", though I'm not really familiar with how Fuji does their in-camera processing, though most dedicated cameras I've used tend to leave heavier processing to the user's discretion (via Lightroom, Photoshop, Capture One, etc). The bottom-right of the corner of the two images (Top-Left and Bottom) also has a clear difference in sharpness, though this likely comes down to the lens.

[LAwLz]

49 minutes ago, Zodiark1593 said:

If I had to guess, the Top-Right image is most likely the iPhone. The mushy details, particularly in the grass is indicative oh aggressive noise reduction that is pretty characteristic of iPhones. Top left is probably the Samsung. The slightly warm color toning and gradients does indicate some form of processing, and smartphones are generally very good at this. The bottom one is probably the Fujifilm then. The details I cannot describe as either mushy, nor sharp, and there isn't a ton of subtlety of the color. Probably very minimal processing was done. It feels a bit more "raw", though I'm not really familiar with how Fuji does their in-camera processing, though most dedicated cameras I've used tend to leave heavier processing to the user's discretion (via Lightroom, Photoshop, Capture One, etc). The bottom-right of the corner of the two images (Top-Left and Bottom) also has a clear difference in sharpness, though this likely comes down to the lens.

Interesting analysis. 

I'd like to add that none of the three images in the "3 picture image" is the fuji camera. That was posted separately later. I have posted 4 images so far. 

[Zodiark1593]

6 minutes ago, LAwLz said:

Interesting analysis. 

I'd like to add that none of the three images in the "3 picture image" is the fuji camera. That was posted separately later. I have posted 4 images so far. 

Still a fun exercise.

 

Tbh, I looked at the fourth one too, but didn't think that it was for the comparison. That image shows a lot more detail, and raw noise in it, so much so that I was under the impression that it was shot closer up (and thus a higher resolution), or wasn't subjected quite to the same levels of compression (that often cuts graininess itself).

 

If I were to use that shot in the comparison, it's not even a fair fight.

 

 

[LAwLz]

20 minutes ago, Zodiark1593 said:

Still a fun exercise.

 

Tbh, I looked at the fourth one too, but didn't think that it was for the comparison. That image shows a lot more detail, and raw noise in it, so much so that I was under the impression that it was shot closer up (and thus a higher resolution), or wasn't subjected quite to the same levels of compression (that often cuts graininess itself).

 

If I were to use that shot in the comparison, it's not even a fair fight.

I'd also like to add that I didn't really intend for this to be a good comparison. I picked a random image and then cropped out one of the darker parts of it. 

 

I guess I could post pictures with more things in them to make it a more comprehensive test but I can't really be bothered. I just wanted to show that for low light images the S21 Ultra don't struggle even though it has a 108MP sensor vs the iPhone with its 12MP sensor. 

[tikker]

On 1/27/2022 at 2:37 PM, Imbadatnames said:

No it doesn’t you’re still capturing the same amount of light per pixel. Mashing the pixels together doesn’t create more data. Just adding them together doesn’t work.

Changing my camera to take a photo at half res doesn’t magically make it better in low light. 

 Not if using only half the pixels, but if you are summing the pixels and treating them as a single bigger pixel, i.e. pixel binning, then it does work that way. You are correct that you aren't creating more data by doing it. You are, however, increasing the ratio of the signal, which increases from summing, compared to the noise (be it due to low light, readout noise, whatever). This comes at the cost of losing resolution, though, because N pixels are now treated as one pixel. You thus have a choice between a noisy picture with more detail or a cleaner picture with less detail.

 

As an astronomer I can say that pixel binning most definitely works as I've used it myself. I don't know how the cameras discussed here do it, but on telescopes it's implemented on a hardware level in the CCDs and can, for example, be used on faint targets where the noise created by measuring the amount of electrons the CCD has captured (the so-called "readout noise") is large compared to the signal. Let's say that we have four pixels that collected 10 electrons each and that we have a readout noise 5 electrons. That means for an individual pixel we only have a so-called signal-to-noise ratio of 10 / 5 = 2 and that we can measure anywhere from 5 to 15 electrons upon reading out the image (a 50% uncertainty). This will give a very noisy image. With hardware pixel binning you sum all four pixels together so you have 40 electrons, but still a readout noise of 5, which means you only measure 35-45 (a much lower 12.5% uncertainty). Even sub-sampling in post works (be it perhaps to a somewhat lesser extent) as the signal is still there, but just buried in the noise at higher resolution.

 

It's similar to cooking. Measuring out 1 gram of stuff on a scale that's only accurate to 0.5 grams and cooking ten times with consistent results is difficult, but measuring out 10 grams at once and cooking it at once will give 10 consistent portions.

 

[Zodiark1593]

20 minutes ago, LAwLz said:

I'd also like to add that I didn't really intend for this to be a good comparison. I picked a random image and then cropped out one of the darker parts of it. 

