Yi Technology -- Yi-M1

[VidThreeNorth]

I have been using a Yi-M1 since around the second week of August 2017. I received it with the "1.0.20 International" firmware. Originally I intended to use it with that firmware first and then upgrade to the April version "2.0 International" firmware, but I saw that one of the upgrades was for SD-Card compatibility and I felt that I wanted the safest firmware available for storage since a trashed card could mean losing a day's efforts. Actually, since I had not heard of a problem with storage up till now, I think testing "1.0.20" would probably have been safe enough, but I went ahead with the upgrade.

I will not be attempting anything like a "full review" of this camera. I will only report specific issues that I am interested in or specific problems I run into.


Full Reviews are at:

"New kid on the block: YI M1 review"
Published Nov 29, 2016 | Chris M Williams
"New kid on the block: YI M1 review"
My Notes:
According to the publishing date this review was written before version "1.0.20 International" which was the first reasonably stable firmware. There is no indication of an update, so it is best to read this as "historical".
[Added 2016-09-10]

"YI M1 Review", "Camera Reviews / YI Cameras"
"YI M1 Review: Now Shooting!"
Last update "07/31/17: Field Test Part II"
by "Mike Tomkins"

My Notes:
While Tomkins does not state outright which the last firmware he used was, it was clearly the version "2.0 International" (he mentions "mid-April"). Apparently his camera is having "locking up" problems. I have not experienced any yet. I also noticed that his camera was made in China. Mine was made in Indonesia.

"PCMag.com"
"Reviews/Consumer Electronics/Digital Cameras/YI M1"
"By Jim Fisher, December 2, 2016 12:20PM EST"
"YI M1"

My Notes:
The firmware is not identified, but I suspect that it was at best 1.0.20 International, and possibly earlier. This review has the most useful lens analysis Apparently the lenses are both quite good optically. It would be nice if they re-tested it with newer firmware, in particular the auto-focus speed and image quality might have improved.


Firmware 2.0 Changes:

The 2.0 firmware has a new icon on the right side of the screen which opens a menu for selecting the function of the far right "wheel" control. It can redefine the wheel to adjust "Focus Mode", "ISO", "White Balance", "Metering Mode", "File Format" (RAW or JPEG) or "Drive Mode". This adds a second level of quick adjustments to the three dedicated screen buttons on the left which force control of "F-Stop", "Shutter Speed" or "Exposure Compensation" and the "Quick" hardware button on the bottom right outside the screen and the rotary Mode control on the top deck.
[2017-08-26 replaced "while" with "wheel"]

If I had tested the 1.0.20 firmware I could say for certain how much difference having this new icon with is new set of controls changes the usability of the camera, but it seems likely to me that it is a substantial improvement for users.

Another change that I know has occurred is that zooming during 4K video does not seem to cause exposure errors as it did in the past. Apparently this was partly the fault of the metering forcing the aperture fully open. I have no current easy way to test whether it still forces the aperture fully open, but I do not think so.

When the first reports came out about the exposure "flashing" when the zoom was used, I thought that it could be a problem of the processor being over-used, and that it might be best to reduce the frame rate from 30 fps down to 24 fps. But not only has this problem been addressed without reducing the framerate, but I did not notice the camera body becoming unusually hot. So while the processor is probably being heavily utilized, it does not seem to be pushed as hard as some other early or low end 4K video cameras (such as the Panasonic GX850 and the Sony a6300).

"Full HD" video now supports stabilization. I have not tried it yet.


About the 4K video Sensor Crop

The Yi-M1 is not the only camera cropping the sensor during 4K video recording, but the amount of cropping is more than the other Micro 4:3 cameras ("budget" Panasonics). The Panasonics start with a 16M sensor and crop that down to the 3840 x 2160 "4K" video format. In the X dimension this is a crop from about 4592 sensors (the DC-GX850 16 GB largest still picture format is 4592 x 3448) down to 3840 pixels for 4K video for a lens equivalence ratio of about 1.2X (compared to the full Micro 4:3 sensor usage). The Yi-M1 based on a 20MP sensor with X dimension of about 5184 sensors (based on the 5184 x 3888 20MP image format) and crops it to 3840 for 4K for a lens equivalence ratio of about 1.4X which is close to the same reduction from a full size APS sensor down to APS-C (which is roughly 1.5X). All the issues for that degree of sensor cropping apply equally in this situation as the do for using APS-C cropping modes on full frame cameras such as the various Canons, Nikons and Sonys. The importance to me is that the reduction in effective resolution is a significant issue for both the lower end Panasonics, but even worse for the Yi-M1. Personally, I think the reduction in effective sharpness is generally ignorable in the Panasonics, but for the Yi-M1, I will be paying attention to lens selection. In general, I will probably want to stay more with my sharper lenses on the Yi-M1.

Calculating the sensor crop factor compared to a APS Full Frame, the Micro 4:3 sensor is 17.30 mm wide.

