Sony Xperia 10 iii for Video

[VidThreeNorth]

5G Phone

A few years ago, Sony re-organized their divisions and placed its money losing smartphone business under their successful camera division. At the time there was skepticism about why this was done. Looking back, even then, low cost camera sales were falling as people replaced them with their smartphones. That was because smartphone cameras were improving to the point that a lot of people were sufficiently satisfied with results that their smartphones produced. The strategy shifted and three new lines resulted: The Xperia 1, 5 and 10 series products. This last year, after the Xperia 1 ii (second generation) had been selling, Sony introduced their "over the top - line" Xperia Pro. April of 2021, Sony announced their latest upgrades, which included Xperia 1 iii, Xperia 5 iii and Xperia 10 iii. The Xperia Pro continues as-is for now.

With the coming of 5G, in theory, it will not be necessary to upgrade past my 4G HSPDA phones, but experience has indicated to me that eventually the phone companies will attempt to push people into upgrading. That is my first, and in itself, a sufficient motivation to look for a 5G phone. A second reason for upgrading was my past experience with trends in live music concerts. I have attended concerts where "real cameras" were not allowed, but photography and videos using smartphones was acceptable. The future for this is unclear, but for now, I felt I might benefit from an upgrade in my cellphone cameras. I could not afford a top-tier smartphone/camera, but there were some middle range smartphones with 5G and theoretically better video cameras than I had. As for timing, the highest prices phones have supported 5G for over a year now, and phones in the "moderate" price levels have been coming out. Considering all this, I decided to "get it over with".

My early choice was the Sony Xperia 10 iii, but I was not entirely happy with the price of the Xperia 10 iii, and I almost changed my mind. LG has recently announced that they will no longer sell cellphones. This does not mean the warranties will not be honored. The company is still established. But I thought that their phones might be discounted. I hunted around for an LG, but the prices did not really change. They were actually, very low priced in the first place. In the end, I decided to stick with the Sony.

Again, ironically, I am in no hurry to complete a report about this phone and camera. For now, I will continue my look at the Sony Xperia Z5c (which is almost completed now). If you have a specific question about this camera for video, post it, and if I know the answer I will post it.


NOTE: Sample Images are 1600 pixels wide in order to allow C1 (most detail) compression.

Sample Video:
- Stabilization OFF
- Default settings

"MOV_0003.mp4" [not posted]
size 769,510,277 bytes
Modified: July 10, 2021, 15:15:11
Win 8.1 data
Length 02:27
Frame width 3840
Frame height 2160
Data rate 41,699 kbps
Total bitrate 41,856 kbps
Frame rate 30 frames/second (!)
Audio
Bit rate 155 kbps
Channels 2 (stereo)
Audio sample rate 48 kHz

VLC:
Stream 0
Type Video
Codec: H264 - MPEG-4 AVC (part 10) (avc1)
Language: English
Resolution: 3840x2178
Display resolution 3840x2160
Frame rate: 30.053077
Decoded format Planar 4:2:0 YUV full scale
Stream 1
Type Audio
Codec MPEG AAC Audio (mp4a)
Language: English
Channels: Stereo
Sample rate: 48000 Hz.

One of my basic tests is to record scene with moving water in the foreground taking up the lower half of the composition, and fairly highly detailed but fairly still land (usually a river bank) filling the upper half of the frame. Moving water is a hard challenge for video compression and the result is usually filled with artifacting. Because of processing stress of the constantly changing water, often the upper half detail suffers. It is like the processing is being "sucked away" by the moving water. This location did not give me the right composition for such a test, but the principles are there.

2022-0216 22:55 NOTE: On re-reading GSMarena's report for this phone, and looking at my images again, it appears that the Xperia 10 iii supports UHD video ONLY using the main camera. All the zooming is digital based on its 12 MP sensor, which is why it "matches" the un-zoomed image so well, and as noted before, there is no "Ultra-wide" UHD video at all. I will correct my comments below to reflect this.

"01-SX10iii-0003-UHD-1x-20h06m42s141.png" [not posted]

"1x" is the main camera fully used (no zoom). The "ultra-wide" 0.6 zoom is not supported in UHD. This is a surprise since the 8MP sensor sounds like it might be the same as the sensor for the "telephoto" camera.

"01a-SX10iii-0003-UHD-1x-Rsz1600-C1.jpg"

I particularly like the render of the main camera. It is not as overly "contrasty" or over saturated or over sharpened as I have seen in some other "consumer grade" cameras. The low bit rate and resulting loss of image quality is a bit sad, but if I had to, I could release this on YouTube without embarrassment.

"01b-SX10iii-0003-UHD-1x-Crop1600-C1.jpg"
Crop from 1700,900

I framed the "detail crop (b)" to put a small tree in the upper left corner. That tree has lost detail, even compared to trees behind it, indicating that it is not a depth of field issue. The low bit rate (about 42 mbps) means this, and the arti-facting in the water are to be expected.


"02-SX10iii-0003-UHD-2x-20h17m41s896.png" [not posted]

"02a-SX10iii-0003-UHD-2x-Rsz1600-C1.jpg"
2022-0216 CORRECTION: All the UHD video for this phone is done using the main camera. All the zooming is digital based on its 12 MP sensor. It looks good up to 2x, and falls apart as expected beyond that magnification.

"02b-SX10iii-0003-UHD-2x-Crop1600-C1.jpg"
Crop from 1600,610
[Revised 2022-0216]
I put the same tree in the upper left corner again. Over-sharpening is just starting to show up.


"03-SX10iii-0003-UHD-4x-20h26m27s970.png" [not posted]
[Revised 2022-0216]
4x is further "digitally zoomed". I generally find 4x is my personal limit. Beyond this point it gets progressively worse. The maximum zoom on this camera is "10x", which is too much.

"03a-SX10iii-0003-UHD-4x-Rsz1600-C1.jpg"
Ironically, resizing this sample frame down to 1600 un-does the problems caused by scaling it up in the first place.

"03b-SX10iii-0003-UHD-4x-Crop1600-C1.jpg"
Crop from 3800,330
[Revised 2022-0216]
As you might guess, the problems of resizing are evident in this detail crop. The sharpening in particular , though not the worst I have seen, is present. Arti-facting in the water tends to increase too, though somehow, Sony seems to have softened the effect a bit.

[2021-10-10 21:54: Sony Xperia 10 iii Stills is covered in "Sony Xperia 10 iii Stills",
"Sony Xperia 10 iii Stills"]

 

[VidThreeNorth]

Sony Xperia 10iii Report pt02

[2021-08-12 14:06 Re-write: I originally planned to post my test notes roughly as they were made, but "Comment 2" was a combination of notes from at least three different days and it turned out a mess. I have done some re-organizing and re-writing of it for clarity. I have tried not to add too much new info in it at this time because I still think it would be better for me to post my findings as they occur, but I might eventually change my mind about that. There seems to be a lot that needs to be covered just to put specific issues in context. We'll see. . . .
2021-08-19 Separated the Exposure notes from the Focus notes.]

It looks to me like the Xperia 10 iii video capabilities are aimed at what Sony believes are what relative beginners and lower end amateurs would want. To that extent I am not surprised. There are some choices that I would not have made. I do not believe that a beginner would want a product with no future in it. There needs to be adequate room for growth.

The Full HD (aka 1080p) capabilities are, as expected, more complete than the UHD capabilities. Both target a 30 fps frame rate and only 8-bit 4:2:0 YUV are supported for all modes. Digital stabilization is available for both Full HD and UHD and I do think that this was a necessity for amateur and beginner users. I would expect a serious videographer to carry a 'pod everywhere. I do, but that's my point -- I'm already a "serious videographer".


Specific Features:

File Formats, Compression Methods and Combinations (aka Limits):

Video Sizes
- 4K (16:9)
- 4K (21:9)
- Full HD (16:9)
- Full HD (16:9 60 fps)
- 1080x1080 (1:1)
- HD
- VGA

Except for Full HD @ 60 fps, I presume that all frame rates are nominally 30 fps. So far, I have only tested 4K (16:9) (both H.264 and H.265) and Full HD @ 30 and 60 fps.

I have only tested stabilization for 4K (16:9) and Full HD (30 and 60 fps). I will try to update my experiences with other formats if and when I use them, but I have no current intention to try to cover them all.

