Yi Technology -- Yi-M1

[VidThreeNorth]

I installed the Version "3.0-int" firmware last night. I have not had time for thorough testing, and as usual I cannot even predict when I will, but I did some quick testing of issues for 4K video:

Yi 42.5mm F1.8 lens: (no focus ring)
During 4K video:
- cannot set Manual focus (no change from "version 2.0 Int.")
- EV comp is working using dial
- Touch Screen focus works during 4K video

Yi 12-40mm F3.5-5.6 zoom lens: (has a focus ring)
- Manual focus can be set and used (with or without "peaking" -- no change from "version 2.0 Int.")
- EV compensation is working using the dial
- Touch Screen focus works during 4K video

I have not tried face detect during video yet [See Below], which would also be a big asset, but this much goes very far in making the camera a viable video tool. The "touch screen focus" is best used on a tripod or at the least, a monopod, but that is how I work for most of my video recording anyway.

[2017-11-04]
I just ran a quick test of face detection in 4K video using the Yi 12-40mm F3.5-5.6 zoom. It is working. It might not be as good as some other cameras, but it definitely makes this a more useful video camera.

 

[VidThreeNorth]

Adapting Lens To Yi-M1

Equipment Used:
Yi-M1 (Firmware Vers "2.0-Int")
Monopod
Minolta 28mm F2.8 MD lens (Used, under $50 Cdn)
Fotodiox PRO MD - MFT" adapter MFR # MD-MFT-P (~$60 US)

I have been using adapted lenses on my Yi-M1 and my Sony a5000 lately. The Yi-M1 attempts were aimed at use for 4K video. My testing so far have been using a Minolta 28mm F2.8 MD (fully manual) lens for 35mm film cameras through a "Fotodiox PRO MD - MFT" adapter, but without a lens hood.

This set of files are converted from RAW and then exported without alterations (except the cropping, resizing and compressing). In general, I was quite happy with the "PA240002.DNG" image.

Both Source files:
Settings: F5.6-6.3

"PA240002_-rs1640-C5.jpg"
PA240002 is a 20MP DNG file. This version has been resized down to 1640 x 1230, but with more compression than I usually use (Paintshop Pro Compression level 5). I have not provided a corner "Detail" file in this set because I plan to provide them later in actual 4K video frame captures. But since the lens is designed as a 35mm film wide angle, one can expect that "corner" sharpness performance when used on a Micro 4:3 camera is close equaling the center, and I can report that it in fact does hold true. The equivalent 35mm film camera view for this file would be roughly a 56mm lens.

Partial EXIF:
Date time: Oct 24, 2017 17:13:13
width 5200
height 3902
Pixel height 3888
Pixel width 5184
Component config YCbCr
Exposure: Normal Program
Scene capture Standard
mode: Auto exposure
bias: 0.00 ev
time 1/100
ISO 1000
Metering: Center weighted average
Gain control: High gain up

"PA240002a-4K-1920-C5.jpg"
This is a crop from the 20MP file representing the coverage of the camera's 4K video mode. The actual equivalent 35mm film camera equivalent view would be roughly a 78.4mm lens. This jump is a result of the 4K video mode using a reduced sensor area from the full M 4:3 sensor. Again, the Paintshop Pro JPEG compression is level 5.


"PA240002b-Detail-1200-C1.jpg"
Lately I have tried to standardize on 1640 x 1230 pixel "Detail" crops, but for this set I found that the file sizes required too much compression. I have reduced to 1200 x 900 in order to provide minimum JPEG image data loss. I should caution everyone by reminding that "noise" increases file size as well as detail, and I have no way of breaking out the noise component, so while the file sizes appear impressive, do not assume it is all a result of quality. Still, the nature of the composition does imply a fairly high detail component.