Start by calculating the effective width of the Yi-M1 sensor while recording 4K video. [added 2017-09-10]
17.30 mm * (3840 / 5184)
= 17.30 mm * 0.74074
= 12.8 mm

So, for the 4K video mode, the actually used sensor area has a 35mm camera equivalence ratio of 36mm / 12.8mm ~ 2.8, which is close to a 1" (CX) sensor. A 50mm lens will give the view of a 140mm lens on a 35mm camera.


Confirming my Sensor Usage Estimate (above):

Back to the brick walls. I set the Yi-M1 on a tripod and pointed it at a handy brick wall (at an approximate distance of about 10') and used the Yi 42.5mm lens for this series of tests. All focussing and exposure was set to automatic. The lighting changed due to cloud movement.

Pic01 is a 20MP still image reduced to 1640 pixels wide.
Pic02 is a 4K frame capture reducted to 1640 pixels wide.
Pic03 is a center (full resolution detail) crop of Pic01 (1640 x 1230)
Pic04 is a center (full resolution detail) crop of Pic02 (1640 x 1230)
[2017-08-26 NOTE: The 20MP still image was a RAW converted in Corel Paintshop Pro X7. I have not been using the built in JPEG yet.]

Aside from a small vertical discrepancy probably caused by my touching the camera, the center crops cover almost identical views.

Partial EXIF data from the 20MP still:
Exposure program Aperture priority, Scene capture type Standard
Exposure mode Auto, 1/100 sec., f/5.6, ISO 1250,
Metering mode Center weighted average, Gain control "High gain up"

Video settings unknown except No Exposure compensation used.
Everything "default".

[To Be Continued -- someday, but probably not very soon. ]

 

[fmw]

It is an interesting niche to inhabit - the interchangeable lens point-and-shoot. I think it may do reasonably well in the marketplace thanks to its $400 price. Serious photographers, though, are likely to prefer cameras with more control - things like manual focus.

 

[VidThreeNorth]

It is an interesting niche to inhabit - the interchangeable lens point-and-shoot. I think it may do reasonably well in the marketplace thanks to its $400 price.

Your pricing is off. At B&H it cost $300 US for the silver body with the zoom lens. The price for the black body and with both the zoom and the 42.5mm 1.8 lens it is $380 US. I bought the 2 lens package. These prices are closer to stock GoPro Hero5 range.

Serious photographers, though, are likely to prefer cameras with more control - things like manual focus.

The Yi-M1 has manual focussing for still photography including screen magnification and "peaking" for assistance. I have not tried it much but it seems to be working. Apparently it did not have manual focussing for video in version 1.x firmware. I have not been looking for it in version 2.0 yet. It was not mentioned in the changes list, but sometimes changes are not documented in this industry.

Beyond that, it is still too early for me to have a real opinion about the camera overall. My first 4K video test was made with a Panasonic 12-32mm zoom lens because I did not know what quality the Yi lenses had. So my first 4K video looked quite good. I have a bit more confidence in the Yi lenses now. The 42.5mm is very good. The PCMag report numbers for the zoom show that it is "ok". I have another 4K video made with that lens but I have not really looked at it yet.

 

[fmw]

The lenses do not have focusing rings. What you call manual focusing is simply manipulating the motorized AF system with the touch screen. I took the prices from Amazon. Sounds like it is a better deal than I thought. Post some images when you can.

 

[VidThreeNorth]

About the Lenses:

First, I wanted to post a clarification about my comment about the Yi zoom. Optically, from the test numbers it looks mediocre. There is no better way to put it. But if you look at the closest similar lenses from Olympus and Panasonic, their basic numbers are no better. It seems to be common characteristic of "cheap kit zooms". I expect that if I looked at the numbers for Canon, Nikon and Sony they would be similar. But on the Yi-M1, as I noted above, if I want to record 4K video, this lack of "stellar quality" is going to show up, and it will be a bit worse than in the low end Panasonics (GX-850 and GX-85). If I want to make a general rule, actually, if I use the Yi-M1 for 4K video, I think it would be worth it to stick with "prime lenses".

@fmw:

Yi only has 2 lenses. As you say, the 42.5mm has no focus control ring, but the zoom has a focus control ring. It works the same as most of the latest auto-focus camera lenses. You turn the ring and it causes the focusing motor to make the adjustment. The Yi 42.5mm lens is not the only Micro 4:3 lens to not have a focus control ring. My Panasonic 12-32mm zoom is also missing the focus ring, and there may be others too. On some of the cameras, like my Panasonic GF3, there really is no way to manually focus the lens [not even for still picture -- clarification added 2017-09-10]. But I expect that the modern Panasonic and Olympus camera bodies have some kind of focus control on the body. I have never tried any of those cameras except the ones I bought, so I do not know. Now that I know the camera is working, I "packed it away". I have too much other stuff to get done right now. This is why I said in the first message that I will not be adding to this topic for a while. Besides, just as a reference for the potential image quality, there are plenty of test pics available in the reviews I linked above. I posted the "wall" pictures because I had something to say that has not been covered in any of the reviews. Nobody seemed to want to "do the math". I don't blame them really. . . .

