VidThreeNorth
No longer a newbie, moving up!
- Joined
- Oct 21, 2016
- Messages
- 1,175
- Reaction score
- 214
- Can others edit my Photos
- Photos NOT OK to edit
On Dec 29, 2021 "DP Review TV" posted their "Cheap Camera Challenge" on YouTube [Length 10:43 ""]. I think that this was a semi-serious effort. They bought the lowest cost "new" stills and video cameras (the still camera was a discontinued product, but bought "new") from their favorite store and ran their standard tests, and gave fair reports. Chris had a fun time with his Olympus stills camera, but Jordan's experience with his "Safari 5D" action camera was painful and expressed it so. I commend them for "sticking to the rule" of testing literally the cheapest products in their classes, but I commented that better video cameras were available for only a little more, though in retrospect, I do acknowledge that such cameras often need either modifying or working around shortcomings.
When I decided to move up beyond Full HD video (1920 x 1080p) around 2016 - 2017, the prices for the available cameras were a substantial jump. Instead, I decided to convert slowly and make do with whatever I could find to bridge the gap. Among my first efforts was the GitUp "Git2" action camera. This camera supported a "UHD-like" anamorphic format of 2880 x 2160 at 24 fps, which displayed as 3840 x 2160 "UHD". Since I could stretch this in post to UHD, I thought that it might be usable as "B cam" fill later in UHD productions. If it turned out not good enough quality for that, at least it might produce some good enough free standing clips that were a step up from "Full HD". Moreover, GitUp decided to try an "experiment" putting a rectilinear lens in some of the Git2's at a small price increase. This saved me buying such a lens separately and making the change myself. So that is what I bought.
GitUp called it a "90 degree" lens version. Aside from this special version of the Git2, they also sold three lenses separately, all called "90 degree" lenses. I think these were 3.85 mm, 4.1 mm and 4.35 mm focal lengths. I never could understand why they called all three of these lenses "90 degree" lenses despite their varying focal lengths, but only one of those lenses (3.85 mm) could really be called a "90 degree" lens. They were all slightly too long a focal length to actually cover "90 degree" for video. Maybe they thought they were "close enough", I don't know.
[Following link added 2022-01-28]
For more information about "90 degree" lenses, see: "90 Degree Lenses"
"90 Degree Lenses"
The first experimental "Git2 90 degree" (the one I bought) came with a nominal 4.35 mm all glass lens. It had noticeable barrel distortion, but was close enough to linear for my immediate needs. I do not have the specs of the other two lenses handy (hopefully I will find my notes later) but as I recall them, one was a 3.85 mm focal length all glass lens, which had the best linearity (which I will discuss further later), and last was, I believe, 4.1 mm focal length with 1 "plastic" element. This latter lens was used in the Git2P, which is still available today.
The files conformed to "normal" MP4 files with a typical file sizes like (actual sample file "2022_0106_130743_006.MP4") 745,647,986 bytes (3 min, 0 sec @ 24 fps). This particular example file had a video data rate of 32,753 kbps, audio data rate 111 kbps (monophonic, 48 kHz) and total bit rate 32,865 kbps. But for time lapse, this camera puts out files typically like this 737,583,210 bytes [actual sample file "2022_0121_071422_011.MP4"] size 737,583,210 bytes (the same 2880 x 2160 anamorphic formatted frame, 3 min, 0 sec but @ 30 fps), with a video data rate 32,759 kbps without a sound track, giving a file data rate of 32,759 kbps.
Notice the "30 fps" format for the time lapse file. This is possible because processing only 1 frame every second gives the processor plenty of time to get its work done. Other action camera makers have realized this as well. Time lapse became an interest feature which differentiated these cameras, and made them especially appealing for me. In fact, I still use them for time-lapse, even "today" -- though with a caveat that I do recommended considering modifications to the cameras.
One of my first attempts using my Git2 in "factory" condition (no modifications except perhaps a firmware update).
"20170130 Toronto Sunrise Git2 4.35mm 2.8 Lens [4K] v2"
posted Feb 2, 2017 [Length 4:24]
""
2 steps forward, 1.5 steps back
Unfortunately, while I did like the output of the camera as it came, most of the time, I knew that I there were situations where the barrel distortion would be unacceptable. For example, I could not mix it with the output of a better camera and lens because the distortion would become noticeable. Moreover, The Git2, design and construction resulted in imprecise lens to sensor alignment. The center axis of the lens was not necessarily in the middle of the sensor. This made correcting the distortion in post difficult, or impossible, depending on the limits of the software. Looking at the other separate lenses GitUp offered, I decided to buy the "3.85 mm" lens I have mentioned above because it was the most linear lens of the three.
The good news is that the linearity, as I expected, is so good that I can consider it to be "perfect" in that regard. In fact, for linearity, I do not think I have a better lens for any of my current cameras, and that includes my Panasonic Lumix and Sony Alpha cameras.