"PA240009-1640-C2.jpg"

Partial EXIF:
Date time: Oct 24, 2017 17:16:22
width 5200
height 3902
Photometric Interp: Unknown (32803)
Pixel height 3888
Pixel width 5184
Component YCbCr
Exposure: Normal Program
Scene capture Standard
mode: Auto exposure
bias: 0.00 ev
time 1/320
ISO 200
Metering: Center weighted average
Gain control: Low gain up

This image has a serious fogging issue caused by stray light within the body. Immediately after seeing this image I started looking for a lens hood. My next projects were done with a rubber lens hood, but I have a metal lens hood on order.

When adapting lenses that have significantly larger coverage than sensor size, one can expect that body flare can become significant and that an appropriate lens hood will be beneficial. This 28mm lens was intended to be "wide angle". In normal Micro 4:3 usage (full sensor) it is roughly a "standard lens" (56mm) and in 4K video for the Yi-M1 it becomes a 78.4mm equivalent short telephoto. For these latter uses, "narrower" lens hoods become appropriate. But all the light from the designed "wide angle" coverage is coming into the body. Depending on how well "baffled" the body is, light can then be reflected onto the sensor. The "56mm" equivalent use should be perfectly paired with a standard lens hood. The "78.4mm" equivalent use is harder to match. An 85mm lens hood might cause vignetting depending on how "tightly" it has been designed.


Fotodiox PRO MD - MFT" adapter MFR # MD-MFT-P (~$60 US)
As for the Fotodiox adapter, overall, the construction is based on two main pieces. The body of the device appears to be forged, and with minimal machining. The rear mount (to the Micro 4:3 camera) is part of the main body. It cannot be removed and can be considered unrepairable. The front "lens" mount is a separate fully machined piece. In theory, it could be repaired, but I doubt if the repair part will be made available. Overall, it is probably intended to be disposable. This is a problem with this device because the rear mount is not as tight a fit as I would have liked. I tried to "rock" the lens in the mount and it seems secure enough, but how long it will last will depend on how fast the black coating and the underlying metal might wear out. Oddly, a standard Micro 4:3 rear lens cap will not work on this mount. It is too lose and would easily fall off.

The adapter has a tripod mount (which I believe can be removed, though I have not tried to do so), but the front-back length of the mount is short (about 19mm) and I think this could cause dangerous stress in use with long lenses.

 

[VidThreeNorth]

Earlier I mentioned that I wanted to test the touch screen in cold weather. Due to a cold snap, I got a chance to run a test today at about 10:00 EST. Temperature was about -7C. I made a couple of 2-3 min 4K video clips using the "touch screen focus" -- a new feature with the new "Version 3.0-Int" firmware. There were no problems for this brief test. I cannot say more than that. I would take a second camera with me in temperatures below 0C anyway. Actually I always have second camera, so that is nothing special.

 

[Braineack]

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

why? the sub-par quality images it produces?

 

[fmw]

I didn't suggest that it would find a market with you or me.

 

[VidThreeNorth]

Well this puts me in an awkward spot because I seem to be the only one in this forum who has touched this camera, and I have not really paid it much attention -- not even as a video camera, and certainly not as a "still" camera which I take it is what Braineac is talking about. If you are talking about the 4K video capability, which is why I bought it, well, I would have to disagree. If used properly, it can produced some nice clips. Even there I only have a couple of real "work" clips (tagged for a specific project). Just about everything I have done has been testing.

Let me take a moment and comment about my testing reports. I do not go out of my way to produce reports for this forum or anyone beyond me. I go out and test the equipment I have specifically to see if I can use it for specific uses. I make point form notes as necessary -- again it is all aimed at me and what I need to know. If I have information that I think will be interesting or valuable "here" (or elsewhere), then I do a little more work to try and turn my notes into something readable (with variable success). I might make an extra cropped file or resize file or two, but I rarely do more than that. The images I post are good enough to demonstrate something that I found out along the way. They may be flawed in other ways, technically or artistically, but they showed something specific that I needed or wanted to know.

Take that "PA240002" file I posted Nov. 3. That image has so much noise some people might wonder why I used it. But personally, I can see the lens sharpness through the noise. I am used to having to evaluate equipment in this kind of situation. But if it was a situation where, if I were being paid to write a review, I would have waited for another opportunity and taken something better (not that hard really, I just needed a day with better lighting). But that's the point. I cannot take more time to make a special set of images for "other people". It would be nice if I had waited for another opportunity, but unless someone is paying me, I cannot afford the extra time. So, "good enough for me" is what I posted.