 

[fmw]

I think it will find a pretty good market in the U.S.

 

[VidThreeNorth]

As I wrote before, I have not had a lot of time to look at the Yi-M1 and there has been a general range of issues that I needed to settle. Mostly, they have to do with focussing.

On Aug. 13, I took the Yi-M1 to the "Taste of the Danforth" street festival in the east side of Toronto. I was only there briefly. I recorded a few video clips in 4K and 2K, and took a few pictures.

Everything was done using the kit zoom (12-40mm, F3.5-5.6). The camera was on a monopod above my eye level. Without a tilt screen this was difficult to use. Because there is no stabilization for the 2K and 4K video modes, I had to think about composing with later stabilization in mind. In general, I left more at the top of the frames than I usually would. At the bottom of the frames, I thought the speakers would cut off their feet anyway, so I left them close to the bottom of the frame. In retrospect, I think I should have zoomed back a bit and framed a bit lower, including the feet and below. Both clips looked in focus in the screen while I was recording.

If I had been located in front of the stage the scene would have been badly backlighted because there was no back to the stage, which was where the sun was. I generally prefer to work from an angle anyway, but in this case I used a specific angle which put the side of the stage as the background, reducing the backlighting problem. Exposure for the first clip was EV = +1. At some point I changed the exposure to EV = + 1.3. It might have been between the 2 clips or after the second clip. I did not make a note about it, but I found out later in the day that it was set at EV = +1.3, and I do not know if I did this deliberately or by accident. I do think it was deliberate and probably before the 2K clip which seems to be higher exposed, but this might be due to the change in composition with the 4:3 framing. The "Yi-M1" does not allow changes in the exposure compensation during video recording.
[2017-09-24 grammatical corrections.]

Additionally, I set my metering to "Center Weighted". I usually use this method on all my cameras for video and still images because it is generally consistent from camera to camera. It is not necessarily a "better" practice, it is just what I am accustomed to.

Also, I should mention that the Yi-M1 exposure compensation allows +/- 5.0 EV, which is very wide.
[Added 2017-09-10]

When I finally got to review the clips on a computer, I noticed that in the 4K clip, the singer (blue dress) seemed out of focus. I did some frame captures and confirmed it. While the depth of field was very deep, I could tell that the auto-focus had picked the store faces on the street [in the background -- added 2017-09-10] to focus on rather than the performers. As far as I am concerned, this clip is not acceptable.

On the other hand, the 2K clip focussed on the performers as I had wanted. This clip is usable, and in fact, overall, I like it. The sound is not wonderful, but then again, it was not that good coming from the stage.


Other problems:

I cannot use a "very slow zoom" on this lens. This is very annoying because it is a technique I used often. On my Sony CX240B camcorder in particular, I can zoom so slowly many people will not even notice that I am zooming. The Canon R40 can slow zoom too, but the control is a bit harder to use. When I tried to "slow zoom" the Yi "12-40" the zoom ring would stick and it caused a series of small jumps ("judder"). If I grip the zoom ring harder to try to get it to move, then the problem gets worse. The zoom ring is too flexible and it is the friction of the zoom ring against the lens barrel that is causing the sticking. The design and the choice of materials are the cause of the problem, and it is unlikely to improve with use -- it is not going to "wear in".


Back to Focussing:

During 4K recording there does not seem to be a way to "fine tune" the [automatic] focusing system. I tried the "face detect" but it seems to be disabled. The "face detect" box did not appear. The "touch focus" using the touch screen is also disabled. So whatever way the normal auto-focus works is what is happening. I think that the reason it did well with the 2K might have been because of the 4:3 format. There was simply less of the extra background in the composition to confuse the auto-focus. In that case, I think I am going to like using the 2K video a lot more.
[edited 2017-09-10 clarification]


Video Manual Focus Works BUT:
[Edited subtitle 2017-09-10]

One thing I checked later is that for video, the manual focus systems PARTIALLY work. Here is the situation for 4K and 2K recording. The other video modes are probably the same, but I have not tested them:

IF you have a [Micro 4:3 auto-focus lens with a] focus ring on the lens,
THEN you can select manual focussing BEFORE you start recording. You can chose "Manual" or "Manual w/Peaking". Once recording has started, this mode cannot be changed until the recording has stopped. [Clarified 2017-09-10]
- The display magnification does NOT work during these video modes.

IF the lens does NOT had a focus ring (like the 42.5mm prime lens)
THEN whatever setting you have selected for focussing will be ignored during video recording and it will use "Continuous Auto-Focus".