UHD can be recorded in H.264 or H.265 (sometimes called HEVC). This is my first camera that supports this compression format. But it is limited to only UHD. If you switch to Full HD, then the selection does not change, but the recording is automatically in H.264.

[2022-02-17 08:48 clarified and corrected]
In UHD, There is no video support for the "Ultra-wide" camera at all -- only the "Normal" camera is supported.

[2022-02-17 08:48 New]
For Full HD, all three cameras are supported, but with a catch. I tested this using a finger to block each camera, to be sure what was going on, and it turned out a bit trickier than I expected. On the right side of the screen view, there is a small circle with a number in it, indicating the Zoom state. If you touch the circle and drag up and down, the zoom will adjust smoothly. If you drag the zoom up from 1x through 2x, the main camera will zoom up to 2x and past. If you tap on the circle it will cycle between 0.6x (ultra-wide), 1x (normal) and 2x (tele) in jumps. In this case, the 2x is the telephoto camera, and zooming past 2x is the telephoto camera being zoomed. At least this is usually what happens. I did find that occasionally, the main camera was zoomed 2x when tapping the circle, which might be considered a bug, but usually it switched cameras.

UHD also disables Tracking auto-focus, leaving only "general" autofocus capability.

Apparently "Smile Shutter" does not work in 4K modes, but I have not tried "Smile Shutter", so I cannot confirm this.

Stabilization:

[2022-0217 09:00 Correction and expansion]
Stabilization does not work in a "factory fresh" state. To get it to work requires "training". Turn on the video camera with stabilization ON and then wave the phone/camera around in the air in an "infinity" pattern ( an "8" on its side). This takes a minute or two.

Even then, the stabilization does not quite work, at least not in 4K when zoomed out past 4X. It might have some other limits too. It failed in my own tests which resulted in recordings with vertical oscillations.

When system firmware is updated, the stabilization data is lost and the camera/phone must be trained again.


It's "Android 11 . . . ."

Maybe someday Sony will fix this stabilization problem, but it brings up another strangely bad choice by Sony. The reporting of the version of system firmware is limited to saying that it is "Android 11". Specific revisions in a format like "8.1.1" used commonly in the industry is not being followed. So I cannot tell you that "it's a problem in 11.1.0" so that you can check easily if you have later firmware which might be fixed.
[2021-0812 14:08 clarified]

Screen Brightness:

According to GSMarena.com the screen seems to provide 343 NITS brightness (or maybe 345 -- their report gives both results but in different locations) when manually adjusted, or 559 NITS when brightened automatically due to ambient conditions.

Focusing:

Autofocus seems to be slow and not always reliable. I have not tested enough to isolate conditions for this problem.

There is no focus peaking and no focus area magnification, but then again, there is no manual focus, so that much is not something one might expect.

As mentioned above, Full HD (and the other "smaller formats") features Tracking auto-focus, while UHD is limited to a "general" auto-focus. The Full HD tracking focus works generally well but I need to test it more. For UHD it seems to prefer the middle of the screen, but I have seen cases where it preferred the foreground (bottom of the screen).
[2021-08-19 02:36 separated out the Exposure notes.]

Exposure:

EV compensation is controlled by a vertical slider on the screen. Range is not marked, but for stills it is marked EV -2 to EV +2. I assume that this is the same for video. Although I have not done specific testing yet, the results seem to indicate that this is correct. There are also no zebras, histograms nor waveforms to help exposure adjustment, and those should still be helpful since EV compensation is still available.

Orientation:

There is no indicator for vertical alignment, which even my old Z5c has.

Heat and Battery Life:

I am still early in the testing of this phone camera, but there is no indication so far that heat or battery life is a particular problem.

Gamma and Colours:

There is some indication that I need to look more closely at the gamma curve(s) in the Xperia 10 iii. I would have done that eventually anyway, but there seems to be "something going on there" that I will need to find out about.

Other Features:

One nice touch which is mainly cosmetic is a blue filter over the lenses. Clearly this was done just to make the phone look a bit unique and "cool". If you understand camera sensors you will know that actually adding a bit of a blue filter like this is actually irrelevant because the individual RGB filters used by the sensor array cover the necessary separations. Also, this "cosmetic" filter is very weak compared to the dynamic range of the cameras. But yes, I do have to admit that I like the slightly understated look it gives the phone.


Anyway, after looking at the results and prepping some files to upload, a couple of points are clear:

1. I really would prefer a camera with more basic controls available. Many control options are available for stills but are not available for video. That includes contrast and selectable focus tracking, and saturation.

2. The gamma seems to be modified by very strong "tinkering" at both the high and low ends. I am not sure how I feel about that, even on its own, but it would be hard for me to "match" it on my other cameras. Some of my other cameras can produce similar curves, but I tend not to use them.

The general image quality is "ok", and as I wrote before, in some cases, I would not be embarrassed by them if I uploaded them to YouTube. But it is the "in some cases" that bothers me. It can be a problem of probabilities:

- if the subject matter is not too far
- if the weather is about right -- an experienced photographer will understand that a bright sunny day is not necessarily a "good" thing
- if I can stabilize the phone (a phone is not a good shape for holding horizontally without something to brace it with, particularly for telephoto work)
- if I can see the screen (it is the brightest I have owned so far, but it is still not among the brightest available).
And in the end: if the bitrate covers me.


Bitrates:

The bitrate problem was a surprise, so that became the target of my first "detailed look." And all this work was done with a final question overall: Is this a better phone/camera for video than what I already have?

Sony Xperia 10iii Screen Captures:
[2022-02-17 14:23 added screen captures]

"01-DragZoom.jpg"
- The on-screen zoom control when dragging it up or down.

"02-TapZoom.jpg"
- The on-screen zoom control when tapped (jumping cyclically 0.6x, 1x and 2x).

 

[VidThreeNorth]

The Bit Rates:

Right now, I am concentrating on the H.264 bitrate at 4K. This is something that has been a particular annoyance for me while I have been researching these phones. The reports are being written by people who do not understand the importance of bitrates, and bitrates are generally not even reported. Generally? I have not personally seen it mentioned at all -- anywhere.

First, I can report that the bitrates I have seen so far, for clips longer than 2 minutes (and even for clips longer than 1 minute), I have not seen a clip higher than 43 mbps/sec., which is just a bit higher than I reported before. I have made enough recordings now, that I am confident that I do not expect to see a higher number in the future without a software change. If you want to leave a bit of headroom for the future, let's call it 45 mbps. Looking at my other cameras, at 24 and 30 fps, UHD (3840 wide), H.264, my Sony a6400 records at 60 and 100 mbps, my Panasonic G85 records at 100 mbps, my Sony FDR-AX53 records at 60 and 100 mbps, and my Yi-M1 records at 80 mbps. In fact, the most up to date cameras are recording H.264 UHD at 120 mbps and only a bit higher. The Olympus OM-D EM-1 mk iii claims much higher but the results are are far below the claimed maximum.
[2021-08-19 02:04 corrected many typos.]


So what happens below 45 mbps?
[Note: Sample files are generally 1600 pixels wide, and 900 high.]

Compared to Sony a6400 w/Sigma 30mm F2.8 lens
[2021-08-14 14:33 Correction: I forgot that I changed the location of my custom profile from PP10 to PP9. This control video was recorded using Default PP10 which is HLG2 with BT2020.]

The "control" sample clip was recorded using a Sony a6400 at 30 fps and 100 mbps using default Picture Profile 10, which is HLG2 gamma with BT2020 colours. This was actually not what I intended to do, but I suppose that it might be slightly more helpful since it is a more typical setting than what I had intended.