And really, I do have a version of the picture I actually like. I like the wild collage of textures, and there is some subtle eye direction and movement. But that is after "processing", which mainly was getting rid of the noise....

Braineack said:
> fmw said:

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

> why? the sub-par quality images it produces?

Here I would like some clarification. Are you writing about the still images in the main reports, or do you mean the stuff I posted "here"? Because, really, I haven't posted much in the way of still images at all. Mainly because I have not been using this camera for still images. Almost everything I have posted has been frame captures. And there I would have to say that a couple of the frames have shown the camera actually can do well for video. The 2K video of the "Creepy Crawlies" song turned out quite well, and I have received a couple of compliments from it. Then again, "taste" being what it is, I have had compliments and a couple of "likes" for the "James Blonde" clip, which is unarguably, "photographically poor".

Also the 20MP "brick wall" file turned out really nicely -- if you like that sort of thing.

As for the formal reports by DP and Image Resources, etc., mostly the JPEGs were done with old firmware, and Yi claims that the "ver. 3.0-Int" firmware does better. I do not recall having a JPEG file from the earlier firmware to make a comparison, so I will not be able to comment about that. I am hoping that one of the major reviewers will cover the Yi-M1 again with the latest firmware, so I can find out along with everyone else. I have just started taking still images with JPG+RAW set, but again, whether I get a good picture out of it in the near future is doubtful. I have little incentive to make the effort.


About RAW:

Now here, the Yi-M1 is about as good as it can be with a good sensor. If you stick a good lens in front of it, mostly you can get the same image you will get from a top end Panasonic (exactly the same sensor) or Olympus (from what I have seen, if it is not the same sensor, it is certainly not a better sensor). The only caveats are 1. if you are using auto-focus, the three camera brands work differently, and 2. there is no optical image stabilization (neither lens based nor sensor shift based) in the Yi -- only software (aka "digital") stabilization, and even then, not in either every situation.

I will summarize this: For still images, from RAW it can, at least sometimes, be as good as the best you can get from a Panasonic or Olympus. From JPEG? I don't know, but they are claiming it is "better" now.


fmw said:

> I didn't suggest that it would find a market with you or me.


Well, yeah, now that is a question isn't it? Exactly who is this camera good for? Actually, at version "2.0-Int" it was usable, and with practice, and careful use, I have been able to get a foot into the 4K video world. And at this point, that is all I want. I still prefer "Full HD". Also, the 2K (4:3 format) video has been a revelation for me. I do intend to use that capability.

With version "3.0-Int", I am just now evaluating whether I am going to use it for more than that much.

But would I tell a "beginner" to buy one of these? I think that having really good autofocus is important for a beginner's camera. Even with what I have seen so far on "3.0-Int" I do not think it is good enough for a beginner. But I need to do more testing to be sure. Right now I would recommend something else.

But if it isn't a "beginners" camera, and it isn't an "experts" camera (which is where I am place you two), then that makes it a good camera for "me" and maybe not much more. Now, that's a really small market.

Ok, well, maybe it has a good market as a "C-roll" 4K video camera in a general sense. If they can get some video people to try it out with the new firmware, they might actually have something here.

 

[VidThreeNorth]

Yi Firmware "Version 3.0-Int"

About the Focus:

Yi has changed the main focus method. In versions "1.0.20-Int" - "2.0-Int" focus seemed to consider the whole screen and chose the focus distance that covered the greatest proportion of the whole screen. This resulted in anomalous focusing, often behind the main subject. "Single point/touch" focussing was available as an option for still pictures, but not for video. Now, the primary focus mode is "single point/touch". When the camera starts, the middle section of the 81 on the screen is selected as the chosen focus area. Another area can be selected at any time by touching the view screen. In "touch shoot" mode a still picture can be automatically taken when focus has been achieved (there may be occasional missed focus shots, there were under 2.0-Int, but I have not tested it under 3.0-Int yet).