 

I guess I could post pictures with more things in them to make it a more comprehensive test but I can't really be bothered. I just wanted to show that for low light images the S21 Ultra don't struggle even though it has a 108MP sensor vs the iPhone with its 12MP sensor. 

Oh absolutely, it holds up brilliantly vs the iPhone. Pixel binning definitely affords a lot of flexibility that can make the best use of a very limited sensor size. When binning pixels, you also sidestep the potential downside of diffraction as well. If the sensor readout speed is fast enough, the phone can probably even shoot images at varying levels of pixel binning to further bump up dynamic range.

[mr moose]

3 hours ago, tikker said:

 

 

As an astronomer I can say that pixel binning most definitely works as I've used it myself.

 

 

I was going to raise this example of how it works and all, but I didn't want to complicate a concept that people are already struggling with.

[tikker]

9 minutes ago, mr moose said:

I was going to raise this example of how it works and all, but I didn't want to complicate a concept that people are already struggling with.

Maybe it was a futile and complicated attempt, but I was in the mood for some argueing haha. I has even been a sanity check recently to verify I hadn't lost information so it would've been bad if it didn't work.

[mr moose]

1 minute ago, tikker said:

Maybe it was a futile and complicated attempt, but I was in the mood for some argueing haha. I has even been a sanity check recently to verify I hadn't lost information so it would've been bad if it didn't work.

I'm not completely upto date with it myself, but when I was into photography I remember talking to amateur astro-photographers and they were always talking about taking multiple long exposure photos then adding them together to get the right amount of light.  but with new hi res binning sensors that isn't required as much. 

 

 

[Arika]

9 hours ago, Imbadatnames said:

Snip

Stop grasping at straws and just answer the question.

 

The more you attempt to dodge answering lawlz, the more your argument just looks like a fanboy's ramblings.

[Imbadatnames]

17 hours ago, LAwLz said:

Then can you please elaborate?

I don’t think I can at this point you’re just not getting it 

17 hours ago, LAwLz said:

 

Because if I didn't do that you could just look at the resolution and figure out which was which. I am starting to suspect that that's why you want me to post the original as well, so that you can "cheat" and pick the iPhone image without having to judge the quality of the images.

So you’re editing the images then asking people to judge? 

17 hours ago, LAwLz said:

Also, it is 1:1, in the section I posted.

If I tear out a page of a book, would you say I have altered the text on the page? Of course not. The text is still the same. It's just not the entire book. I want you to evaluate the section I posted.

It’s not a book it’s an image. 

17 hours ago, LAwLz said:

 

 

Why does it matter if it's not technically the original anymore if what I posted is a pixel perfect representation of the original images? It's just that it's a small section of it.

You’re editing the images. 

17 hours ago, LAwLz said:

I wonder totally understand your frustration if I had resized the image, or applied more compression to it or some other destructive process, but in this case nothing like that has happened, and that was intentional.

You are literally destroying the image, you are irreversibly removing data. 

17 hours ago, LAwLz said:

 

 

ISO 1600

1/4 sec shutter speed

f/2.8 aperture

Far to fast for the image dude ISO is stupidly high too as you can see by the noise.

17 hours ago, LAwLz said:

 

The resolution of the images.

If I post the original images you can either just look at the EXIF and see "Camera maker: Apple" and "Camera maker: Samsung". Even if I scrub the EXIF you will still be able to just look at the resolution. The iPhone takes pictures with a resolution of 4032x3024. The Galaxy S21 Ultra takes pictures at 4000x3000.

To be honest the only reason I looked at resolution was because the image was clearly low res. I know the difference between Fuji pictures and phone pictures because I have a Fuji Camera anyway. 

17 hours ago, LAwLz said:

If I post the full images you get several ways of figuring out which is the iPhone image and which is the Galaxy S21 Ultra image. I created this test in a way where you can't cheat.

You didn’t create a test, you intentionally destroyed the images and took them down to 2MP which invalidates any comparison outside of colour science which is subjective, for example I bought a Fuji camera because of their colour science in JPEG files straight out of the camera. 

17 hours ago, LAwLz said:

Can we just be honest for a minute and admit that the difference in camera performance isn't that big? The Samsung phone certainly don't "struggle" with taking decent photos. It's in the like top 5 smartphone cameras that you can get. Some might prefer the iPhone's camera in some scenarios, and some might prefer the S21 Ultra's camera in some scenarios. I think there are certainly merits to these high megapixel sensors. Do I think we need 200 megapixels in a phone camera? Nope, I think that's beyond the practical limit of our current sensor tech. Do I think 108 megapixels are beyond that limit? Probably. 

You’re suddenly wanting to be honest after days of just straight up not posting original photos, claiming editing images were original photos until pressed about it and not understanding why knocking down resolution shouldn’t be done when comparing images? Oh let’s compare cameras between this smartphone from 2004 and an A7SIV, but let’s change the resolution to 1x1 first just so you don’t know by the resolution. 