The not-so-good news is that the new lens had other problems. First, although I have not done formal tests, I do think that the 3.85 mm lens is not quite as sharp as the original 4.35 mm lens. Moreover, the flare pattern of the new lens, in general, is a bit worse than the original, though I actually prefer the flare in the new lens. If you look at the above noted "Sunrise" video, you will see that the original lens (which was used in that video) had one large, strong magenta "reflection" flare of the sun, which appears exactly opposite the sun, though the center axis on that lens. The posted video was recorded with a crop similar to the current camera's "2.2x zoom" crop. The new lens has a big magenta flare around the sun itself, then a smaller one close to the axis, past but close to the axis, and then another green spot, also on the opposite side of the axis, but a bit further away than the small magenta spot.
"01-Original-Git2-LensSJ5000plus-rsz1600-C1.jpg"
This is from a video recorded without "zoom", using the original "4.35 mm" lens which I removed from my "Git2" and installed into an SJCAM SJ5000plus, which uses a Panasonic sensor which is also 1/2.3" diagonal.
"02-Git2-NewLens-rsz1600-C1.jpg"
This is a frame from a video recorded on the Git2 with the "3.85 mm" lens that I bought separately from GitUp later. The Git2 firmware has settings that allow lower contrast and sharpening. I am using a low contrast and low sharpening. Both the SJ5000plus and the Git2 have EV control. I could have recorded this and the previous sample at higher EVs, but I was worried about the sun coming out and ending up with clipping.
"03-NewLens-Flare-Rsz1600-C1.jpg"
This sample is from a video I recorded on the Git2 with the 3.85 mm lens, but using a 2.2x crop. I have pasted it on a grey background at the relative size of a 1x zoom, and then re-scaled the final image. So the grey "frame" represents the field of view of an uncropped ("un-zoomed") video. It is intended to show that the flares are not that big if you are recording without any "zoom" cropping. I actually do like this lens.
"04-Git2-Rsz1600.jpg"
This is the Git2 with the new (3.85 mm) lens installed. The front element protrudes about 1/2 mm beyond the lens barrel and this results in a lot of undesirable flare. I made a temporary paper "lens hood" which helps a lot. It will not help much when the sun is actually in the composition.
"05-Git2-DetailCrop1600-C1.jpg"
[Uploaded 2022-01-28]
I thought about uploading a different "detail crop" because this image is a bit under-exposed, and as such, is a bit grainy, but realistically, considering it was such a cheap camera, and currently is not even available in this form (it was replaced by the "Git2P" which is currently still being sold) it is actually quite good performance, so this is "good enough". I still have not found my test notes (from about five or six years ago), but I believe that I focused it as about 1M. and everything else is covered by depth of field and sharpening. This is roughly how most action cameras work.
When I decided to move up beyond Full HD video (1920 x 1080p) around 2016 - 2017, the prices for the available cameras were a substantial jump. Instead, I decided to convert slowly and make do with whatever I could find to bridge the gap. Among my first efforts was the GitUp "Git2" action camera. This camera supported a "UHD-like" anamorphic format of 2880 x 2160 at 24 fps, which displayed as 3840 x 2160 "UHD". Since I could stretch this in post to UHD, I thought that it might be usable as "B cam" fill later in UHD productions. If it turned out not good enough quality for that, at least it might produce some good enough free standing clips that were a step up from "Full HD". Moreover, GitUp decided to try an "experiment" putting a rectilinear lens in some of the Git2's at a small price increase. This saved me buying such a lens separately and making the change myself. So that is what I bought.
GitUp called it a "90 degree" lens version. Aside from this special version of the Git2, they also sold three lenses separately, all called "90 degree" lenses. I think these were 3.85 mm, 4.1 mm and 4.35 mm focal lengths. I never could understand why they called all three of these lenses "90 degree" lenses despite their varying focal lengths, but only one of those lenses (3.85 mm) could really be called a "90 degree" lens. They were all slightly too long a focal length to actually cover "90 degree" for video. Maybe they thought they were "close enough", I don't know.
[Following link added 2022-01-28]
For more information about "90 degree" lenses, see: "90 Degree Lenses"
"90 Degree Lenses"
The first experimental "Git2 90 degree" (the one I bought) came with a nominal 4.35 mm all glass lens. It had noticeable barrel distortion, but was close enough to linear for my immediate needs. I do not have the specs of the other two lenses handy (hopefully I will find my notes later) but as I recall them, one was a 3.85 mm focal length all glass lens, which had the best linearity (which I will discuss further later), and last was, I believe, 4.1 mm focal length with 1 "plastic" element. This latter lens was used in the Git2P, which is still available today.