Face detection and tracking are now available as an alternative in all recording modes. I have only tested this capability a bit, so I cannot comment on its effectiveness.

Having tested the main focus mode with videos and still pictures, I think the 3.0-Int focus system shows a marked improvement over 2.0-Int and previous firmware.

About the Exposure System:

The exposure compensation system however, has both a major improvement and a regression.

Previously, exposure compensation had to be set before a video was recorded and it could not be changed during a video. Under "3.0-Int" exposure compensation can now be adjusted at any time during video recording. This is a major improvement. However, I found that when video recording starts, any exposure compensation setting is "reset" to "+0.0". So every video clip starts with no compensation. It cannot be pre-set before a video recording commences. That means that the start many clips one has to make the exposure compensation and then remove that part of the clip in editing. This is a very annoying problem, and I hope Yi corrects it quickly.

Stabilization:

Stabilization is only available during video recording and only in resolutions from "Full HD" and below. It does seem to work well within those parameters, but it is more limited than most "optical" stabilization systems, and it also means more "sensor cropping".

Is the Yi-M1 Good Enough For A Beginner?

I think that the current focus system could be handled by a beginner with reasonable expectation of success. The exposure system was adequate back with version "2.0-Int" and, except for the video exposure compensation resetting at the start of every clip, it is even better now that it can be adjusted during video recording. We can only hope that the video exposure compensation problem will be addressed quickly.

Yes, at around $300 - $380 US (depending on the kit), I think it can be considered for beginners. But I also have an old Panasonic GF3 sitting on a shelf. Except for the lack of 4K video, I think the GF3 does better. The focus system seems more reliable, and has more focus mode options available. So if I extrapolate the GF3 to the current GX850 would the Panasonic GX850 be worth paying the difference for? I would say that If I had the extra money, I would still recommend buying the Panasonic, or some of the other possibilities, most of which lack 4K video but would be better still cameras.

Uploads:

"PB210004a-rs1640-C2.jpg"
This is a 20MP still image taken with the 12-40 Yi zoom is properly focussed, though really it is hard to tell since the depth of field covers almost everything. Only some very far background objects are easily noted as out of focus. Exposure has been compensation +1.7 to compensate for the backlight condition. Other EXIF data: Color representation sRGB, F-stop F5.6, Exposure time 1/250 sec, ISO-200, focal length = 40mm, Max aperture 4.97, metering mode Center weighted average.

"PB210004b-Crop1640-C1.jpg"
A full resolution detail crop from the above picture.

"PB210011.MP4-18h19m09s319a-rs1920-C1.jpg"
A frame from a 4K video, resized down to "Full HD". Again, focus is correct even following the motion. Exposure has been compensated again, though it is not clear by how much. Probably +1.7 - +2.0.

 

[VidThreeNorth]

Well this is frustrating. About a week ago I went out and took some still pictures specifically to find out about the Yi-M1's "improved" JPEG support in "version 3.0-Int" and then last night the computer it was on died. That was the computer I do most of my video work on. Thankfully, because I still also use a "weaker" computer, I can still get some work done while I wait to fix the broken one, but that might take a week.

Anyway, a few days ago I posted some "close-up" pictures taken with the Yi 42.5mm short-tele/macro lens. I set that up specifically with a ruler to show the size of the object photographed (about 3" high and about 3" wide). The next day I decided to do a similar set of pictures without the ruler and slightly different lighting.
"Beginner Equipment Questions"
Beginner Equipment Questions.

For these pictures I forced "F8" and allowed the ISO and exposure time to be selected by the camera. In the previously posted pictures I "side-lighted" the toy, just to make it different from my usual approach. This pair is more typical, with it facing the "key-light" (Rembrandt style).

