OK so here it is. I took some time this evening with an old Galaxy S, a new(ish) Galaxy S2, a shiny new Galaxy S3, and an iPhone 4S (for good measure).
Problem statement: Samsung has been accused of equipping some versions of the Galaxy S3 with an inferior image sensor, providing degraded low light sensitivity when compared to the Galaxy S2 and other top phones. This is bad because customers expect to be purchasing a tested and true level of quality with their new device, not buying into a game of chance "which sensor do you get with your $200 and two year contract?"
Background: Samsung, like most large companies, has a limited supply chain. Many times with a product launch such as large as this, they will take shortcuts. Usually, these shortcuts are not major problems. Apple does this all the time, they use memory modules from different manufactures, capacitors from different places, etc. Samsung does this as well. It only becomes a problem when something major, such as the camera/battery/glass/etc is enough different to change the user experience. It has been proven that on the Galaxy S2 and Galaxy S3 samsung has sourced their image sensor from two locations:
Methods: Well, this is no lab test. I tested these phones side by side, one at a time. I took at least 3 pictures for each image shown and selected my favorite. The first images use auto everything because that is going to be the most common setting (and in the case of the iPhone, the ONLY setting, minus HDR mode). Following this I enabled night mode on the androids and took another set of images. Pictures were selected with 100% bias towards what I thought looked nice overall, nothing more. No tripod was used, but I tried to keep each phone as steady as possible. It's my job to perform microsurgery (on occasion) so I think I'm qualified to hold the phone still enough. No other settings were messed with.
I did not try to find good angles for the phones, instead I placed them in situations where the lighting was poor/hellish. The pictures were all taken indoors with few/no lights on, with ambient light messing up the shots, and no flash (which would have made all of these shots 10000x better). It was overcast and late in the evening.
First off, camera firmware for the S2 and S3
The iPhone should be shaking with all this epicness in front of it
In this set, the iphone would autofocus both outside and inside so both images are shown. No matter where I tapped to focus with the Galaxy's, I could not get the chair to come out clear. I could have adjusted the settings manually I guess. I re-sized the S1 image here accidentally (it's 5MP not 8MP). The iPhone, S1 and S3 all did well, with the edge going to the iPhone and S1 (surprisingly)... the S2 and S3 were very temperamental.
This set threw in some motion. I wouldn't try to quantify it, however. When that dog gets excited there isn't a camera anywhere fast enough to avoid motion blur. The framing was difficult as I was trying to keep the dog in the picture, but anyway, the S3 on auto had my favorite results overall. With night mode enabled
the S2 and S3 both looked good, I feel the S2 took the best shot with night mode.
In this set, the S3 night mode and HDR beat everything else hands down. The S2 looked good, but I had issues keeping it still, the S3 was very sharp each time despite the increased exposure.
Getting into pure darkness without a flash (who does this???), I only compared the S2 and S3. The S1 did surprisingly well here... but it's old tech so nobody cares. The iPhone was just black screen, nothing to compare. As for my take on these... yes, the S3 is REALLY red. That means the S3 has a crappy IR filter. But all hope is not lost... anybody that knows anything about night vision knows that IR and near-IR light is the most valuable. The fact that the sensor is able to collect this much IR really helps it "see" in the dark, even if it's primarily one color. While this may cause some minor problems at normal light levels (very minor, since this light is hardly present, remember the lights are almost totally off here), by passing the S3 image through some very minor filters in photoshop the image comes out beautiful. The third image looks almost identical to how the scene looked to my own eyes, slightly yellow hue from the dim lights, redish wood table. IMO at least.
Moving into some real situations, here is a macro shot of the S1 and S3
And in finale, my dirty car. The S1/2/3 all had good color reproduction, with the S2/3 having more detail, and the S3 the best overall. The iPhone kept trying to focus on the wheel (It either thought it was a face, or Siri was getting hot for the rims) which was the cause for the bad white balance.
In conclusion, in three generations of camera's, the biggest change is in detail. In all of the S1 images, noise is present in much higher amounts, although it held in WAY better than I expected for its age. As for the S3 being worse than the S2, for at least these tests I do not believe this to be the case. If anything they are very comparable. While some see the excess near-IR on night shots as a bad thing, remember it is always a tradeoff. Less IR (red) makes the shot more balanced naturally, but since the other wavelengths will be so sparse you will have little (or no) detail. At least with the red, you can regain vasts amounts of detail, and post process a little bit later to get a very nice picture. On a side note, I noticed the S3 compresses its pictures more than both the S2 and the iPhone4S. The only phone with smaller images was the S1, and it has a 5mp camera. This may not be a factor for anything, but I did find it interesting.
To not offend any apple fans, considering the fact they have zero control over the functioning of the camera, it did pretty well. Had the flash been used (like they should have been), I believe the S3 and iPhone would have been pretty much tied. I don't like apple, but I did try my best to be fair and take good pictures with all phones.
I do apologize for not having the GDxxxx firmware S3, as many were looking forward to seeing it in this comparison. I hope this helps somebody out there regardless. Lastly, unadulterated pictures can be found https://picasaweb.google.com/117440914475223218617/CameraComparison?authuser=0&feat=directlink with all the exif data you could ever need if you can sort through them.