The files conformed to "normal" MP4 files with a typical file sizes like (actual sample file "2022_0106_130743_006.MP4") 745,647,986 bytes (3 min, 0 sec @ 24 fps). This particular example file had a video data rate of 32,753 kbps, audio data rate 111 kbps (monophonic, 48 kHz) and total bit rate 32,865 kbps. But for time lapse, this camera puts out files typically like this 737,583,210 bytes [actual sample file "2022_0121_071422_011.MP4"] size 737,583,210 bytes (the same 2880 x 2160 anamorphic formatted frame, 3 min, 0 sec but @ 30 fps), with a video data rate 32,759 kbps without a sound track, giving a file data rate of 32,759 kbps.
Notice the "30 fps" format for the time lapse file. This is possible because processing only 1 frame every second gives the processor plenty of time to get its work done. Other action camera makers have realized this as well. Time lapse became an interest feature which differentiated these cameras, and made them especially appealing for me. In fact, I still use them for time-lapse, even "today" -- though with a caveat that I do recommended considering modifications to the cameras.
One of my first attempts using my Git2 in "factory" condition (no modifications except perhaps a firmware update).
"20170130 Toronto Sunrise Git2 4.35mm 2.8 Lens [4K] v2"
posted Feb 2, 2017 [Length 4:24]
""
2 steps forward, 1.5 steps back
Unfortunately, while I did like the output of the camera as it came, most of the time, I knew that I there were situations where the barrel distortion would be unacceptable. For example, I could not mix it with the output of a better camera and lens because the distortion would become noticeable. Moreover, The Git2, design and construction resulted in imprecise lens to sensor alignment. The center axis of the lens was not necessarily in the middle of the sensor. This made correcting the distortion in post difficult, or impossible, depending on the limits of the software. Looking at the other separate lenses GitUp offered, I decided to buy the "3.85 mm" lens I have mentioned above because it was the most linear lens of the three.
The good news is that the linearity, as I expected, is so good that I can consider it to be "perfect" in that regard. In fact, for linearity, I do not think I have a better lens for any of my current cameras, and that includes my Panasonic Lumix and Sony Alpha cameras.
The not-so-good news is that the new lens had other problems. First, although I have not done formal tests, I do think that the 3.85 mm lens is not quite as sharp as the original 4.35 mm lens. Moreover, the flare pattern of the new lens, in general, is a bit worse than the original, though I actually prefer the flare in the new lens. If you look at the above noted "Sunrise" video, you will see that the original lens (which was used in that video) had one large, strong magenta "reflection" flare of the sun, which appears exactly opposite the sun, though the center axis on that lens. The posted video was recorded with a crop similar to the current camera's "2.2x zoom" crop. The new lens has a big magenta flare around the sun itself, then a smaller one close to the axis, past but close to the axis, and then another green spot, also on the opposite side of the axis, but a bit further away than the small magenta spot.
"01-Original-Git2-LensSJ5000plus-rsz1600-C1.jpg"
This is from a video recorded without "zoom", using the original "4.35 mm" lens which I removed from my "Git2" and installed into an SJCAM SJ5000plus, which uses a Panasonic sensor which is also 1/2.3" diagonal.
"02-Git2-NewLens-rsz1600-C1.jpg"
This is a frame from a video recorded on the Git2 with the "3.85 mm" lens that I bought separately from GitUp later. The Git2 firmware has settings that allow lower contrast and sharpening. I am using a low contrast and low sharpening. Both the SJ5000plus and the Git2 have EV control. I could have recorded this and the previous sample at higher EVs, but I was worried about the sun coming out and ending up with clipping.
"03-NewLens-Flare-Rsz1600-C1.jpg"
This sample is from a video I recorded on the Git2 with the 3.85 mm lens, but using a 2.2x crop. I have pasted it on a grey background at the relative size of a 1x zoom, and then re-scaled the final image. So the grey "frame" represents the field of view of an uncropped ("un-zoomed") video. It is intended to show that the flares are not that big if you are recording without any "zoom" cropping. I actually do like this lens.
"04-Git2-Rsz1600.jpg"
This is the Git2 with the new (3.85 mm) lens installed. The front element protrudes about 1/2 mm beyond the lens barrel and this results in a lot of undesirable flare. I made a temporary paper "lens hood" which helps a lot. It will not help much when the sun is actually in the composition.
"05-Git2-DetailCrop1600-C1.jpg"
[Uploaded 2022-01-28]
I thought about uploading a different "detail crop" because this image is a bit under-exposed, and as such, is a bit grainy, but realistically, considering it was such a cheap camera, and currently is not even available in this form (it was replaced by the "Git2P" which is currently still being sold) it is actually quite good performance, so this is "good enough". I still have not found my test notes (from about five or six years ago), but I believe that I focused it as about 1M. and everything else is covered by depth of field and sharpening. This is roughly how most action cameras work.
Attachments
Last edited:
![[No title]](/data/xfmg/thumbnail/31/31977-2b717e032201241cbeae8226af23eba4.jpg?1619735136)