Sony a6400 clip:

"C0001.MP4"
Size: 1,686,155,057 bytes
Created July 28, 2021, 16:44:55
Video
Length 02:14
Frame width 3840
Frame height 2160
Data rate 98,816 kbps
Total bitrate 100,353 kbps
Frame rate 29 frames/second [sic]
Audio
Bit rate 1537 kbps
Channels 2 (stereo)
Audio sample rate 48 kHz

VLC:
Stream 0
Type: Video
Codec: H264 - MPEG-4 AVC (part 10) (avc1)
Resolution 3840x2178
Display Resolution 3840x2160
Frame rate 29.970029
Decoded format Planar 4:2:0 YUV
Stream 1
Type Audio
Codec PCM S16 DB (twos)
Channels: Stereo
Sample rate: 48000 Hz
Bits per sample: 16

Frame Capture:

"01-Sa64k-C0001-21h01m32s927.png" [Not Posted]
Size 12,352,670 bytes
Created July 29, 2021, 21:01:34
Dimensions: 3840 x 2178
Bit depth 24

"01a-Sa64k-C0001-Rsz1600-C1.jpg"
Size: 1,439,419 bytes
Created: July 31, 2021, 00:58:13
Program name Paint Shop Pro 19.00
Dimensions 1600 x 908
Bit depth 24
Resolution unit 2

Processed: [Corel PaintShop Pro X9]
Smart Photo Fix

Brightness
Overall 0 [Recommended 28]
Shadows -20 [Recommended-50]
Highlights 30 [Recommended 20]
Saturation 24 [Recommended 12]
Focus 51
White Balance No
Black 10
White 18

"01x-Sa64k-C0001-Processed.png" [Not Posted]
Size 20,192,421 bytes
Created July 31, 2021, 01:09:20
Dimensions: 3840 x 2178
Bit depth 24

"01xa-Sa64k-C0001-ProRsz1600-C1.jpg"
Size: 1,822,375 bytes
Created: Jul 31, 2021, 01:10:54
Program name Paint Shop Pro 19.00
Dimensions 1600 x 908
Bit depth 24
Resolution unit 2

Detail Crops:

This set of files is actually a good long term reference beyond this particular report, so I am including two "detail crops". These can be re-assembled the first directly above the second for a picture 1600 pixels wide by 1800 pixels high.

"01xb-Sa64k-C0001-ProCrop1600-C1.jpg"
Size 1,879,120 bytes
Created July 31, 2021, 01:20:38
Program name Paint Shop Pro 19.00
Dimensions 1600 x 900
Bit depth 24
Resolution unit 3

1st Crop start @ 2450,1130

The 1st detail crop shows the trees, grasses, shadows and light sand coloured rocks in the river. There is also fairly "flat" water to the left and rapids to the right. By itself it gives a wide range of situations where the effect of the bitrates and compression operate.

"01xc-Sa64k-C0001-ProCrop1600-C1.jpg"
Size 1,753,4004 bytes
Created July 31, 2021, 01:26:17
Program name Paint Shop Pro 19.00
Dimensions 1600 x 900
Bit depth 24
Resolution unit 3

2nd Crop start @ 2450,1131

The 2nd detail crop shows mainly the river rapids. Because so many pixels change between frames and have to be rebuilt, even at 100 mbps the result is a lot of artifacting which in this case, sort of resembles granular noise. The main thing to notice is that the artifact blocks are fairly consistently small.


The "Normal" Camera:

"02-SX10iii-0004-1X-21h44m21s879.png" [Not Posted]
Size 12,753,070 bytes
Crated July 29, 2021, 21:44:22
Dimensions 3840 x 2178
Bit depth 24

"02a-SX10iii-0004-1X-Rsz1600-C1.jpg"
Size 1,715,512 bytes
Created July 31, 2021, 00:46:54
Program name Paint Shop Pro 19.00
Dimensions 1600 x 908
Bit depth 24
Resolution unit 3

"02b-SX10iii-0004-1X-Crop1600-C1.jpg"
Size: 1,564,759 bytes
Created July 31, 00:45:46

Detail Crop from 2150,1150

In the 1X frame with EV=0, I found some pixels as high as R255, G255, B240. There might be some even worse. With 2 of 3 channels clipped, I would count those pixels as absolutely clipped. In the shadows, I found some R0, G2, B5, down to R0, G3, B0, which shows the scene exceeded the dynamic range of the profile (and possibly the sensor itself). I think that I would have preferred an exposure around EV ~ -0.67. This is where I miss having something like a histogram or zebras to help me get the exposure right. I do think that the camera might be automatically adjusting the gamma curve a bit at the ends to hold off clipping. When comparing this result with the "2x" clip where I reduced exposure by about 1 stop, either some of the rocks in this capture should have clipped white (R255, G255, B255) or the shadows in the "2X" segment should have hit R0, G0, B0. Having neither result is "suspicious".

I chose the detail crop to put a rock near the bottom right corner that shows a bit of macro-blocking at around 1445 x 730. Otherwise, this frame turned out quite good. Looking at the rapids (which I only have a small mount in the detail crop), the artifacting is worse "here and there" but actually not as bad as I was expecting.

[2022-0216 23:17 As mentioned previously, upon re-reading GSMarena's report and looking at my images, it apparently, the Xperia 10 iii supports UHD video ONLY using the main camera, digitally zoomed. The "tele" camera is NOT used. I will try to correct m comments to reflect this.]

"03-SX10iii-0006-2X-21h33m05s377.png" [not posted]
Size: 10,921,823 bytes
Created July 29, 2021, 19:13:16
Dimensions: 3840 x 2178
Bit depth 24

"03a-SX10iii-0006-2X-Rsz1600-C1.jpg"
Size: 1,537,905 bytes
Created: July 31, 2021, 00:55:27
Program name Paint Shop Pro 19.00
Dimensions: 1600 x 908
Bit depth 24
Resolution unit 3

"03b-SX10iii-0006-2X-Crop1600-C1.jpg"
Size: 1,141,002 bytes
Created July 31, 2021, 00:53:43
Program name Paint Shop Pro 19.00
Dimensions: 1600 x 900
Bit depth 24
Resolution unit 2

Detail Crop from 2400,1300

In the 2X frame: I dropped the exposure to -1.0 in order to contain the highlights. Even still, I found at least one R255, G255, B244 and at the other end, in the shadows, I found some R5, G11, B3 down to R6, G13, B0. So looking at the contrast in the "middle range", I think that there was some automatic "tinkering" at the far ends of the gamma to hold the image in-bounds. Stated another way, the lack of outright clipping might indicate that the gamma is being adjusted automatically to some degree to try and keep the far ends inbound. I think I would rather have a "flatter response" and a real manual contrast control. Sharpening is more evident in this frame than the "1X" frame, but it is generally not excessive.

These clips were all recorded at about 16:00 "standard" time (17:00 "daylight saving" time), which is not that close to "golden hour". If you look for the shadows, you can tell that the sun was still fairly high. The Xperia 10 iii seems to render a bit warm. My personal adjustment of the a6400 clip was more neutral. Compared to the Xperia choices, it seems "cool" but actually I do not think it was.

Conclusions:

The main point of this test was to see if the low bitrate was going to cause problems, or whether it is "generally adequate". First, putting this into the context of the whole camera, I expect that a low bitrate will be less of a problem if the contrast is high, and motion is low. In my a6400 recordings I am keeping contrast on the low side, so the higher bitrate helps prevent "macro-blocking". The Xperia 10 iii does not have manual contrast controls (at least not that I have found so far) and seems to be fitting the gamma to create a "rec709" look. The resulting gamma slope is probably helping to reduce macro-blocking in particular. Looking at my previous posts and this set, so far, I like the results generally, though I wonder about how much leeway I have to make adjustments in post. At this point, I am also wondering if there is much recording time adjustments I have available too.

Again, looking at this test, because of the "rapids", it is a sort of "torture test" for codecs and bitrates. While the results were not horribly bad, I do count that small amount of "macro-blocking" in the rock as a small failure, and the sporadic large artifact blocks in the water was a similar failure. If I was recording an evening rock concert (the other kind of rock), I do not know what would happen. I see reason to be hopeful, but the bitrate does not leave a lot of "headroom". I think I would prefer it if Sony could have squeezed in at least the 60 mbps that is the "low end" of my other Sony cameras.


I wrote before that as far as recording something to post directly to YouTube, the results would generally not embarrass me. I think that assessment was about right. But for including in a "serious project", or to use it in a situation that "pushes beyond normal conditions", I do not think I could trust it. It's pretty good, but. . . .

[2022-0216 23:26 edited]
Is it at least the best phone camera I have for video? In general, yes, but here's the surprise: My Xperia Z3c records UHD at 30 fps and 60 mbps and if you can keep the sun behind you (so the flare is not a problem), I think it might be able to do a better job than the Xperia 10 iii's "Normal" lens. But that is a tightly restricted case, and realistically, I think the zoom on the Xperia 10 iii performs better than the Z3c's camera. So yes, do I think that the Xperia 10 iii is my best current phone camera for video? I'm not sure. I think generally the answer is yes, but with the caveat that I could still prefer the "Z3c" in some cases where I expect that I might want to use the clip later in a larger project.