IF you are using an ADAPTED lens (or any other lens that does not have the electronics for auto-focus),
THEN you can select "Manual" focus or "Manual focus w/Peaking". On adapted lenses, you might also be able to adjust the aperture manually (depending on the lens and the adapter), and thus have depth of field control as well. But again, the Screen Magnification will not work during recording. [Clarification added 2017-09-10]


Still Not Tested:

I have not tested the colour temperature selection or "Scene" modes yet.


Best Approach:

If I am going to do "critical 4K video work", well first, I would probably not even use the Yi-M1 if it was that important, but if I do use it, then the best approach would be put the camera on a tripod and use a lens with a focus ring. I would set the focus to "Manual" and then use the magnifier to check the focus before starting the recording, and if possible, LOCK the focus and start the recording. I would avoid trying "focus pulling" because during the video recording I would not have the view magnifier available to check the re-focussing.
[Slight reformatting 2017-09-10]

Really, I think I might use the "2K" video mode quite a bit. The lack of excess background seems to help the focus work well enough, and on YouTube I think a lot of people will like the format.


About Video Formats:

The "2K" video format is a "GoPro" format and as such is it not new. I was surprised that they did not include another "intermediate" format. 2560 x 1440p is [a format that is] available in a lot of the Action Cameras. I like using it on the Git2. I was surprised that this format was not included on the Yi-M1. [Clarification 2017-09-10]

Further Corrections:
GoPro Hero5 Black 4:3 resolution formats are:

- 2.7K 4:3 @30/25 fps 2704x2028
- 1440p 4:3 @30/25 and 24/24 1920x1440

As far as I know, the 2048 x 1536 format was first used by Yi.

In the long run, this will not make a difference since video editors will handle the re-sizing (although I have some concern whether Pinnacle will allow outputting to any of these large 4:3 sizes). The 1440p 4:3 was the actual size that I was thinking about when I called the "2K" format a "GoPro format". The Yi "2K" format would resize downward to the 1440p 4:3 format nicely, including post-production stabilization.
[Above added 2017-09-12]

Pixel Counts:

3840 x 2160 = 8,294,400

2560 x 1440 = 3,686,400 [This format is not on Yi-M1]
[Clarification added 2017-09-10]

2048 x 1536 = 3,145,728

1920 x 1080 = 2,073,600

1280 x 720 = 921,600

If Yi were t0 add a 2560 x 1440 resolution mode it might be possible to use the full width of the sensor which would make better use of the lenses than the current 4K mode.
[2017-09-24 Added above paragraph.]

The 2560 x 1440 format is a 16:9 format which scales nicely both up to "4K" and down to "Full HD". The processing power needed for its support is only a bit more than the "2K" format and likewise the file sizes are only a bit bigger.

I do not know whether there might be problems with processing power if the 4K video is "enhanced" with future firmware changes, but both 2K and "2560" formats should leave enough processor "head room" for such firmware changes in the future.

Also, both "2K" and "4K" video formats are currently only available in 30 fps. Providing 24 fps (a "Pro" video speed) would also be appreciated.


Another Recommended Firmware Change:

I think that one firmware change I can recommend would be to provide a "narrow zone" auto-focus option for video. I would limit the focus to about the inner 1/2 of the frame and put a box of some kind indicating it. Either that or I would try to enable the "touch focus" system for video.


Aperture Working during 2K and 4K Video:

On a later test I confirmed that the aperture is not locked fully open during 2K and 4K video recording, which was mentioned in some of the reports prior to firmware 2.0 International.
[Added 2017-09-10]


About the uploads:

The "detail" files are full resolution.

 

[VidThreeNorth]

Two Issues that I will Eventually Get Around Too:

I will eventually get around to checking the "bit-rates" for 2K and 4K video. I am not that concerned with them. Visibly, they look pretty good.

I am concerned about how the controls work in cold weather. A few years ago I found that my LG cellphone' s touch screen would become unresponsive in cold weather. The temperatures below freezing were a problem. I do not know exactly how cold it was when the problems started. I remember a couple of times when I was using the cell phone camera for short videos and I could not stop the app. I had to pull out the battery. If the touch screen on the Yi-M1 fails in cold weather, then this camera can only be used indoors and in "fair weather". Even if this turns out to be a problem, I can find good use for the camera, but many people could end up very upset if this is a problem. I will probably not get around to this issue till winter.

 

[VidThreeNorth]

The final version of the 2K video, which features "Olivia and the Creepy Crawlies" singing "Gatsby" is at:

"20170813 Gatsby_Taste of the Danforth_Toronto"
""

I would like to thank them for their permission to make this video available on YouTube!

Processing was Magix Movie Edit Pro 2016 Plus 64-bit on my ASUS M32BF AMD A8-6500 and Windows 8.1. The included "proDAD Mercalli 2.0" plug-in did a particularly good job stabilizing the video, and particularly, dealing with "shutter roll" which occurred early in the video as I was "getting set up" and the zoom "judder" later in the video, caused the sticky manual control ring.

Unfortunately, a bug in the program stopped the title screen from fading out, but someday I might be able to fix that.