Problem statement: Samsung has been accused of equipping some versions of the Galaxy S3 with an inferior image sensor, providing degraded low light sensitivity when compared to the Galaxy S2 and other top phones. This is bad because customers expect to be purchasing a tested and true level of quality with their new device, not buying into a game of chance "which sensor do you get with your $200 and two year contract?"
Background: Samsung, like most large companies, has a limited supply chain. Many times with a product launch such as large as this, they will take shortcuts. Usually, these shortcuts are not major problems. Apple does this all the time, they use memory modules from different manufactures, capacitors from different places, etc. Samsung does this as well. It only becomes a problem when something major, such as the camera/battery/glass/etc is enough different to change the user experience. It has been proven that on the Galaxy S2 and Galaxy S3 samsung has sourced their image sensor from two locations:
Methods: Well, this is no lab test. I tested these phones side by side, one at a time. I took at least 3 pictures for each image shown and selected my favorite. The first images use auto everything because that is going to be the most common setting (and in the case of the iPhone, the ONLY setting, minus HDR mode). Following this I enabled night mode on the androids and took another set of images. Pictures were selected with 100% bias towards what I thought looked nice overall, nothing more. No tripod was used, but I tried to keep each phone as steady as possible. It's my job to perform microsurgery (on occasion) so I think I'm qualified to hold the phone still enough. No other settings were messed with.
I did not try to find good angles for the phones, instead I placed them in situations where the lighting was poor/hellish. The pictures were all taken indoors with few/no lights on, with ambient light messing up the shots, and no flash (which would have made all of these shots 10000x better). It was overcast and late in the evening.
First off, camera firmware for the S2 and S3
The iPhone should be shaking with all this epicness in front of it
In this set, the iphone would autofocus both outside and inside so both images are shown. No matter where I tapped to focus with the Galaxy's, I could not get the chair to come out clear. I could have adjusted the settings manually I guess. I re-sized the S1 image here accidentally (it's 5MP not 8MP). The iPhone, S1 and S3 all did well, with the edge going to the iPhone and S1 (surprisingly)... the S2 and S3 were very temperamental.
This set threw in some motion. I wouldn't try to quantify it, however. When that dog gets excited there isn't a camera anywhere fast enough to avoid motion blur. The framing was difficult as I was trying to keep the dog in the picture, but anyway, the S3 on auto had my favorite results overall. With night mode enabled
the S2 and S3 both looked good, I feel the S2 took the best shot with night mode.
In this set, the S3 night mode and HDR beat everything else hands down. The S2 looked good, but I had issues keeping it still, the S3 was very sharp each time despite the increased exposure.
Getting into pure darkness without a flash (who does this???), I only compared the S2 and S3. The S1 did surprisingly well here... but it's old tech so nobody cares. The iPhone was just black screen, nothing to compare. As for my take on these... yes, the S3 is REALLY red. That means the S3 has a crappy IR filter. But all hope is not lost... anybody that knows anything about night vision knows that IR and near-IR light is the most valuable. The fact that the sensor is able to collect this much IR really helps it "see" in the dark, even if it's primarily one color. While this may cause some minor problems at normal light levels (very minor, since this light is hardly present, remember the lights are almost totally off here), by passing the S3 image through some very minor filters in photoshop the image comes out beautiful. The third image looks almost identical to how the scene looked to my own eyes, slightly yellow hue from the dim lights, redish wood table. IMO at least.
Moving into some real situations, here is a macro shot of the S1 and S3
And in finale, my dirty car. The S1/2/3 all had good color reproduction, with the S2/3 having more detail, and the S3 the best overall. The iPhone kept trying to focus on the wheel (It either thought it was a face, or Siri was getting hot for the rims) which was the cause for the bad white balance.
In conclusion, in three generations of camera's, the biggest change is in detail. In all of the S1 images, noise is present in much higher amounts, although it held in WAY better than I expected for its age. As for the S3 being worse than the S2, for at least these tests I do not believe this to be the case. If anything they are very comparable. While some see the excess near-IR on night shots as a bad thing, remember it is always a tradeoff. Less IR (red) makes the shot more balanced naturally, but since the other wavelengths will be so sparse you will have little (or no) detail. At least with the red, you can regain vasts amounts of detail, and post process a little bit later to get a very nice picture. On a side note, I noticed the S3 compresses its pictures more than both the S2 and the iPhone4S. The only phone with smaller images was the S1, and it has a 5mp camera. This may not be a factor for anything, but I did find it interesting.
To not offend any apple fans, considering the fact they have zero control over the functioning of the camera, it did pretty well. Had the flash been used (like they should have been), I believe the S3 and iPhone would have been pretty much tied. I don't like apple, but I did try my best to be fair and take good pictures with all phones.
I do apologize for not having the GDxxxx firmware S3, as many were looking forward to seeing it in this comparison. I hope this helps somebody out there regardless. Lastly, unadulterated pictures can be found https://picasaweb.google.com/117440914475223218617/CameraComparison?authuser=0&feat=directlink with all the exif data you could ever need if you can sort through them.
Last edited:
