Sony a5000 Video

[VidThreeNorth]

I am doing some general analysis of the Sony a5000 video capability.

Full HD (1920 x 1080) Video Crop

There is a slight crop when using Full HD video. I have known that for some time, but I am just getting around to testing it. For this test I wanted to make sure there was no adjustments made by the camera's processing, so I used a Pentax 50mm F2.0 lens, manually focused. I set up a test using a tape measure, and I recorded a video at Full HD, and then I took a still picture at full resolution. I captured a frame from the video, and these are the results.

[2020-06-28 15:40 I recorded the video and the still picture both at F5.6]

"00005.MTS"
Created May 17, 2020, 10:05:51 - 10:06:58
Size 180,879,360 bytes
Video
Length 1:05
Frame width 1920
Frame height 1080
Data rate 21,826 kbps
Total bitrate 22082 kbps
Frame rate 23 frames/second [sic]
Audio
Bit rate 256 kbps
Channels 2 (stereo)
Audio sample rate 48 kHz

DSC01293.JPG
2020-0517 10:08
Width 5456
Height 3632
Resolution unit 2
Compressed bits/pixel 2
Exposure 1/1000 sec.
ISO speed ISO-100
Meter mode: Center Weighted Average
Brightness 4.74765625
Sharpness Soft
White balance Auto

Looking at the X axis alone:
Missing ~15 pixels on left
Missing ~25 pixels on right
Total missing ~40 pixels
Picture Width 5456 pixels
so approximate crop relative to sensor = (5456 - 40) / 5456 * 100
= 5416 / 5456 * 100 ~ 0.993%
The "Crop Factor" relative to the sensor size is ~ 1.007
Compared to Full Frame = (1.5 * 1.007)
~1.51

As I have written before, this is generally ignorable, though occasionally I have had to adjust my zoom or camera position slightly.

NOTE: You might notice that the tape measure is not exactly midway through the frame. I felt it was close enough. . . .
[2020-06-28 15:41 Because I used the Pentax lens, there should not be any distortion correction. Distortion correction could change the relationship between the location of a sensor and the pixel where it would normally be represented. If this had not been done, then locating the tape through the middle of the picture would have simplified calculating a reverse compensation. So really, the location of the tape actually should make no difference. But if I had detected some "correction", then being closer to the middle line is easier to work with, so would have been preferable.]

I have uploaded the Full HD captured frame at highest detail ("a5k-1080p-16h03m57s255-C1.jpg"), a resized copy of the still picture ("DSC01293a-Rsz1803-C1.JPG") and crops of the far left end of the still picture ("DSC01293b-Left-C3.JPG" slightly reduced detail) and the far right end of the still picture ("DSC01293c-Right-C4.JPG" slightly reduced detail). If you feel like counting the pixels to see the difference, have fun. . . .

 

[VidThreeNorth]

I have been meaning to test my Sony a5000 in order to find out the best video available on it. At this point, it is very late for me to get around to this, but I do not expect to get rid of this body in the near future, so there is still good reason to learn how to get the most out of it. Over the Winter, I did run some tests looking for a good sunset, but that effort completely skipped over the more normal recording situations. So for this set of tests, I am starting with the Standard profile, and then later I will be expanding to others.

Contrast Calculations:
(This is a copy from a post regarding testing of my G85.)

This is not a standardized method of calculating contrast. It is just a method that I worked out based on the equipment that I have -- specifically my "datacolor Spyder Checkr 25" colour reference chart. It is, so far, only relevant to my own collection of equipment.

Procedure:

Across the "top" of the colour reference chart, from left to right, there are six greyscale sample squares, ranging from darker to lighter. I am calling these reference colours "Black", "Black + 1", "Black + 2", "White - 2", "White - 1" and "White". I am calcuating an estimated "Luma" value for each sample as follows:

First, I am using Corel PaintShop Pro X9, using the "eyedropper" to read the colour of five semi-random pixels. I am testing these pixels in the following order: center, upper left, upper right, lower right, lower left. I test pixels in a pattern roughly like a die. Specifically, I am avoiding edges of the colour samples and yet, spreading them out. The Red, Green and Blue values for each pixel are read as 8 bit values (0 - 255) and recorded as such.

The values are cumulated for each component colour and the averages are kept. Then the three averages are further averaged into an 8-bit "luma" value (representing the brightness of the sample colour).

Lastly the contrast is calculated by subtracting "White - 1" - "Black + 1".

If one or more component colours of any pixel of a sample colour = 0 or 255, then the Luma is marked as "Clipped".

If the calculated Luma value of a sample is 20 or below, then the value is indicated as "Below 20" as a warning that it is below "legal black".