"PB250001a-rs1640-C1"
"PB250001.JPG"
Partial EXIF:
Width 5181
Height 3888
Bit depth 24
Color Representation [AdobeRGB}
F-stop f/8
Exposure tme 1/100 sec.
ISO speed ISO-400
Exposure bias 0 step
Focal length 43mm
Max aperture 3.65
Metering mode Center Weighted Average
35mm focal length 85
Exposure program Aperture Priority
Original JPEG file size 6,057,637 bytes


This picture is slightly "front focussed" with the back character a bit out of focus giving it some dimensionality, if you want it.

"PB250007a-rs1640-C1"
"PB250007.JPG"
Partial EXIF:
Width 5181
Height 3888
Bit depth 24
Color Representation [AdobeRGB}
F-stop f/8
Exposure tme 1/100 sec.
ISO speed ISO-500
Exposure bias 0 step
Focal length 43mm
Max aperture 3.65
Metering mode Center Weighted Average
35mm focal length 85
Exposure program Aperture Priority
Original JPEG file size 6,624,626 bytes
Created Nov 25, 2017, 13:08:14

This picture puts the both characters within the depth of field. Dropping further to F11 would have been better, and was within the scope of the lighting.

 

[VidThreeNorth]

Yi-M1 Standard Image:

This is an issue that I have left "hanging" since I loaded the "3.0 Int" firmware. Does the Yi-M1 generally create good image files internally or do you need to do everything yourself from the RAW file?

The problem is that aside from some technical aspects that are generally agreed upon, this is largely a matter of taste. Looking at the files posted in the previous post (the Sally and Snoopy images) and the files that I am posting this time, and the other pictures I have taken (still not that many), I think I can say that overall, I think the Yi-M1 does a good basic job of rendering pictures.

This is a picture I took this morning. It was from my first "snow day" set for this camera this season. Unfortunately, in retrospect, I misjudged the exposure compensation. I wanted to maintain the sky (cloud) detail, so I only raised the exposure +0.3 EV. I think I should have raised it about +1.0 EV. That means I lost shadow detail, and I had to deal with a lot of noise. But overall, the Yi-M1's version did well with noise removal, and I did not feel that it "overdid" the processing.

I am posting my own render effort as the main image, but the "detail" crop is from the Yi-M1 created JPEG file. The Yi-M1 did better then me.

"PC150013.JPG"
- this is the original JPEG file version which has not been posted

Partial EXIF:
Pixel Height 3888
Pixel Width 5184
Component config YCbCr
Color space Uncalibrated
White Point 0.31 0.33
Primary Chromatics 0.64 0.33 0.21 0.71 0.15 0.06
Exposure program "Normal program"
Scene capture Standard
Exposure mode "Manual exposure" * wrong
Exposure bias 0.30 ev
Exposure time 1/80 sec.
F number f/5.6
Max aperture f/5.6
Focal length in 35mm 80mm
ISO 640
Metering mode Center weighted average
Gain control High gain up


"PC150013.DNG"
Conversion parameters:
Temperature 5973
Tint 4
Brightness 1.0



"PC150013 -1d-Rsz-1641-C1.JPG"
- This is the version I made from the DNG (raw) file using the following steps:

My Processing:
"Digital Noise Removal"
Noise Correction
- Small 50
- Medium 50
- Large 50
- Correction blend 70
- Sharpen 0

"SmartFix"
Brightness
Overall 4 [Recommended 28]
Shardow 0
Highlights 10
Saturation 19
Focus 60
Black 8
White 12

- I changed the format of the picture removing some foreground leaving a 3:2 image ratio in the final crop before resizing it to 1641 wide.

"PC150013-JPG-Crop-1640-C1.JPG"
- This is an unadjusted "detail crop" from the JPEG created by the Yi-M1.

If you check some of the earlier reports by the usual sources I have listed above, the main criticism seem to be "over doing" the noise removal and, I think "over-sharpening". From the pictures I have seen so far, I think these issues have been well addressed. The detail crops I have posted recently and in this picture look well balanced. One might disagree for a particular image -- which happens with all cameras, but overall, it is doing quite well.