Other Thoughts:

The lack of thorough analysis by various "phone reviewers" bothers me now. I am really wondering now what other phones are like. Does even the Xperia 1 iii have a more sufficient bitrate? It better! What about the Samsungs and Apples, and the various budget brands? Those reviews are far too impressed with over-saturated colours, and how fast they can make their screens scroll. . . .


Video Stats:

Sony Xperia 10 iii Clips:

"MOV_0004.mp4"
- zoom 1x (normal lens 1x)
- exposure +0
- 2nd composition, rough level (tripod bubble)
- stabilization OFF
- caps from 0:10

Size: 692,930,766 bytes
Created July 28, 2021, 17:00:32
Video
Length 02:11
Frame width 3840
Frame height 2160
Data rate 42,135 kbps
Total bitrate 42,291 kbps
Frame rate 30 frames/second
Audio
Bit rate 156 kbps
Channels 2 (stereo)
Audio sample rate 48 kHz
VLC:
Stream 0:
Type: Video
Codec: H264 - MPEG-4 AVC (part 10) (avc1)
Language: English
Resolution: 3840x2178
Display Resolution: 3840x2160
Frame rate 30.053529
Decoded format: Planar 4:2:0 YUV full scale
Stream 1
Type: Audio
Codec: MPEG AAC Audio (mp4a)
Language: English
Channels: Stereo
Sample rate: 48000 Hz


"MOV_0006.mp4"
- zoom digital 2x, [edited 2022-0216]
- exposure -1 stop
- adjusted level using app
- caps from 0:10

Size: 669,242,394 bytes
Created July 28, 2021, 17:05:00
Video
Length 02:06
Frame width 3840
Frame height 2160
Data rate 41,975 kbps
Total bitrate 42,131 kbps
Frame rate 30 frames/second
Audio
Bit rate 155 kbps
Channels 2 (stereo)
Audio sample rate 48 kHz
VLC:
Stream 0:
Type: Video
Codec: H264 - MPEG-4 AVC (part 10) (avc1)
Language: English
Resolution: 3840x2178
Display Resolution: 3840x2160
Frame rate 30.053093
Decoded format: Planar 4:2:0 YUV full scale
Stream 1
Type: Audio
Codec: MPEG AAC Audio (mp4a)
Language: English
Channels: Stereo
Sample rate: 48000 Hz

 

[VidThreeNorth]

Things I found recently:

The Xperia 10 iii has still and video "screen capture". I have tested video capture while making a Full HD video and that much worked. I have not tested capturing the screen to a video while making a UHD video yet. I plan to make a specific test video during a recording of a UHD video.

[2022-02-17 13:22 Added the following quotes from the manual.]
From page 45 onward in the manual:

Still:

"1. Long press the volume down key (A) and power key (B) at the same time."


"To take a screenshot of a recently used app

1. Tap [] (Overview button in the navigation bar.

2. Swipe left or right to select the app of which you want to take a screenshot, and then tap [Screenshot]"


Video:

"1. Swipe down twice from the top of the screen to open the Quick setting panel.

2. Swipe left or right to find [] (Screen Record icon), and then tap it. . . . .

3. Follow the on-screen instructions to start recording. . . ."


Slow Motion:

Also there is a specific slow motion feature which records 120 fps. I have not tried that yet.

 

[VidThreeNorth]

Sony Xperia 10 iii Video Report 01: UHD 2X Telephoto Camera

Looking at my early sample videos, I have been fairly pleased by the results, but I thought that there was something odd about the truncated highlights. I felt that there was some "adjustment" being made that was stopping clipping from occurring, and as such I thought that it would be difficult to match the appearance from the Xperia 10 iii and a real camera. I would have tested the gamma anyway, but it became more important to me.

I have made a wide variety of test pictures and video clips, but I am only writing up my basic test of the UHD video captured on the telephoto lens (nominally the "2x" video output). This is what I expect that I might use. Not only was I surprised by the results, but I have no clear understanding why Sony has done this.

What Sony Did:

The still photos from this phone are fairly typical. Some people will like them, and others might find them disappointing, but their characteristics are not particularly unusual in any way. On the other hand, Sony did manage to do something odd in this particular video mode. I do not know if this oddity is in the other video modes, but I expect that they are consistent with this test.

The color values for this mode range from 0, to ~251. No colour component (R, G or B) reaches 255. There does not seem to be a curve up to 251 either. It seems like an abrupt stop at 251. The only guess I can make about why this was done (and I emphasize that I am guessing -- I have no "inside" information, and I have never seen this done before in any other camera), is that it might have been to avoid differences in gamma response characteristics of the different sensors. If this is true, then it might have been done by other cellphone makers with similar multiple camera systems. That is something I do not know, and unfortunately, even if you do a lot of research, you might not find out. The community of "cellphone reviewers" does not seem to understand anything like "dynamic range" beyond maybe having heard the term.

Anyway, lets look at the images:


Sony a5000 w/20mm F2.8 lens [Control]
- "Standard" finish, EV = +0.0

You might have noticed that the exposure of the "Control" picture is not compensated, which is rare for my tests. On my a5000, I usually use adapted lenses, often with full manual control (focus and aperture). When working this way on this camera I eventually started using the "Center" metering mode, which is similar to a "Spot metering mode on a film camera. On the a5000, "Center" is a bit larger than "Spot" but cannot be moved -- it is always in the center. The only other mode is "Multi" which is the default.

Since I used the "Center" metering, which is exactly where the test card was located, the test card was perfectly exposed. So no compensation was needed, and the result also caused the background to be under exposed, which was not important to me. In most cases if I use "Center" it is with a "Zebra" set to 100%. I point the camera at the main subject, set my exposure (usually the ISO and shutter speed are already set, so aperture is all I need), and then compose. If I have time, I might re-adjust the exposure a bit further based on the zebras, which is my generally preferred exposure, but if I am working manually, I am more likely busy adjusting focus and composition, and leaving the fine tuning of the exposure to post.

On the other hand, the Xperia 10 iii metered, what appears to be an average or slightly center-weighted pattern which resulted in an un-adjusted exposure about 1 stop high in the first test sample frame. The compensated exposure in the second test sample frame is corrected and the result is a good acceptable exposure.


Control: Sony a5000 w/Sony 20mm F2.8 lens

"01-Sa5k-Control-EVplus0.jpg" [Not Posted]

This is a "Standard" render with no changes to contrast, saturation nor sharpening.

Black Luma 41.0
Black + 1 Luma 86.6
Black + 2 Luma 145.9
White - 2 Luma 193.8
White - 1 Luma 225.5
White Luma 242.1

Contrast: (White - 1) - (Black + 1) = 225.5 - 86.6 = 138.9

This is, or was, Sony's idea of typical contrast. I have not done a statistical survey about it, it does seem to be that recent Sony APS-C and Full Frame cameras default to a bit less contrast. I emphasize that this is just my "impression". The original is not a video frame, but it was a full 20 MP still image.


"01a-Sa5k-Control-EVplus0-Crop1600-C1.jpg"
Crop start: 2100, 1420

Since the point of this post is the cellphone's video capability, I am not bothering with a resized image of this picture -- just this crop.


"02-UHD-0007-EVplus0.png" [Not posted]
Exposure EV = +0.0

Black Luma 55.9
Black + 1 Luma 114.0
Black + 2 Luma 169.7
White - 2 Luma 218.7
White - 1 Luma 251.1 [clipped?]
White Luma 251.4 [clipped?]

Contrast 251.1 [clipped] - 114.0 = 137.1 [clipped]

This is the first image in this set taken with the Xperia 10 iii. Both the "white - 1" and "white" sample patches showed pixels with R, G and B maxed out at "251". I have labeled them "clipped" but with a question mark, though really there is not doubt that these are clipped. The "contrast" calculation therefore cannot be trusted.

