Moving to big pixel size 1.55um gets best camera review for Pixel phone. It seems HTC10 is also same sensor which is the second highest score in DXO list. May be a pixel size increase trend will start.
Well, not really. Phones already have more or less maximised sensor size in respect to their thickness. And I doubt anyone will go under 12M pixels because of high demand for both 4K video and 3:2 stills. That sets the low bar at around 12M pixels.
The only resonable outcome is to have more of these cameras for various reasons. 3D, multiple fields of view, better signal to noise ratio, Moire-reduction (three single color channel cameras) and such.
I still don't get this high res trend in phones. My apsc mirrorless 24Mpixel camera with kit lens has an effective resolution of 5-6 Mpixel (according to DXO). What is the effective resolution of a smartphone camera with small plastic lens? 1-2 MPixel? Why having 4k video sensor in that case while we are limited by the plastic lens? Is it just marketing or am I missing something here?
I agree that its not much real value, and mostly marketing driven. But they dont have much of an alternative. Customers want 4K, camera manufacturers and image sensor manufacturers have to deliver or loose customers.
Is DxOMark truly reliable? Looks like they are for-profit software consultancy for camera makers, make them a profit company. Will assume that they have the power to silently shame the companies that choose not to license its software or services.
I followed the DXO review 2 years ago for a Sony phone. The score was the highest they had way above Samsung. I bought the phone and the camera was the poorest I have seen despite having 20 mpix. I don't trust their reviews. It's rates high the highest bidder irrespective of any image quality.
Depends on which sony phone you talk about. The z series always had some of the best cameras out there and surely the best sensor. The problem was a poor software behind it (wrong WB, wrong exposure, missed focus, etc.). In a controlled test environment I'm sure those phones delivered the best performance.
I don't think the overtly change results but they definitely have a habit of focusing on performance criteria that are in line with their top providers. Also its clear that camera companies are cooking Raws as the same sensor can get different results in different cameras and DxO is not taking that into account.
This manipulated, false camera jitter is not realistic and is optimized to match this 200Hz gyroscope for this demonstration. Who jitters their camera around like that in handheld? OIS is optimized for normal hand held movements and probably would not reduce this amount of falsely imposed fake jittering altogether. 200 Hz gyroscope with its lack of subtlety compared to OIS shows better in this video, but not in real life circumstances. I hate these kinds of manipulated comparisons, and DxO, shame on you, you know this was a manipulated scenario. I know you know better. Looks like Google just placed the cameras on top of the car and rode over rough streets in SF or had a person in the sun roof shake the heck out of the system. This demo from the people who co-invented early autonomous vehicles and they can't even provide a realistic demo of normal human camera shake?
I conduct these kinds of tests daily. IMHO, it should have been mentioned from DxO. Like I said, they know better. DxO had the total platform to test, not just the sensor.
The EIS for video is limited to FullHD. Also, to my eyes it didn't seem too impressive. The white balance and exposure seem spot on. Though this sensor seems promising. I am guessing that Samsung would do an even better job. I find their photo quality stellar. I hope Samsung make a version of this with better HDR, OIS and with that f1.7 lens. The Google Pixel is also grossly over-priced considering its poor design, lack of weatherproofing and expandable memory. It will most likely fail as a smartphone. Sad, considering that the IMX378 sensor is a PDAF version of the IMX377, thus having much faster autofocus and thus better at HDR too.
Anyone know if the EIS is working for live view or for post processing after video has been captured. If it is for live view, i think this performance is amazing.
Which sensor is used?
ReplyDeleteThere is no official word on that. Unofficially, many websites say it's Sony IMX378 in the rear camera and IMX179 in the 8MP front camera.
DeleteOops, it's official, on Google Pixel site:
Deletehttps://madeby.google.com/phone/specs/
Moving to big pixel size 1.55um gets best camera review for Pixel phone. It seems HTC10 is also same sensor which is the second highest score in DXO list. May be a pixel size increase trend will start.
ReplyDeleteWell, not really. Phones already have more or less maximised sensor size in respect to their thickness. And I doubt anyone will go under 12M pixels because of high demand for both 4K video and 3:2 stills. That sets the low bar at around 12M pixels.
DeleteThe only resonable outcome is to have more of these cameras for various reasons. 3D, multiple fields of view, better signal to noise ratio, Moire-reduction (three single color channel cameras) and such.
I still don't get this high res trend in phones. My apsc mirrorless 24Mpixel camera with kit lens has an effective resolution of 5-6 Mpixel (according to DXO). What is the effective resolution of a smartphone camera with small plastic lens? 1-2 MPixel? Why having 4k video sensor in that case while we are limited by the plastic lens? Is it just marketing or am I missing something here?
DeleteI agree that its not much real value, and mostly marketing driven. But they dont have much of an alternative. Customers want 4K, camera manufacturers and image sensor manufacturers have to deliver or loose customers.
DeleteEIS
ReplyDeleteI wonder how it compares at x2 zoom with iPhone 7plus in terms of resolution.
ReplyDeleteWhich translates to: I wonder how it compares to the magnifying lens of iPhone 7plus when it tries to beat it with basic post-processing.
DeleteQuite a Apples/Peas comparison...
This comment has been removed by the author.
ReplyDeleteWhere does EIS solution come from? MSM8996 pro? or Google Algorithm?
ReplyDeleteI really want to know the other Phones using different AP will use that someday.
Is DxOMark truly reliable? Looks like they are for-profit software consultancy for camera makers, make them a profit company. Will assume that they have the power to silently shame the companies that choose not to license its software or services.
ReplyDeleteI followed the DXO review 2 years ago for a Sony phone. The score was the highest they had way above Samsung. I bought the phone and the camera was the poorest I have seen despite having 20 mpix. I don't trust their reviews. It's rates high the highest bidder irrespective of any image quality.
DeleteDepends on which sony phone you talk about. The z series always had some of the best cameras out there and surely the best sensor. The problem was a poor software behind it (wrong WB, wrong exposure, missed focus, etc.). In a controlled test environment I'm sure those phones delivered the best performance.
DeleteI don't think the overtly change results but they definitely have a habit of focusing on performance criteria that are in line with their top providers. Also its clear that camera companies are cooking Raws as the same sensor can get different results in different cameras and DxO is not taking that into account.
DeleteThis manipulated, false camera jitter is not realistic and is optimized to match this 200Hz gyroscope for this demonstration. Who jitters their camera around like that in handheld? OIS is optimized for normal hand held movements and probably would not reduce this amount of falsely imposed fake jittering altogether. 200 Hz gyroscope with its lack of subtlety compared to OIS shows better in this video, but not in real life circumstances. I hate these kinds of manipulated comparisons, and DxO, shame on you, you know this was a manipulated scenario. I know you know better. Looks like Google just placed the cameras on top of the car and rode over rough streets in SF or had a person in the sun roof shake the heck out of the system. This demo from the people who co-invented early autonomous vehicles and they can't even provide a realistic demo of normal human camera shake?
ReplyDeleteIf you mean the comparison video, I think it came straight from Google. DxO does not have any relation to it.
DeleteI conduct these kinds of tests daily. IMHO, it should have been mentioned from DxO. Like I said, they know better. DxO had the total platform to test, not just the sensor.
ReplyDeleteThe EIS for video is limited to FullHD. Also, to my eyes it didn't seem too impressive.
ReplyDeleteThe white balance and exposure seem spot on. Though this sensor seems promising. I am guessing that Samsung would do an even better job. I find their photo quality stellar.
I hope Samsung make a version of this with better HDR, OIS and with that f1.7 lens. The Google Pixel is also grossly over-priced considering its poor design, lack of weatherproofing and expandable memory.
It will most likely fail as a smartphone. Sad, considering that the IMX378 sensor is a PDAF version of the IMX377, thus having much faster autofocus and thus better at HDR too.
This comment has been removed by the author.
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteAnyone know if the EIS is working for live view or for post processing after video has been captured. If it is for live view, i think this performance is amazing.
ReplyDelete