 

[VidThreeNorth]

Yi-M1 "HDR"
- 42.5mm F1.8 Macro/Portrait lens

This project started when I decided to post some pictures of my Sony a5000 using my Minolta 28-100 zoom. I expected to take a couple of quick "snap-shots" and post them. But when I saw the results I was disappointed and ended up working on the DNG files until I got fairly acceptable results.
[see: ""Konica Minolta 28-100mm f/3.5-5.6 D AF Zoom" #9" Konica Minolta 28-100mm f/3.5-5.6 D AF Zoom]

After I was done, it remained on my mind and I had a couple of approaches for improvement. The first was a simple change in the exposure to EV = +1.0. The second was to finally try out the HDR capability of the Yi-M1. Later in the evening I decided that it was a good idea for me to take the extra time and try out the HDR capability of the camera, so the decision was made to continue this as a new project.

One reason I have not gotten around to trying the HDR mode on the Yi-M1 was because the camera does not save a RAW (DNG) file when it is used this way. I guess that this is probably because it is doing so much and taking its time just making the HDR. If there were an option to save a RAW, what kind of file would it be? I would be happy it the camera saved the RAW image indicated by the EXIF data.


About HDR:
NOTE: This forum has a specific area to discuss HDR which I have not had time to look into yet.

As far as I know, "HDR" and "WDR" mean the same thing. Both terms were coined by people trying to cope with the problem of making pictures where the dynamic range of the subject matter exceeded the inherent reciprocity of the hardware. The techniques go back to the concepts like "dodging and burning" and other techniques developed in film photography to cope with the same problem. Some companies might be trying to define certain techniques as one or the other, but I have seen the terms used interchangeably, and I have not seen anyone state a specific difference. If anyone knows, or even "thinks" otherwise, let me know.


This Image Set:

I used a slightly different but similar setup as yesterday, but the lighting is a bit less diffused this time. For this set of images I also decided to control the depth of field by using "Aperture Priority" and working with F8 and F11, and in the latter pictures selecting the point of focus with the touch screen.

I am not planning on posting "detail" crops for this set.


"PC210003a-rsz1640-C1.JPG"

I started this project with three images taken "regularly" with EV = +1.0 and DNG files. This is the last image with DNG. All the rest are HDRs. This is my "control" image. If I had taken yesterday's pictures with EV = +1.0 then I would have probably still worked with some of the DNG files, but I would have had an easier time, and probably better results. I probably would have posted this picture as-is. There is sufficient shadow detail to show the equipment using the adapter's tripod mount and how high it raises the camera. I was not trying to make it look "pretty", I was only trying to show the equipment I had been using lately.

Partial EXIF
Dimensions 5184 x 3888
Color representation Uncalibrated (AdobeRGB)
F-stop f/8
Exposure time 1/100 sec.
ISO speed ISO-3200
Exposure bias +1 step
Focal length 43 mm
Max aperture 3.65
Metering mode Center Weighted Average
35mm focal length 86
Exposure program Aperture Priority (forced F8)


"PC210005a-rsz1640-C1.JPG"

This file replicated most of the settings of "03" except that it is set for HDR and no RAW. Notice how that exposure time jumps from 1/100 sec. to 1/5 sec. and the ISO drops from ISO-3200 down to ISO-200. I did not do that, the HDR setting did that. Also, the picture is slightly zoomed/cropped compared to the control picture. If size is critical, one might have to anticipate, and maybe even calculate this difference. Also, the exposure is still EV = +1.0. In picture "11" below, using EV = +0, shadow detail starts to go away. So correct exposure is still a "judgement call" and bracketing is a good idea. I consider this image to be "almost perfect". The highlights have been muted bringing out the "AF 28-100" label and the exposure of the bright area on the lens mount is also reduced. Part of the camera body and parts of the tripod mount are sitting a bit too close to the "floor black" than I would like, but there is no detail of interest in those areas anyway. Likewise, at the front of the lens there is some silver area which is just inside the "white" value.