"02a-UHD-0007-EVplus0-Rsz1600-C1.jpg"
Resized

"02b-UHD-0007-EVplus0-Crop1600-C1.jpg"
Crop Start: 1220, 800


"03-UHD-0008-EVmns1.png" [Not posted]
Exposure EV = -1.0

Black Luma 28.7
Black + 1 Luma 70.1
Black + 2 Luma 119.7
White - 2 Luma 162.3
White - 1 Luma 201.3
White Luma 235.6

Contrast = 201.3 - 70.1 = 131.2

This contrast calculation can be trusted, and it is similar to the "control" image. The colour representations might be a bit different, but it does look like you could edit the output of the Xperia 10 iii with, maybe a Sony a6300 and get a good match that far. If you look at the detail crop though, it is not really wonderfully sharp, which could be an issue.

"03a-UHD-0008-EVmns1-Rsz1600-C1.jpg"
Resized

"03b-UHD-0008-EVmns1-Crop1600-C1.jpg"
Crop Start: 1220, 800

Not much that I can say beyond what the numbers show, and my remark about the image quality being less than what I could get with a Sony ILC with a good lens. The sharpness is probably good enough if I were editing down to Full HD. I would have to be very careful about exposure if I wanted to accurately match this phone with a better camera. It is not a "sunny day" camera. With it's limited dynamic range, one would want to have cloudy days to record. The phone is waterproofed, so rain is not a problem. Or one could fall back on stills.

I have a number of files with specific options such as Full HD and "slow motion", but from what I have seen, I am not expecting surprises. Which is to say, one can extrapolate the results from what I have posted so far. That being the case, I am not in a hurry to continue posting more findings. I probably will do so eventually, but it is not a priority.

 

[VidThreeNorth]

So, about an hour after I posted the above report I checked my phone annnnnnd, a firmware update was available. I updated the phone and checked the buggy stabilization (UHD @ 10x) and it looks like it has been fixed.

I will not bother analyzing my previous videos that I have not posted yet. Instead, I will start with fresh recordings. Yeah, it will take a while longer. I think I am happy about this. I'll know for sure when I see the new files.

 

[VidThreeNorth]

Stabilization Report:

First, the build number for this update is reported as "62.0.A.3.109 release-keys". As for the stabilization issues I have mentioned, my first tests showed some stabilization issues, but later they seemed to go away. I thought about this and I am wondering if stabilization has an adaptive component in its firmware? If so, it is possible that part of the reason I had problems with the older firmware, and had problems when I first used the new firmware, is because I do not use stabilization much. So it is possible that it had to "learn" how to work, and now that I have used it a bit, it is working better. That might mean that there was nothing "wrong" with it before. It just needed to be used. I don't know. I will have to watch what happens in the future. But I really have little incentive to use stabilization. I prefer to mount the phone on a 'pod when I use it for video work, and I have not been using the phone for stills much.

 

[VidThreeNorth]

The Screen Capture Function:

This report is actually about the "Screen Capture" capability in Sony's implementation of Android 11 on the Xperia 10 iii. I suspect that this is a direct Android feature and not just something that Sony added, but I have not researched that. So I cannot say if you will see it in other Android phones. I have seen a capture in a Sony Xperia 5 iii promo video which implies to me that this feature is, at the least, probably also in that phone and also the Xperia 1 iii.

The "Screen Capture" supports optional stills or video, and the video can capture video and optional audio. The audio takes the microphone input except when the "screen capture" is on top of actual (camera) video recording. In that case the audio recording for the screen capture video is simply blanked until the main video recording is done.

I had no particular reason to record a screen capture, so I randomly decided to record my playing a round of Mobilityware's "Freecell" implementation. I recorded a few games, trying to record a short one. I settled on a recording roughly 10 minutes long. Luckily, I decided to watch the recording before posting it. The recording turned out well enough technically, but it was seriously boring. It is not the first boring thing I have ever recorded, but I decided to spare everyone the experience and decided to post some captures from the video instead.

About The Game Demonstrated:

Mobilityware Freecell has a typical vertical layout or an optional horizontal layout. I recorded both vertical and horizontal versions. The vertical version looks much like the versions I have played on a computer, but the horizontal version is interesting because it moves the 8 spots that are normally at the top of the screen, dividing them and putting the 4 "Free Cell" spots on the left of the screen and the 4 collection spots on the right of the screen.

The file stats show a surprisingly high compression, but I think this is partly due to the subject matter. There is no indication of the compression used. Having seen a few screen capture videos, I am guessing that the compression is probably around 95%. I did not bother to record audio.
[2021-09-23 20:41 The compression spec for my video editor is the opposite from Corel Paintshop Pro JPEG compression. 99 is the least compressed and results in the largest files.]

If I find a better subject to record, I might post a full video, or maybe more captures, but I do not currently consider this a priority.


File Stats:

"02-HorFullWon-095648.mp4"

Win 8.1:
Size 159,420,734 bytes
Created 10:32:28
Video
Length 08:55
Frame width 2520
Frame height 1080
Data rate: 2,377 kbps
Total bitrate: 2,377 kbps
Frame rate 34 frames/second

VLC:
Stream 0
Codec H264 - MPEG-4 AVC (part 10) (avc1)
Language English
Type Video
Video resolution: 2520x1080
Buffer dimensions: 2528x1090
Frame rate 34.155820
Decoded format: Planar 4:2:0 YUV
Orientation: Top left
Color primaries: ITU-R BT.601 (625 lines, 50 Hz)
Color transfer function: ITU-R BT.709
Color space ITU-R BT.601 Range
Chroma location: Left

 

[VidThreeNorth]

Web Browsing:

This is a video issue. I am trying to find a Web Browser that does well with YouTube. This is sort of a semi-random collection of issues I have run into. If anyone has a recommendation I would appreciate it.

I Want ONLY Horizontal Viewing. As a general practice, after years of experience with tablet computers and phones, I have found that I want a true web browser and NOT a special app. YouTube has an App. I have tried it. I don't like it. I want a general purpose standard HTML browser with good security and proper support of general Web capabilities. That means fairly complete support of HTML, JavaScript, and Java. I want to view standard desktop computer pages and not the cut-down "mobile" pages. Since I want the desktop pages, I want to view the using the horizontal layout. This is an issue on this Sony Xperia 10 iii. On the system level it defaults to vertical. The only way to view horizontal is to use the auto-rotate. If it is in horizontal using the auto-rotate and I lock it, it changes back to vertical.

Firefox Browser by Mozilla version 92.1.1 (Build #2015832083:

This browser relies on the auto-rotate to support horizontal. It does not lock in that screen mode. Also, the controls for the video do not show up, and because of the phone's screen layout, the bottom of the videos is cropped leaving me about 2/3 of the vertical content of the screen. The video does not fill the screen vertically. The top is filled with page formatting.

Dolphin Browser version 12.2.9

Horizontal works well. The full YouTube controls work fitting the view to show the full video within the browser frame, or optionally filling the screen top - to - bottom (shrinking horizontally as appropriate) putting icon control indicators around the video. The only thing that goes wrong is that sometimes those controls stay visible during the video. Again, the only problem I found was that it rotates to vertical as usual. It cannot be locked in horizontal.

Maxthon Browser MX6 version 6.0.0.3490 (NOTE: This is NOT the same as the "MX" Player which is a media player).

Basically the same problems as Mozilla. The full video image cannot be seen at once. And again, the screen cannot be locked in horizontal mode.

Samsung Internet version 15.0.2.47

Based on the Firefox browser. Bottom of the video is cut off the same as with FireFox. And again, it auto-rotates to vertical.

Chrome by Google, version 93.0.4577.82

Fills the screen vertical properly. Also auto-rotates.
[2021-09-25 17:57 Correction:
The web pages has the normal size options for viewing the video. The first time I tried it, the icons did not show up, but today they displayed properly and seem to be working.]

Conclusions (for now):

The Dolphin Browser works best for YouTube, and also for general web browsing.

It would be nice if I could lock the screen Horizontal on the system level, but since that is not available, it would be good to be able to over-ride the layout option from within the browser app. Also there is a major security issue which is in some or all of the browsers. The phone's location must be on. When I did this testing, I ran into the problem with one of them, and I left it on for the rest of the tests, so I do not have a record of which programs had the problem. If I can, I will add the info later. On the system level I set it to "[Allow] Only when app is in use". I do not feel this is acceptable. In fact, it will probably not matter since that "auto-rotate" problem makes all the browsers too annoying for me to bother with.

[2021-09-25 18:00 added following, and also attached files.]