"00001.MTS"
Pentax 50mm F2 @ F5.6

Size 187,662,336 bytes
Created May 17, 2020, 09:38:50
Video:
Length 00:01:07
Frame width 1920
Frame height 1080
Data rate 21983 kbps
Total bitrate 22239 kbps
Frame rate 23 frames/second [sic -- actually 23.976024 per VLC]
Audio bitrate 256 kbps
Channels 2 (stereo)
Audio Sample rate 48 kHz

Standard profile, "baseline" (EV = +0.0, Contrast = +0,
Saturation = +0, Sharpening = +0)

This file was recorded at the same time as the file referred to in the June 6 2020 post (comment #11) for the G85 test at:
"G85 w/12-60mm F3.5-5.6 lens"

"Sa5k-00001-Standard-EV0-Cntst0-22h20m08s773-C1.jpg"

"Black" Luma: 54.3
"Black + 1" Luma: 95.1
"Black + 2" Luma: 148.9
"White - 2" Luma: 200.5
"White - 1" Luma: 233.3
"White" Luma: 254.7 (Clipped)

Contrast: 233.3 - 95.1 = 138.2


"00002.MTS"
Pentax 50mm F2 @ F5.6

Size 194,543,616 bytes
Created May 17, 2020, 09:53:47
Video:
Length 00:01:10
Frame width 1920
Frame height 1080
Data rate 21969 kbps
Total bitrate 22225 kbps
Frame rate 23 frames/second [sic -- actually 23.976024 per VLC]
Audio bitrate 256 kbps
Channels 2 (stereo)
Audio Sample rate 48 kHz

Standard profile, "EV = -1" (EV = -1.0, Contrast = +0,
Saturation = +0, Sharpening = +0)


"Sa5k-00002-Standard-EVminus1-10h52m32s964-C1.jpg"

"Black" Luma: 37.1
"Black + 1" Luma: 59.7
"Black + 2" Luma: 93.5
"White - 2" Luma: 136.1
"White - 1" Luma: 176.1
"White" Luma: 209.2

Contrast: 176.1 - 59.7 = 116.4


"00003.MTS"
- NOTE: Exposure probably -1 as well as contrast reduction.
Pentax 50mm F2 @ F5.6

Size 180,289,536 bytes
Created May 17, 2020, 09:57:50
Video:
Length 00:01:05
Frame width 1920
Frame height 1080
Data rate 21907 kbps
Total bitrate 22163 kbps
Frame rate 23 frames/second [sic -- actually 23.976024 per VLC]
Audio bitrate 256 kbps
Channels 2 (stereo)
Audio Sample rate 48 kHz

Standard profile, "Preferred" (EV = -1?, Contrast = -3,
Saturation = +0, Sharpening = -2)


"Sa5k-00003-Stnd-CntMns3-ShrpMns2-17h27m13s320-C1.jpg"

"Black" Luma: 50.1
"Black + 1" Luma: 68.4
"Black + 2" Luma: 96.9
"White - 2" Luma: 134.1
"White - 1" Luma: 164.3
"White" Luma: 194.5

Contrast: 164.3 - 68.4 = 95.9

Results:

First, even in in video "00003.MTS" the sharpening is still a bit high. I think if I use the "Standard" profile, I will probably reduce sharpening all the way down to -3.

I have not changed the saturation. I often see people reducing saturation and bringing it back up in post. This seems like it might be a misunderstanding on how saturation works. I think I might reduce saturation a bit, but not much -- not unless I am really trying to go monochrome. In general for both contrast and saturation, working in 8-bit source with 8-bit target output, it is actually better to try to record as close as possible to what you want in your final result. Making big, unnecessary changes is where "banding" is coming from. For contrast, yes, I would prefer to work with slightly lower than target contrast because that will help retain detail at the extremes. And saturation also, it is good to have slightly less than your target output -- again to avoid "max-ing" out a colour to an inaccurate hue. But I have rarely encountered situations where recorded colours could not bear some correction, or a bit more saturation.

Having said that, I think a case can be made to use colour correction filters in extreme cases.

Also, the contrast in the baseline clip was surprisingly high, especially when compared to the EV = -1 clip. I wondered if I might have had "Dynamic Range Optimizing" on by mistake. I don't think so though. For the near future, I think I want to try to test some other profiles before looking closer at the "Standard" profile. I think I will probably end up preferring another profile anyway, so I might as well take a look at some others first.

 

[VidThreeNorth]

I am currently backlogged reviewing files for 2021, so I might still have something to post in this topic, but I will not be taking anything into 2022. I sold my a5000 "recently". I guess, I might have gotten fairly good money for it, if I had been in the mood for "hard negotiating", but in fact, I did not. The new owner is likely a beginner, and I did not want to start out a future potential hobbyist (or maybe even professional) with a bad taste in the mouth.