I do not know if Pinnacle would have allowed me to render a 2048 x 1536 output file, so for now, I might not have any other options.


Re "2K" video format:

GoPro Hero5 Black 4:3 resolution formats are:

- 2.7K 4:3 @30/25 fps 2704x2028
- 1440p 4:3 @30/25 and 24/24 1920x1440

As far as I know, the 2048 x 1536 format was first used by Yi.

In the long run, this will not make a difference since video editors will handle the re-sizing (although I have some concern whether Pinnacle will allow outputting to any of these large 4:3 sizes). The 1440p 4:3 was the actual size that I was thinking about when I called the "2K" format a "GoPro format" (above). The Yi "2K" format would resize downward to the 1440p 4:3 format nicely, including post-production stabilization.

I tested the "2K" video format when I make the "brick wall" recordings and it uses the full sensor array. The captures I made showed the same coverage as the 20MP still picture. This also means that it uses the Micro 4:3 lenses fully, so there is no need for any special considerations about lens performance or angle of coverage.
[Added 2017-09-24]

 

[VidThreeNorth]

Bit Rate for Yi "2K" video mode:

I would prefer to wait until I have a few video clips to take stats from but lately I have been using my camcorders for video work, so I have not accumulated video files taken on the Yi-M1 yet. But the "Olivia and the Creepy Crawlies" video has been made available, so at the very least if I take the stats from the camera file for that video, you can at least have some idea of how well this video mode works.

These are the stats that I got for the original camera file from the VLC media player and the Window 8.1 OS:

VLC stats:

Stream 0
Codec: H264 - MPEG-4 AVC (part 10)(avc1)
Language English
Resolution 2048x1538
Display resolution: 2048x1536
Frame rate: 29.970029
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


Windows 8.1 Properties (partial stats):

Size: 804 MB (843,740,520 bytes)
Created August 13, 2017, 15:51
Video:
Length: 03:44
Frame width 2048
Frame height 1536
Data rate 30002kbps
Total bitrate 30130kbps
Frame rate 29 frames/second [Windows Properties rounds the frame rate downwards 2017-09-24]
Audio:
Audio bit rate 127 kbps
Channels 2 (stereo)
Audio sample 48kHz

Calculation:

Video bit rate:
30,002 kbps /1024
~ 29.3 Mbps

Overall bit rate:
Approximate Mbps: 30130 / 1024
~ 29.4 Mbps

Confirm:
224 sec x 29.4 = 6,590,375 Mbits
6,590,375 Mbits / 8 = 823.9 MBytes
(~ 863,887,360 bytes) (a slight discrepancy, close enough)

So the target bit rate is probably ~29.5 Mbps

Yi Website accessed 2017-09-16 15:48

No bit rates are provided by Yi Technology in the specs in the manual nor online.


Comparing the Yi "2K" format with good "Full HD" camcorder files:

Yi's "2K" 2048 x 1536 = 3,145,728 pixels
"Full HD" 1920 x 1080 = 2,073,600 pixels

29.5 Mbps * (2,073,600 / 3,145,728)
~ 19.4 Mbps

Compare this to a Canon Vixia HF-R70: MP-4 files:
Full HD 1920 x 1080 @ 30 fps
- High Quality = 24 Mbps,
- Standard Quality = 17 Mbps

So, on a "per pixel" basis, the Yi records its "2K" video at a better data rate than a Canon R70's "standard" "Full HD" quality, but not by a lot.

Or looking at it on a "per frame" basis, it is 100 x 29.5 / 24
~ 23% better quality than Canon R70's High Quality 30 fps "Full HD" frames.

Considering that the Canon Vixia HF-R70 is a well respected consumer camcorder the numbers are very good. It is no surprise that the YouTube uploaded final file looks as good as it does.

 

[VidThreeNorth]

"4K" Video Bit Rate Calculation:

About the Test Clip:

"P8120005.MP4" [Recorded 2017-0812, Firmware 2.0 International]
- I will not be posting this clip anywhere. It is nothing more than some panning and some zooming. The sample capture shows about all there is to see really.

As I wrote above, I had not intended to try calculating bit rates until I had a variety of clips from which I could chose. But then I decided to post calculations based on the "2K" clip that was the basis for the "Olivia and the Creepy Crawlies" performance video, and since I had done that much, I looked through my "4K" clips to see if I had one that might give sufficient stats that I could post. I also decided to check further and see if any of the previous reviews had stats. When I checked their Yi Technology's English website I did not find any, but it is possible that they have published them somewhere else where I would have missed it. Or an astute writer could have asked the question to Yi and the information provided directly to them. I found such a list in dpreview.com's:

According to the "DPreview.com" stats page for the Yi-M1, the video bitrates (with an early firmware, possibly pre- "1.0.20 International") were as follows:

"3840 x 2160 @ 30p / 75 Mbps, MOV, H.264, AAC
1920 x 1080 @ 60p / 30 Mbps, MOV, H.264, AAC
1920 x 1080 @ 30p / 15 Mbps, MOV, H.264, AAC
1920 x 1080 @ 24p / 15 Mbps, MOV, H.264, AAC
1280 x 720 @ 60p / 15 Mbps, MOV, H.264, AAC
1280 x 720 @ 30p / 10 Mbps, MOV, H.264, AAC
1280 x 720 @ 24p / 10 Mbps, MOV, H.264, AAC"

See: "https://www.dpreview.com/reviews/yi-m1/2"
- accessed 2017-09-19 14:57

Looking at the above table, the first thing to note is that there is no "2K" bit rate. I assume that this is because the "2K" video format was added later. So this calculation will be aimed at verifying the "4K" bit rate (75 Mbps).


The test clip was one of my first on this camera. At the time, I did not know if the Yi lenses were good or not (it turns out that the lenses were both respectable, particularly the 42.5mm) so I used my Panasonic H-FS12032 (12-32mm, F3.5-5.6 manual zoom) lens. This lens is widely owned. Most owners of Panasonic Micro 4:3 cameras have this lens, so it is a good reference. A prime lens would be a bit sharper, but I did not have access to my 14mm lens. Exposure is compensated (EV = +2.0) for shadow detail, but is still within range to maintain cloud detail.

This particular video clip tested panning and zooming. It was mainly recorded at 32mm but I zoomed all the way to 12 and back to 32. It lost focus for a while zooming in [toward telephoto -- corrected 2017-09-24], but got it back fairly quickly. The +2 exposure was intended to pickup shadow detail but, as mentioned, without losing the clouds. Because this clip used a lot of zooming and panning, it might not be the sharpest throughout, but then again, it is probably representative of what can typically be expected. Focus is automatic, so panning and zooming probably resulted in "hunting" throughout, but it only "lost focus" obviously once during the "zoom-in" test, and focus was re-acquired quickly after the zoom ended.


VLC Codec Info:
Stream 0
Type: Video
Codec: H264 - MPEG-4 AVC (part 10)(avc1)
Language English
Resolution 3840x2160
Frame rate: 29.970029
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

Win 8.1 Properties (partial)
1.32 GB (1,418,384,200 bytes)
Video
Length 02:29
Frame width 3840
Frame height 2160
Data rate 76023 kbps
Total bit rate 76151 kbps
Audio
Bit rate 127 kbps
Channels 2 (stereo)
Audio sample rate 48 kHz.

Calculations:
Video bitrate 76023 kbps / 1024 ~ 74.2 Mbps
Total bitrate = 76151 kbps / 1024 ~ 74.4 Mbps

The 74.4 Mbps bit rate is close enough to the estimated 75 Mbps, and the overall image quality (and sound quality) are good enough to confirm that the camera is performing as well as these numbers would imply.

I have recently recorded some 4K "stock" using the Yi zoom. Most of it was manually focussed and showed good sharpness and detail, and yes, the bit rate for those clips are a bit higher despite high motion in the content from blowing leaves and plants. eg. clip P9210006.MP4 Video data rate 76,955 Kbps, Total bit rate 77,084 Kbps (~75.3 Mbps).
[2017-09-24]

The bit rate is very respectable but not earth shattering. Yes, if you pay a lot more, say for a Panasonic GH5, you can get better results, but I do not know of anything with the Yi-M1's flexibility (ie a proper interchangeable lens system with focussing and aperture controls) within this price range that has this performance. Also, I have to wonder whether a sharper lens might make some difference? As I wrote earlier, lens sharpness could have a stronger effect on the Yi-M1 than on cameras with less or no sensor reduction, and while the Panasonic H-FS12032 is a very good lens compared to similar "kit zoom" lenses, a good "prime" lens would be substantially better.

Recently I have found out that the Yi-M1 "4K" video quality is probably better than the "Z-Camera E1" bit rate, which according to Videomaker, achieves about 60 Mbps, though the E1 might have other advantages I have not heard about.

Comparing this to the Canon R70 at 30 fps Full HD. Scale the Yi "4K" bit rate down to a Full HD equivalent number (74.4 Mbps / 4) ~ 18.6 Mbps. So this particular clip shows a bit more detail per pixel than a Canon R70 recording Full HD @ 30 fps in its "Standard Quality" 17 Mbps. and only a bit less than the 19.4 Mbps that the camera achieved on the "2K" video. On a per frame basis, the 74.4 Mbps is still a huge increase over the Canon's "High Quality" Full HD 30 fps' 24 Mbps.


Captured Sample Frame:

The captured sample frame was during a pan, so the focus might be "hunting" a bit, but it looks representative of the best this setup can do. In fact, because the depth of field is very deep, I am not sure where it is focussed. I think it is focussed on the signal light stand, but it is very difficult to tell.