Partial EXIF
Dimensions 5184 x 3888
Color representation Uncalibrated (AdobeRGB)
F-stop f/8
Exposure time 1/5 sec.
ISO speed ISO-200
Exposure bias +1 step
Focal length 43 mm
Max aperture 3.65
Metering mode Center Weighted Average
35mm focal length 86
Exposure program Aperture Priority (forced F8)


"PC210007a-rsz1640-C1.JPG"

This file has slight composition changes (the angle of view) and F11 is used to increase the depth of field. The focus distance has been adjusted by using the touch screen focus in order to best use the depth of field. I tried a few focus points and this one turned out the best. I do not think there is a good way to check depth of field on this camera.

Partial EXIF
Dimensions 5184 x 3888
Color representation Uncalibrated (AdobeRGB)
F-stop f/11
Exposure time 1/2 sec.
ISO speed ISO-200
Exposure bias +1 step
Focal length 43 mm
Max aperture 3.65
Metering mode Center Weighted Average
35mm focal length 86
Exposure program Aperture Priority (forced F11)


"PC210011a-rsz1640-C1.JPG"

This file has another composition changes (angle of view) and it tests setting the Exposure Bias = +0.0. The average brightness of the image is darker, but the highlights and "floor black" still seem to be "contained." Comparing the two, I think it is best to adjust the exposure compensation as if it was a normally exposured picture. The lettering at the front of the lens is falling out of the depth of field, but I think if I had focussed a bit more towards the front of the camera I could have covered it without the "Sony" label losing sharpness.

Partial EXIF
Dimensions 5184 x 3888
Color representation Uncalibrated (AdobeRGB)
F-stop f/11
Exposure time 1/5 sec.
ISO speed ISO-200
Exposure bias +0 step
Focal length 43 mm
Max aperture 3.65
Metering mode Center Weighted Average
35mm focal length 86
Exposure program Aperture Priority (forced F11)


Conclusion:

The HDR setting can be useful, but yes, it takes a bit of practice to get the most out of it. It still helps to set the exposure compensation appropriately.

NOTE: Due to upload problems the rest of the files were uploaded 2017-12-22 09:31.

 

[jcdeboever]

micro contrast is lacking.

 

[VidThreeNorth]

micro contrast is lacking.

What are you referring to?

 

[jcdeboever]

micro contrast is lacking.

What are you referring to?

The look of the images. Not impressed at all.

 

[VidThreeNorth]

Yi-M1 Touch-Screen Focus Test
(Partial report -- for now, consider this a partially uncompleted test. I considered abandoning this test and re-starting it with the "3.0 Int" firmware. Shortly after I upgraded the firmware, the flowers were gone, so now I would have to pick a completely new subject, or wait till next year.)

Firmware version "2.0 Int"
Lens: Yi 42.5mm Portrait/Macro

In late August I decided that I wanted to test the accuracy and speed of the screen touch focus (and shoot) functions on the Yi-M1. I thought it should be easy. I just had to set up the camera on a tripod and use the screen touch to alternate between a "far" focus point and a "near" focus point. I thought that about 20 pictures of each to get the speed of the focus. I ran the test twice for about 80 total files.

When I looked at the files on the computer I found out that the "time" on the files were reported down to the minute rather than down to the second. When I checked the EXIF internal information I found the same thing. The problem is that I would need many pictures to get a good calculation of the speed. Or I needed a better method of testing. I could use a stop-watch (which I have), or perhaps I could set up a camcorder on another tripod behind the test unit and take a video of test. That would be best because I could post the video and people could see the speed directly.

I never did get around to running the re-test. But I did have the ~80 files of the test images.

Much later I found that by looking for the EXIF info from the DNG file using Corel Paintshop Pro X7, I could read the file time down to seconds.

The composition was high contrast illuminated flower with white petals against a background of trees, many of which were in shadow. My first set was done with EV = +0, which was not a terrible choice. But for my second set I decided that I should expose for the background in order to give the auto-focus a better chance. I can justify this choice because there are certainly situations when the main subject is in shadow and lightening the images is correct. Thinking about this later, maybe I should have adjusted the exposure downward for the flower petals, because they were a "real" point of interest and not just an imaginary possibility.