17 hours ago, LAwLz said:

Of course they can't compete with a DSLR, but do phone cameras even have to? They are different categories of devices and therefore don't compete with each other.

Speaking of DSLRs, Sony have started doing pixel binning in some scenarios on their mirrorless cameras. The Alpha 7S III and the Alpha 1 do pixel binning when recording video for example since it helps with HDR recording. If you want a serious sensor then look up the IMX661. It's a large format sensor from Sony that does pixel binning too. I wouldn't be surprised if we start seeing more cameras with pixel binning in the future. It provides great flexibility and the best of both worlds (high resolution and large pixels) without that big drawbacks.

You do know the A7S range aren’t high res that’s the A7R? The Sonys are also Full frame sensors so they don’t have the issue of gathering light on a tiny phone sensor. 

[Imbadatnames]

20 hours ago, Spotty said:

The forum will display a preview image in the thread that is limited to maximum 1000 pixels width/height. If you right click and save you are only downloading the preview image. To view the full image you need to click on it first.

Already done, photo was still 2MP out of a 16MP camera. Poster is editing photos to 2MP and asking to compare it.

[mr moose]

1 hour ago, Imbadatnames said:

You didn’t create a test, you intentionally destroyed the images and took them down to 2MP which invalidates any comparison outside of colour science which is subjective,

 

 

Simply cropping a photo does not alter the pixel density, compression or any other aspect.   It simply removes parts of the image that are not selected.

 

 

[Imbadatnames]

1 hour ago, mr moose said:

 

Simply cropping a photo does not alter the pixel density, compression or any other aspect.   It simply removes parts of the image that are not selected.

 

 

But you’re also blowing it up to a point where it’s a pixelated mess, also if the photo is already edited in one way how do you know it’s not edited in another? Ontop of the utter dogshit settings they’ve used for the Fuji 

[tikker]

  

On 1/27/2022 at 3:53 PM, LAwLz said:

can you evaluate these three images? I'd like to hear your opinion on which one is the best.

They were all taken with night mode enabled, since you are so insistent on making the software as different as possible even though we are trying to evaluate hardware.

 

  Hide contents

 Just because I feel like it: all look good, but I prefer the one on the top right. Both the left one and bottom one feel rather sharpened, perhaps even oversharpened, to me. The central brick in the third row from the bottom, for example, looks more natural in the top right pic to me, but has a harsher edge in the other two. Putting the Fujifilm one next to it as a reference (assuming an actual camera camera is better here than a smarthphone camera).

 

1 hour ago, Imbadatnames said:

You are literally destroying the image, you are irreversibly removing data. 

Yes cropping removes data, but if you are not comparing that section of the image it's irrelevant information and hence has no impact on the comparison, so the full size does not matter. You don't lose information about the part you care about by cutting out the part you care about. Cutting out a section of the image does not compromise image quality in that section.

2 hours ago, Imbadatnames said:

You didn’t create a test, you intentionally destroyed the images and took them down to 2MP which invalidates any comparison outside of colour science which is subjective, for example I bought a Fuji camera because of their colour science in JPEG files straight out of the camera. 

Alright, let's do some math here to determine how "destroyed" the photos are. Remember, they are cropped not "taken down" to 2 MP:

Spoiler

I have opened each image in a new tab to ensure I have the full size versions. These are:

 

3-picture mosaic: 2498 x 1524

Fujifilm reference: 2087 x 1003

 

Now we know from this information:

On 1/28/2022 at 1:15 PM, LAwLz said:

Don't worry, the Fuji picture was not compressed to 2MP. 

I got the full image on my computer and it was that I used to crop it.

 

  

that the Fujifilm was shot at 26 MP. A 2087x1003 area crop means that the subject covers 8% of the photo.

 

Now I assume each photo with each device was shot more or less the same, such that we can assume more or less the same coverage in the final shot. Combining that with the following information:

19 hours ago, LAwLz said:

The iPhone takes pictures with a resolution of 4032x3024. The Galaxy S21 Ultra takes pictures at 4000x3000.

An 8% crop of this same subject in each area would result in 4032*3024*0.08 = 975,421 pixels, or just under 0.98 MP for the iPhone. For the Galaxy S21 Ultra the crop would be 4000*3000*0.08 = 960,000 pixels or 0.96 MP.

 

Assuming the aspect ratio of the crop has remained constant, since we know the 2087x1003 size of the Fujifilm, the aspect ratio is 2.08. We can deduce that the same 8% area crop in iPhone and S21 would be 1424x685 and 1413x680, respectively. Horizontall that adds up to 2837 and vertically to 1365. Now the Fujifilm one seems to cover a bit more area to the right, and cropping it quick-and-dirty by eye to match the tree to the others I arrive at something more like 1860 pixels wide. This changes the aspect ratio for the crop to 1860/1003 = 1.85, which would reduce the crops in the phone captures to 1343x726 (iPhone) and 1332x720 (S21 Ultra), which would sum to 2675 horizontally.