I tested the two best browsers with the YouTube App disabled. The app cannot be uninstalled, but in theory, if either browser was "handing off" YouTube streams to the the app, this should have stopped it. That way, I tested the performance of each browser directly addressing system calls or internal included routines.
[2021-12-07 01:42 Section rewritten for clarity.]


"screen-20210925-142538.mp4" [not uploaded]
Nigel Danson, Chrome browser

Size: 173,688,820 bytes
Created September 25, 2021, 14:25:40
Video Length 02:22
Frame width 2520
Frame height 1080
Data rate 9,694 kbps
Total bitrate 9,694 kbps
Frame rate 32 frames/second

VLC:
Stream 0
Codec: H264 - MPEG-4 AVC (part 10) (avc1)
Language: English
Type Video
Video resolution: 2520x1080
Buffer dimensions: 2528x1090
Frame rate: 32.517646
Decoded format: Planar 4:2:0 YUV
Orientation: Top left
Color primaries: ITU-R BT.601 (625 lines, 50 Hz)
Color transfer function: ITU-R BT.709
Color space: ITU-R BT.601 Range
Chroma location: Left


"screen-20210925-144046.mp4" [not uploaded]
Thomas Heaton, Dolphin browser

Size: 113,950,168 bytes
Created September 25, 2021, 14:40:46
Video Length 05:08
Frame width 2520
Frame height 1080
Data rate 2,939 kbps
Total bitrate 2,939 kbps
Frame rate 34 frames/second

VLC:
Stream 0
Codec: H264 - MPEG-4 AVC (part 10) (avc1)
Language: English
Type Video
Video resolution: 2520x1080
Buffer dimensions: 2528x1090
Frame rate: 34.258220
Decoded format: Planar 4:2:0 YUV
Orientation: Top left
Color primaries: ITU-R BT.601 (625 lines, 50 Hz)
Color transfer function: ITU-R BT.709
Color space: ITU-R BT.601 Range
Chroma location: Left


"Chrome-0_06-C1.jpg"
Web page opened, horizontal mode

"Chrome-0_09-C1.jpg"
Web page opened, showing video control icons

"Chrome-0_15-C1.jpg"
Video displayed full screen -- the instruction disappears after a few seconds.

"Chrome-0_50-C1.jpg"
Vertical mode of web page, screen capture is also auto-rotated, shrinking the video and text.

"Chrome-1_51-C1.jpg"
Horizontal full screen display of video with clean display.

"Dolphin-0_40-C1.jpg"
Web page opened in horizontal mode.

"Dolphin-0_45-C1.jpg"
Web page opened in horizontal mode, scrolled down to see the title.

"Dolphin-1_20-C1.jpg"
Video displayed horizontal full screen with control icons.

"Dolphin-3_00-C1.jpg"
Video displayed full screen vertical

"Dolphin-4_07-C1.jpg"
Horizontal full screen display of video with clean display

 

[VidThreeNorth]

I received the latest firmware update 2022-02-12 17:21 which is Build number "62.0.A.3.131 release-keys". There are some new test modules to confirm the phone is operating properly, but nothing refers to the camera system. I am not going to start my tests over again. I have done some new tests, and I am still evaluating the results. Nothing looks "earth shattering" though, so I am not in a hurry. I tried a bunch of pictures of trees in snow (extremely high contrast) using the HDR function and a quick look at the results was disappointing. I think they all failed, mainly due to motion blurring, despite daylight conditions (roughly mid-day). Maybe the extra cold of the day made me too shakey. I do not really know. . . .

 

[VidThreeNorth]

This is a short note on the current update state of my Xperia 10 iii. On Mar 23, 2022 I installed a firmware update resulting in "62.0.A.9.11 release-keys".

The firmware immediately prior to that had created a bug in the charging system. That previous firmware stopped charging from occurring while the phone was OFF. I think that most people probably would not have noticed this because I think that most people probably leave their phones on during the night while charging. Since I am not currently using my Xperia 10 iii as a phone, I prefer to charge it while it is off, which should be the quickest way to charge it. If I attempted this, the charging graphic would appear, but the animation did not start, and, as far as I could tell, charging was not occurring. The solution was simply to turn on the phone and then charge it until 100%, and then unplug it and turn it off.

This above noted firmware version fixed the problem and the phone charges while off, subject to what I have noted before (it depends on which way the type C plug is inserted).

There was also a recent announcement that Android 12 is becoming available. The Xperia 10 iii is listed as a candidate for this upgrade. I do not know if there will be significant changes to the camera in the Xperia 10 iii. In the higher models there do seem to be new features for video.

 

[VidThreeNorth]

The Sony Xperia 10 iii with Android 12 as a Photographic Device:

General stats:

The zoom range, in theory ranges from "12mm F4.0" equivalent ultra-wide to 54mm F4.0 (telephoto) which is digitally zoom-able out to about 4x for Full HD video, (or about 2x for stills). Video is limited to 30 fps for both Full HD and UHD. There is a 120 fps slow motion available (I have only used it for a short sample file so far, giving me the following data: (conformed to 30 fps, actually about 30.07 fps according to VLC) 1280 x 720p, Video data rate 12,420 kbps, Total Bitrate 12,575 kbps (sound is recorded). This implies a record rate of about 50 mbps, which is about what I expect from Sony.
[2022-04-24 12:54 I left out Full HD at 60 fps, which is also usable for "slow motion". Some people seem to use 60 fps as a regular playback speed.]


Summary:

Knowing that the Android 12 update was coming for my Xperia 10 iii brought me to a decision point for my future projects for this phone. I have not intended to write much about the phone beyond photographically related issues, and the Xperia 10 iii, being the bottom of the product line meant that Sony did not give it the capabilities that it did for the other higher models. Having put effort into using it specifically to see what it could do, I found it was pretty much what I expected. It is a good, slightly better than minimal "stills camera". It has reasonably versatile Full HD video, but only what I would call borderline minimal UHD video.


Pricing (in perspective):

My Xperia 10 iii XQ-BT52 5G, Dual 128GB Flash, 6GB RAM
Costs about $440 US, proportionally, the Xperia 5 iii (XQ-BQ72) costs about 1.5x, The Xperia 1 iii (XQ-BC72) costs about 2.0x, and the Xperia Pro-I (XQ-BE72) costs about 3.0x

If you could have found a used Xperia 5 ii (which would have been "last year's model" as of when I bought my 10 iii), it should have costted about the same as my 10 iii "new - in box".

Alternatives:

My Xperia 10 iii competed mainly against the Apple iPhone 12 mini, in price and screen resolution, though the screen was larger. The full size iPhone 12 is priced more against the Sony Xperia 5 iii. Photographically, for stills, the Xperia 10 iii and the iPhone 12 mini are roughly equal, but different. The apple product has only two main cameras, but they are 12MP each whereas my Sony has three main cameras, but they are 12MP for the main "Wide" camera, but only 8MP each for the "Ultra wide" and Telephoto cameras. For me, the big gain is the "telephoto" camera, which suits my use case for "stills" in particular because I find the longer focal length more commonly suits my compositions. But the Xperia 10 iii's UHD video can only use the main "Wide" camera, and even then the quality of the output file is limited to barely useful dynamic range and a maximum 30 fps.

Since I have no direct experience with the Xperia 5 ii or iii, I can only guess whether I would really find it that good or not. I do feel that I have enough "trusted reports" to say that I believe that I would find it useful, but yes, I could be wrong.

My most important summary would be that if have about $2,000 US right now, and you are going into video, but you know that you should be buying a 5G phone "right about now", then I would suggest that you should look to buying something like an A6400, Canon RP (or even an R might be possible), Nikon Z50, Panasonic G9, or OM Systems EM5iii, and the next model up Sony phone, the "Xperia 5 iii" instead. The main difference is that the Xperia 5 iii should be a reasonably useful UHD video camera adding a lot of versatility to your system. It is also a "superior" stills camera.

Sample Images:

My first video file was a Full HD test with stabilization on an overcast day.

"MOV_0047.mp4" (not posted)
Windows:
Size 261,303,407 bytes
Created Apr 21, 2022, 12:12:43
Video
Length 02:08
Frame width 1920
Frame height 1080
Data rate 16,150 kbps
Total bitrate 16,306 kbps
Frame raate 29 frames/second [sic]
Audio
Bit rate 156 kbps
Channels 2 (stereo)
Audio sample rate 48 kHz

VLC:
Codec: H264 - MPEG-4 AVC (part 10) (avc1)
Language: English
Type Video
Video resolution 1920x1080
Buffer dimensions 1920x1088
Frame rate 29.715032
Decoded format
Orientation Top left
Color primaries: ITU-R TB.601 (625 lines, 50 Hz)
Color transfer function ITU-R BT.709
Color space ITU-R BT.601 Range
Chroma location: Left
Stream 1
Codec MPEG AAC Audio (mp4a)
Language: English
Channels: Stereo
Sample rate: 48000 Hz
Bits per sample 32

Sample frame:

"01a-15h58m44s477-rsz1600-C1.jpg"
Adjusted and Resized:
Corel Smart Photo Fix recommended
Brightness:
Overall 28
Highlights 10
Sat 8
Focus 10
Resized to 1600

"01b-15h58m44s477-crop1600-C1.jpg"
Detail Crop:
[1600 x 900 Crop from 1900,1070]
- no adjustments

 

[VidThreeNorth]

More Perspective about the Xperia 10 iii with Android 12

In order to put my feelings about the Xperia 10 iii with Android 12, I am posting a couple of videos by other people which relate.

First, I have noted that the Xperia 10 iii is essentially a "Full HD" video camera. In this regard, I have other "Full HD" camcorders, including a Sony CX-405, which has about 30x optical zoom and three Canon camcorders, and I do still use Full HD "sometimes". The reason I am mainly recording in UHD is because I already have three superior cameras/camcorders, and I do not find the drawbacks of using UHD generally significant. But Full HD is actually still "good enough". Here is a piece by Tony Northrup from January 2020 where-in he says why he went back to Full HD despite having used UHD for a while. I do not know if he is back in "the UHD camp" at this point, but as he states, most people who view YouTube do so on mobile phones and "iPads" and they cannot see the difference between Full HD and UHD. If you check around, I believe you will find that Full HD is still used by many YouTubers even today.

"4K Failed: the TRUTH about HD vs 4K",
Posted Jan 5, 2020 by "Tony & Chelsea Northrup", [Length 9:04]
""

This next link is to a review of the next model up Sony Xperia 5 iii. It is the only review I have found of the Xperia 5 iii using Android 12. The report is based on a "loaner" unit provided by Sony, and the Android 12 installation seems to be a "pre-release" version. Also, I have not seen other posting by "TK Bay" before, and do not know his reputation, but this report seems about what I expect when I compare it to my experiences with my Xperia 10 iii:

"Sony Xperia 5 III 5G Review, Android 12 & Top Features (Cameras, 120fps Gaming, With Chapters)",
posted Jan 11, 2022 by "TK Bay", [length 16:08]
""

As I have written above, considering the price of the Xperia 5 iii is about 1.5x the cost of an Xperia 10 iii, I think if you really want to use the phone's cameras, it is likely you will find the Xperia 5 iii a better buy.

 

[VidThreeNorth]

While 3D rendering is generally thought of as a "gaming feature", that was not where the technology started, and there are other possible uses. Personally, I thought that it would be a good way of creating backgrounds or environments for photography. I never got around to using it that way, but then again, there are a lot of things I have not gotten around to doing yet. Anyway, I recently ran the "GFX Benchmark to test its performance. Gamers would find it weak compared to today's "flagships". What I found was that first, it completed the benchmark. This was a first for me. I have never had a phone or computer that could actually complete this benchmark before. As for the "power", well, as expected, this being a low cost device, it was far below the top performers. But still, if I pre-render a video to use as a background for a production, it appears that I would be able to use it for that purpose.

Here are the "top line" results -- the most generally useful number. I will add to this later someday, but it is time consuming. The benchmark displays the results in good detail, but not as a text file. I have to re-type all the output data by hand.

[tests ran on Apr 30, 2022]
[Report Last Edited 2022-09-17
- Added comparison data. Also re-ran 1 test.]

Sony XQ-BT52
Xperia 10 iii, Android 12

GFX Benchmark results

3D API-VULKAN
3D API-GL OpenGL ES 3.2
Memory 5.4 GB

(2022-0908 21:08 UPDATE)
Version 5.0.5
(previous test using Version 5.0.4)


High-Level Tests


Format:
Test Name
Device
Frames
(Frames - previous run)
- Cmp1 [Best device score reported]
- Cmp2 [Closest device score]


Aztek Ruins OpenGL (High Tier)
ES 3.1 | Adreno (TM) 619 | 2394 x 1080
514.8 Frames (8.0 Fps)
(513.9* Frames (8.0 Fps))
- NVIDIA Shield tablet, NVIDIA OpenGL 644.1 Frames
- Apple iPad Pro, Apple | Metal 512.9 Frames

4K Aztek Ruins OpenGL (High Tier) Offscreen [First Run]
ES 3.1 | Adreno (TM) 619 | 3840 x 2160
158.7* Frames (2.5 Fps)
- (NEW TEST, no comparison results available yet)

1440p Aztek Ruins OpenGL (High Tier) Offscreen
ES 3.1 | Adreno (TM) 619 | 2560 x 1440
370.3* Frames (5.8 Fps)
(364.8* Frames (5.7 Fps))
- Samsung Galaxy S8, Adreno 540, SM-G950x, OpenGL 463.0 Frames
- Nvidia Shield Tablet, Vulkan 360.1 Frames

Aztek Ruins OpenGL (Normal Tier)
ES 3.1 | Adreno (TM) 619 | 2394 x 1080
807.5* Frames (13 Fps)
(807.5* Frames (13 Fps)) - no change
- NVIDIA Shield tablet, NVIDIA OpenGL 1028 Frames
- LG G Flex 2, F510, LG OpenGL 818.4 Frames

1080p Aztek Ruins OpenGL (Normal Tier) Offscreen
ES 3.1 | Adreno (TM) 619 | 1920 x 1080
983.8* Frames (15 Fps)
(979.9* Frames (15 Fps))
- Samsung Galaxy S8, Adreno 540, SM-G950x, Vulkan 1236 Frames
- Samsung Galaxy S8, Adreno 540, SM-G950x, OpenGL 1029 Frames

Aztek Ruins Vulkan (High Tier)
VULKAN 1.0 | 0x6010900 - Adreno (TM) 619 | 2394 x 1080
525.2* Frames (8.2 Fps)
(525.2* Frames (8.2 Fps)) - no change
[Note: Re-ran the test 2022-05-18. My best score so far]
- NVIDIA Shield tablet, NVIDIA OpenGL 644.1 Frames
- Apple iPad Pro, Apple | Metal 512.9 Frames

1440p Aztek Ruins Vulkan (High Tier) Offscreen
VULKAN 1.0 | 0x6010900 - Adreno (TM) 619 | 2560 x 1440
380.1* Frames (5.9 Fps)
(380.0* Frames (5.9 Fps))
- Samsung Galaxy S8, Adreno 540, SM-G950x, OpenGL 463.0 Frames
- NVIDIA Shield tablet, NVIDIA OpenGL 391.4 Frames

4K Aztek Ruins Vulkan (High Tier) Offscreen
VULKAN 1.0 | 0x6010900 - Adreno (TM) 619 | 3840 x 2160
166.8* Frames (2.6 Fps)
- (NEW TEST, no comparison results available yet)

1440 Aztek Ruins Vulkan (High Tier) Offscreen
VULKAN 1.0 | 0x6010900 - Adreno (TM) 619 | 2560 x 1440
380.1* Frames (5.9 Fps)
- Samsung Galaxy S8, Adreno 540, SM-G950x, OpenGL 463.0 Frames
- NVIDIA Shield tablet, NVIDIA OpenGL 391.4 Frames

Aztek Ruins Vulkan (Normal Tier)
VULKAN 1.0 | 0x6010900 - Adreno (TM) 619 | 2394 x 1080
830.7* Frames (13 Fps)
(830.7* Frames (13 Fps)) - no change
- NVIDIA Shield tablet, NVIDIA OpenGL 1028 Frames
- Nvidia Shield Tablet, Vulkan 932.5 Frames

1080p Aztek Ruins Vulkan (Normal Tier) Offscreen
VULKAN 1.0 | 0x6010900 - Adreno (TM) 619 | 1920 x 1080
1,045* Frames (16 Fps)
(1,043* Frames (16 Fps))
- Samsung Galaxy S8, Adreno 540, SM-G950x, Vulkan 1236 Frames
- Nvidia Shield Tablet, Vulkan 1046 Frames

Car Chase
ES 3.1 | Adreno (TM) 619 | 2394 x 1080
687.8 Frames (12 Fps)
(687.8 Frames (12 Fps)) - no change
Sony Xperia XZ1 Compact (G8441) OpenGL 2983 Frames
Zebra Technologies TC56 OpenGL 686.4 Frames

1080p Car Chase Offscreen
ES 3.1 | Adreno (TM) 619 | 1920 x 1080
855.6 Frames (14 Fps)
(855.6 Frames (14 Fps)) - no change
- Qualcomm Adreno 540 (dev bd) qcom OpenGL 1,449 Frames
- Xiaomi Mi Mix, Xiaomi OpenGL 853.3 Frames

1440p Manhattan 3.1.1 Offscreen
ES 3.1 | Adreno (TM) 619 | 2560 x 1440
904.8 Frames (15 Fps)
(902.4 Frames (15 Fps))
- Apple iPad Pro (12.9", 2nd Gen), Apple Metal 2,721 Frames
- Samsung Galaxy Note FE (SM-N935), Sam OpenGL 905.7 Frames

Manhattan 3.1
ES 3.1 | Adreno (TM) 619 | 2394 x 1080
1,316 Frames (21 Fps)
(1,316 Frames (21 Fps)) - no change
- Sony Xperia XZ1 Compact (G8441), Sony OpenGL 3,726 Frames
- Razer Phone, Razer OpenGL 1,325 Frames

1080p Manhattan 3.1 Offscreen
ES 3.1 | Adreno (TM) 619 | 1920 x 1080
1,610 Frames (26 Fps)
(1,606 Frames (26 Fps))
- Apple iPad Pro (12.9" 2nd Gen) A. Metal 4,738 Frames
- Samsung Glxy Note FE (SM-N935), Sam OpenGL 1,609 Frames

Manhattan
ES 3.0 | Adreno (TM) 619 | 2394 x 1080
1,893 Frames (31 Fps)
(1,886 Frames (30 Fps))
- Sony Xperia XZ1 Compact (G8441), Sony OpenGL 3,724 Frames
- Alcatel 6076S, Alcatel | OpenGL 1886 Frames

1080p Manhattan Offscreen
ES 3.0 | Adreno (TM) 619 | 1920 x 1080
2,264 Frames (37 Fps)
(2,251 Frames (36 Fps))
- Apple iPad Pro (12.9" 2nd Gen) A. OpenGL 6,761 Frames
- Apple iPhone SE, Apple Metal 2,252 Frames

T-Rex
ES 2.0 | Adreno (TM) 619 | 2394 x 1080
3,042* Frames (54 Fps)
(3,041* Frames (54 Fps))
- Razer Phone OpenGL 4163 Frames
- Xiaomi Mi Mix, OpenGL 3040 Frames

1080p T-Rex Offscreen
ES 2.0 | Adreno (TM) 619 | 1920 x 1080
4,068 Frames (73 Fps)
(4,043 Frames (72 Fps))
- Apple iPad Pro (12.9" 2nd Gen) A. OpenGL 12,509 Frames
- Xiaomi Mi Note 2, OpenGL 4064 Frames

 

[VidThreeNorth]

GFX Benchmark Part 2

This is the rest of my results of my "GFX Benchmark" run:
[Updated 2022-08-12 - This test was run on Android 12 "build number 62.1.A.0.617"]

Low-Level Tests


Tessellation (* NOTE: new report format)
ES 3.1 | Adreno (TM) 619 | 2394 x 1080
674.0 Frames (22 Fps) - (updated results)
(674.1 Frames (22 Fps)) - (originally posted numbers)
- ZTE Axon M (Z999) 1820 Frames (best)
- Xiaomi Mi Pad 2 OpenGL 675.2 Frames (nearest)


1080p Tessellation Offscreen
ES 3.1 | Adreno (TM) 619 | 1920 x 1080
1,508 Frames (25 Fps)
(1,508 Frames (25 Fps))
- NVIDIA Shield Android TV OpenGL 8,299 Frames
- Teclast X98 Plus II OpenGL 1,493 Frames


ALU 2
ES 3.0 | Adreno (TM) 619 | 2394 x 1080
889.5 Frames (30 Fps)
(889.1 Frames (30 Fps))
- Irbis TZ11 OpenGL 1,879 Frames
- Nextbook NX16A11264 Ares 11 OpenGL 892.6 Frames


1080p ALU 2 Offscreen
ES 3.0 | Adreno (TM) 619 | 1920 x 1080
2,239 Frames (37 Fps)
(2,238 Frames (37 Fps))
- Apple iPad Pro (10.5 inch) Metal 10,659 Frames
- Motorola Moto X Pro OpenGL 2,238 Frames


Driver Overhead 2
ES 3.0 | Adreno (TM) 619 | 2394 x 1080
1,780* Frames (59 Fps)
(1,784* Frames (59 Fps))
- Apple iPad Pro (12.9-inch, 2nd Gen) Metal 1,799 Frames
- Apple iPad Pro 9.7 Metal 1,784 Frames


1080p Driver Overhead 2 Offscreen
ES 3.0 | Adreno (TM) 619 | 1920 x 1080
5,961 Frames (99 Fps)
(5,971 Frames (100 Fps))
- Apple iPhone 7 Metal 11,097 Frames
- Apple iPad 9.7 (5th Gen) Metal 5,918 Frames


Texturing
ES 3.0 | Adreno (TM) 619 | 2394 x 1080
6,086 MTexel/s
(6,028 MTexel/s)
- Apple iPad Pro (12.9 inch, 2nd Gen) Metal 22,387 MTexel/s
- Apple MacBook (Retina 12-inch 2016) OpenGL 6,016 MTexel/s


1080p Texturing Offscreen
ES 3.0 | Adreno (TM) 619 | 1920 x 1080
6,330 MTexel/s
(6,285 MTexel/s)
- Apple iPad Pro (12.9 inch, 2nd Gen) Metal 20,557 MTexel/s
- Motorola XT1650 (1.8 GHz) OpenGL 6,329 MTexel/s


Special Tests

Render Quality
ES 2.0 | Adreno (TM) 619 | 1920 x 1080
2,510 mB PSNR
(2,510 mB PSNR)
- Lenovo K1 HD (2014) OpenGL 4,479 mB PSNR
- Samsung Galaxy S9+ (Adreno 630, SM-G965) OpenGL 2,510 mB PSNR


Render Quality (high precision)
ES 2.0 | Adreno (TM) 619 | 1920 x 1080
3,632 mB PSNR
(3,632 mB PSNR)
- Lenovo K1 HD (2014) OpenGL 4,479 mB PSNR
- Samsung Galaxy S9+ (Adreno 630, SM-G965) OpenGL 3,632 mB PSNR


Comments:

It has been about a year since Sony announced the Xperia 1 iii, 5 iii and 10 iii, and, as mentioned, I have just recently upgraded my Xperia 10 iii to Android 12. Sony has just started to post about its coming announcement of its Xperia 1 iv. But so far, that is all. There has been no mention of a "5 iv" or a "10 iv". It seems likely to me that the lower line products will not be made this year -- similar to halting of the lower camera lines. Samsung has been introducing its 2022 - 23 lines as well, and Apple's schedule seems "unfazed" in terms of roll-outs, though there production numbers might be affected. I have started to post my thoughts about the Xperia 10 iii, but in a lot of ways, my feelings are not what I would call "conclusive". So I am posting "raw data" and "testing in progress" as much as possible, instead.
[2022-06-15 updated: Sony announced the Xperia 10iv at about the time they announced the Xperia 1iv, but no Xperia 5iv has been announced, so now -- June 15, I think it is safe to say there will not be such a phone.]

If you are looking for a new phone this year, what I post might be helpful. That is about all I can promise. It will be as accurate as I can be, but all I have to work with is this one phone. Aside from GSM Arena, I cannot even say that I have found any other good resources out there. You will have to take the information I post and find your own way to make use of it.

Good luck with that. . . .

I will try to post more, but I do not have much planned.