I do not know if the camera can be set to use the "touch" to adjust both the focus and the exposure together. I think that that would be the best solution. It might be possible. I have not gotten that far yet.

Still, that does inadvertently turn this into a "dynamic range test" and I did find out a bit about this camera and its sensor.

NOTE: No "highlight recovery", "noise reduction" or "sharpening" were used in the initial images, so these are "worst case" in those regard. The "P9030007-2" image was re-converted with an extra correction to the exposure which can be considered an attempt at "highlight recovery".

Unfortunately, the "1640" size detail crops turned out to be too big, and I had to compress them beyond the minimum, so more detail has been lost. If the file name ends in "C2" it means that I used compression level 2 which is the next step down from "C1" which is minimum compression. "C4" is two more levels down. I might try to re-do them again someday, but it might take me a while.

[Uploaded 2018-01-13] Correction Files:
I have uploaded new "Detail" crop files which are 1200 x 900 and saved as JPEG files with "C1" minimum compression. I will remove the previously uploaded "1640" files later.


"P9030006 -1" [from DNG]
Temperature 5260
Tint 14

Partial EXIF:
ASDK-00129
FlashPix version 01.00
Date and time Sept 3, 2017 17:17:19
Image width 5200
Image height 3902
Pixel height 3888
Pixel width 5184
Component config YCbCr
Color space sRGB
Exposure "Normal program"
Scene capture "Standard"
Exposure bias 0.00 ev
Exposure time 1/640 sec.
F number f/3.9
Max aperture f/1.8
Focal length 43.0
ISO speed 200
Metering mode Center weighted average
Gain control Low gain up

"P9030006 -1-Detail-Crop02-C1.jpg"
[Uploaded 2018-01-13]
Crop Starting at "0,2100"
- cropped to 1200 x 900 and saved with minimum compression to retain detail.


"P9030007-1"
Temperature 5416
Tint 19

Partial EXIF:
ASDK-00129
FlashPix version 01.00
Date and time Sept 3, 2017 17:17:21
Image width 5200
Image height 3902
Pixel height 3888
Pixel width 5184
Component config YCbCr
Color space sRGB
Exposure "Normal program"
Scene capture "Standard"
Exposure bias 0.00 ev
Exposure time 1/640 sec.
F number f/3.9
Max aperture f/1.8
Focal length 43.0
ISO speed 200
Metering mode Center weighted average
Gain control Low gain up

"P9030007 -1-Detail-Crop02-C1.jpg"
[Uploaded 2018-01-13]
Crop Starting at "2000,1400"
- cropped to 1200 x 900 and saved with minimum compression to retain detail.


"P9030007-2"
Brightness -1.0
Temperature 5416
Tint 19

SmartFix
Overall 15 [Recommended 28]
Shadows -30 [Recommended 0]
Highlights -15 [Recommended 10]
Sat 16 [Recommended 0]
Focus 0 [Recommended 42] *** No sharpening added in order to check the accuracy of the focus.
White Balance OFF
Black 10 [Recommended 10]
White 0 [Recommended 0]

 

[VidThreeNorth]

"DPreview.com" published a review of the Yi-M1 on Nov. 29, 2016. I have just noticed that they have made the following announcement:

"November 2017: Several issues raised in this review have been addressed in recent firmware. We are now shooting with an updated camera and hope to revise the review to reflect current behavior."

I do not know when this update will be published, but I, for one, congratulate them on their diligence. As far as I am concerned, the poor state of the camera's original release, and the long time it took Yi Technology to get to the "2.0 Int" and then the "3.0 Int" makes it understandable that few of the "Pro" reviewers have taken on this task. These reviewers have a lot of other cameras and equipment to test and correctly focus on the products in which their readers have the most interest. They do not "owe" the manufacturers their attention and efforts. They "owe" them to their readers. The originally released firmware, and even up to "1.0.20 Int" were a waste of the reviewers time.

DPreview's Yi-M1 Review page:

New kid on the block: YI M1 review