Given the white space in the mosaic, it looks like the combined length of crops could be a bit bigger than that file, but the fields of view are also slightly different, so they may just as well have been cropped to the same amount of pixels. Without knowing the exact sizes of the crops, it is hard to say if anything or how much of the photo has truly degraded. Since hey're PNGs, so I think I can assume lossless compression w.r.t. what went in. Is there information lost? Maybe a bit. Has the image been "destroyed"? Not significantly I think. Knowing what the crops are would also bias our comparisons as we could probably tell which is which by seeing the different resolutions or field of views. Based on this I think the comparison given should suffice for a forum blind test.

1 minute ago, Imbadatnames said:

But you’re also blowing it up to a point where it’s a pixelated mess, also if the photo is already edited in one way how do you know it’s not edited in another? Ontop of the utter dogshit settings they’ve used for the Fuji 

That's a trust issue and the answer will always be you don't. It comes down to what you believe the other party is trying to do. In this case it seems to be that you think they're trying to deceive you. The reality is that you either trust that what is presented to you is as described or don't. If you don't then the choice is to leave it be or to go and get the devices yourself, do the experiments yourself and present those results. That's also how science works when people don't agree on things.

 

In this case, both phones seem to be shooting 12 MP images, so they're both a similar pixelated mess resolution wise. The fine details is exactly where the differences will show up from the different processing between the phones, be it pixel binning, post-processing, night mode or whatever the phone does to the image, so small sections like these are a good place to look for comparison.

[mr moose]

9 hours ago, Imbadatnames said:

But you’re also blowing it up to a point where it’s a pixelated mess, also if the photo is already edited in one way how do you know it’s not edited in another? Ontop of the utter dogshit settings they’ve used for the Fuji 

 

This is a blind test,  it requires that you don't know anything about the photos until you have chosen which one you think is best, then you evaluate the origin of each photo and what was done to it.  You cannot prove anything by demanding to see all the details of a photo before making note of which one you think is best.

 

 

 

 

 

 

 

 

[Zodiark1593]

9 hours ago, Imbadatnames said:

But you’re also blowing it up to a point where it’s a pixelated mess, also if the photo is already edited in one way how do you know it’s not edited in another? Ontop of the utter dogshit settings they’ve used for the Fuji 

Couple things. Phones use aggressive noise reduction, giving them an artificial advantage in noise performance at a given ISO setting. (My preference, I don’t like noise reduction, I’ll take sharp grain over mush). The other, even at 1600 ISO, the Fuji still retains a hell of a lot more detail than the two phones. Even “dog” settings on the Fuji clearly blow the phones out of the water, its hardly a fair comparison. It was so much better that I didn’t think that image (albeit was uploaded separately) was part of the comparison. If anything, this cements my belief that phone cameras are still very far away from achieving the flexibility afforded to modern DSLR and Mirrorless systems. 

[LAwLz]

25 minutes ago, Zodiark1593 said:

Couple things. Phones use aggressive noise reduction, giving them an artificial advantage in noise performance at a given ISO setting. (My preference, I don’t like noise reduction, I’ll take sharp grain over mush). The other, even at 1600 ISO, the Fuji still retains a hell of a lot more detail than the two phones. Even “dog” settings on the Fuji clearly blow the phones out of the water, its hardly a fair comparison. It was so much better that I didn’t think that image (albeit was uploaded separately) was part of the comparison. If anything, this cements my belief that phone cameras are still very far away from achieving the flexibility afforded to modern DSLR and Mirrorless systems. 

Not really sure what settings would have made the Fuji picture better either.

The shutter speed was already at 1/4 second. Any longer and it would be REALLY hard to take the picture hand-held. Lower ISO? Then the image would be way darker. I guess a bigger aperture would have helped keeping the noise down but it's worth noting that the pictures are at 100% scale. Pretty much all pictures in the dark look noisy when you don't downscale them.

 

Anyway, I feel like my comparison has gotten way too serious and way too many people involved. It was just meant as a quick "I bet you can't tell the difference in these low light shots, so I don't think calling Samsung's camera shit is justified" comment.

 

 

For those wondering:

 

Top left = Galaxy S21 Ultra (Snapdragon 888)

Top right = iPhone 12 Pro

Bottom = Galaxy S21 Ultra (Exynos 2100)

 

 

Not that it's a good comparison of sensor hardware anyway since they were taken with night mode on, which means they are heavily post-processed and my original comment was suppose to only be about the sensor since Imbadatnames said Samsung struggles because of their 108MP sensors.