At EV = +2.0 the shadow detail was passable even without adjustments. As noted above, the clouds were completely "within bounds" meaning no loss of detail. While this frame (and the whole clip really) seem to allow a slight increase to around EV = +2.3, I would not want to risk it in a situation because such clouds are unpredictable. A bright patch could clip. Testing various alterations show that a lot of "hidden" detail has been maintained.

As usual, the "Detail" crop is full resolution. I have also uploaded some variations. The variation of the full frame file shows an "equalized histogram" bringing out the cloud details, and the variation of the "detail" crop is a more subtle histogram adjustment (mid-tone compression 10) which is just enough to bring out shadow details.

"Cap01-09h23m59s387-1920.jpg"

"Cap01-detail-09h23m59s387-1640.jpg"
- a full resolution detail

"Cap01-Alt-09h23m59s387a-HistEq-1920.jpg"
On the full frame, an "equalized" histogram brings out dramatic cloud details.

"Cap01-detail-Alt-09h23m59s387b-MidComp-1640.jpg"
On the detail clip, a milder Histogram Adjust, "Mid-tone compression 10" is a more subtle alteration that is enough to reclaim more shadow detail.

 

[VidThreeNorth]

I put my finding about manual focussing the Yi-M1 during video recording back on my Sept. 8 post, but if you want to see how the manual focus is done, it is demonstrated in the following video:

"Yi M1 - Manual Focus in 4K Video Recording Mode"
posted by "mygiguser" Sep 15, 2017

 

[VidThreeNorth]

I posted my first 4K video from the Yi-M1 to YouTube on Sept. 29, 2017, but I did not post a link to it "here" because I did not feel that it warranted bringing it to this forum's attention. It was a good enough video for my purposes (to provide a record of an event related to the Invictus Games) but it was not up to a high technical standard, and furthermore, the reasons for its short-comings were not primarily due to the hardware. It was partly circumstantial, and partly due to a poor choice on my part. I had a backup camcorder with me (Sony CX240B) which I knew would give me a better chance for a good video, but only in Full HD, yet I decided to take the risk on the Yi-M1 anyway. I will explain the situation a bit:

Lately my Condo management has decided to do work on the building which requires me to re-arrange everything. And yes, I work out of my condo, so that means re-locating things I use for work, including piles of notes and references, and equipment. So I am pressed for time and generally disrupted in everything I do and not by my choice. I wanted to cover the Invitus Games, but I did not have the time, except for this one event -- an evening concert. The week before I had a chance to record some Fall "stock" clips using the Yi-M1 and its kit zoom. I used manual focus, and used it on my usual monopod. That was the closest I came to practicing with this setup.

For this concert, I did not know the venue. Toronto City Hall can be set up in different ways, and this time it had been divided to provide an area for "Wheelchair Tennis". I had not seen how this was done. But I knew I was going to arrive late, so I expected that I would be far from the stage. So I decided to use my new Olympus "ED 40-150mm f4.0-5.6 R" manual zoom lens. For 4K, with the 2.8x crop factor, this performs like a 112 - 420 mm zoom on a 35mm camera. My last 4K attempt using the 12 - 40mm kit zoom for a "music concert" had turned out good except for the focus, so it was not unreasonable to think I could do this, as long as I stayed, perhaps, below 60mm. But this time I would be trying manual focussing.

The big issue was that I did not have time to practice with this setup before the event. Normally, I try to practice with my setup the night before recording--every time. The difference is that when a setup is new to me, I practice longer. With a setup as new as this, I would have liked an hour to check myself out on it, and maybe some time on my backup camcorder too. But I did not get a chance to do any of my usual practice. I just charged my batteries, packed up my gear, planned the trip, and tried to get some sleep.

When I arrived and setup, I found that the crowd was not so bad. The lens forced me to setup further away than I would have with the kit zoom or the Sony. The stage was very wide, which meant I was going to have to manually focus a lot from one end of the stage to the other. And the stage had multiple levels, so I could not lock the up-down tilt on the monopod. In effect, it was like handheld except the camera height above ground was constant. With time to think it over, I could have packed up the Yi-M1 and used the Sony CX240B. But I wanted the 4K video if possible, and my last video turned out fairly well after stabilizing in post. So I decided to give it a try.

The result was worse than I expected. The lighting was dark. I had expected "normal" spotlights. That did not happen. Also, because I was not as familiar with the camera as I thought, I mistook the "Mode" wheel for the EV control. I turned it and the EV did not change. What was going wrong? Also, I could not set "manual focus with peaking". I could only set "manual focus". Two or three songs into the show I looked at the camera and realized what I had done. I had turned the Mode wheel, eventually to Panorama instead of setting the EV. Luckily, there was no effect on the video. Moreover, the lighting apparently did not need much EV correction. I think EV = -0.3 would have been best, but no adjustment at all was good enough. As for focussing without "peaking", I found it was not worse than with "peaking." I find that the "peaking" only helps me if I have magnification, and that is never available during video recording on the Yi-M1.

But the worst thing was that without the locked tilt, and no stabilization, at that distance, the camera shake caused blurred frames that I could do nothing about in post. Stabilizing in post does work as far as aligning the frames, but a blurred frame is still blurred.

So how much of that was due to my lack of practice the night before? And did it make sense to use the Yi-M1 in this situation when I had the Sony CX240B ready in my camera bag anyway?

Did I learn anything? Yes, In this situation I would not bother with the Yi-M1. I would drop back to the Sony and bring home a better Full HD clip. But even after I explain all this does it help anyone else? I do not know. All I can think is "maybe". So I decided to post this for now, but do not be surprised if I replace this someday with a link to a better 4K video, or at least one that shows something more specific about the camera.

"[UHD] 20170926 James Blonde - Cynical - Invictus Games Concerts"
""

 

[VidThreeNorth]

Olympus ED40-150mm f4.0-5.6 R MSC at UHD

The Olympus ED40-150mm f4.0-5.6 R MSC zoom lens is a "budget" zoom lens. I would call it a "semi-kit" zoom because it is often included in a 2 zoom kit for Olympus cameras. This file set shows what it could do in on the Yi-M1 in this particular situation. If you saw the video, you will know that a high percentage of the frames were out of focus or motion blurred, so it took me a bit of work to find a couple of frames that give a real idea of how good this combination could be when used "properly". I deliberately chose frames and detail crops with the band's fans in mind, but the lens can be adequate for general use. It is not a wonderful lens, but it is not terrible either. In "full resolution" you will see typical YUV 4:2:0 aliasing. The reduced "1920" images "reassemble" these anomalies to an extent. Due to the "smoke machine" use, I advise not to assume that typical contrast is demonstrated in these frames. Imagine how the video would have turned out if I could have held onto this quality throughout. That is always the goal.

"Pic01-09h36m53s437-1920.jpg"
The black vertical mic stands in the right foreground are surprisingly sharp in both the 1920 reduced and the 1640 full resolution detail files. I think the auto-focus would have had problems if I had used it. If you look at this frame you see almost everything far behind the guitarist. There are some heads in the foreground but I think maybe 70 - 80% of the frame is behind him. My earlier experience seems to indicate that this composition would have focussed behind the guitarist. In this frame I think I have focussed just a bit in front of him -- on the mic stands themselves.

"Pic01a-detail-1640.jpg"
Full resolution "detail crop of the above frame.


"Pic02-08h59m09s086-1920.jpg"
Ampeg, Marshall and Yamaha would be happy to see this picture. Their logos have shown up very well. I think that the bass is a MusicMan. That logo was a bit too small to show up. The drum kit and mic stands again, are good places to look for the resolution and detail. Looking at the drum sticks, brings up the question of shutter roll. I looked throughout the clip, and yes, I can see some shutter roll, but it did not seem bad. I think that there is significantly more in 2K, which is understandable since the full sensor is being read. The Panasonics have been shown to have less shutter roll than some other 4K video cameras. I think the Yi-M1 is probably about as good. Unfortunately, I have not made a formal test of this issue yet.

Here, the big LED stage display is a significant problem for autofocus. The moire pattern on the screen in the "detail" file shows that the individual LED lights are almost in focus in this frame. I have seen clips where I am certain that a camera's autofocus had gotten fooled by a big screen like this and focussed on it instead of an act in front of it. So again, I believe that autofocus would not have worked for this clip. But in this case it would not surprise me if some other more expensive cameras would have done poorly as well.

"Pic02a-detail-1640.jpg"
Full resolution "detail" crop of the above frame.

 

[VidThreeNorth]

Apparently, on Oct. 30, 2017 Yi Technology released V3.0-int firmware for the Yi-M1 camera. I have downloaded the file and hope to complete the upload soon. Here are the release notes:

"Version: 3.0-int
Release date: 10/30/2017
The firmware is only for the international version of YI-M1 mirrorless camera.
It is strongly recommended that all users upgrade YI Mirrorless App to 3.0.0 to implement new features.
Firmware version 3.0-int update logs:

NEW:
1. Added exposure control in the mode P/A/S/M for video capture
2. Added AF/MF control for video capture
3. Added RAW+JPG format for still capture
4. Added still image option in Time Lapse
5. Added the option to shoot 1-3 images when using the self-timer function with an initial 2 second or 10 second timer setting
6. Added a histogram in live view mode for both still and video capture
7. Added new focus viewing modes with improved image quality including: 6x, 8x and 10x magnification in MF mode
8. Added 2 ways to show the Metering mode in the UI
9. Added the user guide (Restoring factory settings to view)
10. Added multiple selection of images for deletion.
11. Added 2 new options for the display grid
12. AE algorithm optimization
13. Master Guide template display logic optimized
14. Improved AF speed

Fixed:
Fixed some known bugs, and improved the overall stability of the system.

Changed:
Change the thumbnails in each screen from 9 to 12 images."