SJ7 V1.19 Firmware - Adds 4k Rotation - Report Thread

v1.20 has been released.

I just got it. Is there ever going to be real 16:9 4K @ 24p. I still don't get having 25p superview and 24p 2160 at 4:3 (is this even 4K? Why does it not just say 4K?). The 24p is generally unusable for professional film use. Can anyone tell me what the issue is with 16:9 24p 4K? That's really something that should've been in the initial frame rates. All the interpolated SJCams had 4K @ 24p. Rotation is working in fine, color is good and the screen is responsive again.

-Nate
 
Thanks. Guess I know what I'm loading tomorrow morning. [/grin]
 
If you read the official SJCam site, you've read my take on firmware V1.20. Cutting to the chase, it has one big problem and two other problems. The following videos, I think, show th0se problems.

The biggest problem is quite significant. In the frame, there is a zone that appears to be somewhat out of focus. It remains, relative to the edges of the frame, in about the same place. In the videos, it's found on or near the right large gauge (tachometer). As the gauges move around the shot, different places will seem fuzzy. The zone, as I said, remains in about the same place - to the right and below the center of the image.

The second problem is the new version of the EIS (Electronic Image Stabilization) is a step back from V1.15's EIS. Watch the trees and other objects in the distance, and watch how the upper edge of the windscreen moves.

Equipment information: Everything is shot at 1080p at 60 frames per second. The mount is a simple "chin mount" on my helmet using SJCam parts. I used an external battery, as I have in previous videos. The camera was in the "motorcycle" housing, the dive case with an adapter for external power. The mic is the internal mic and, of course, it's recording the sound that came through the case. I haven't altered the sound. The video was edited with Shotcut, but the only edits are setting a common starting and ending point for the "EIS on" runs and the one "EIS off" run. The V1.15 "EIS off" run was lost. More about that later. The video was shot midday on a partly-mostly cloudy day. All three videos were transcoded to 4K to limit the damage done by YT's compression. Transcoding to 4K does not add anything to what's in the original footage.

The first V1.20 video:

The second V1.20 video:

It's sometimes subtle, but the zone that's out of focus should be apparent in both videos. Look at the reflection in the tach or the numbers and the needle in the tach. I don't see a great deal of image stabilization with EIS 2.0 on. While I didn't fully duplicate the "EIS on" route (partly because left turns against the traffic is a pain, partly to keep the neighbors from getting excited about seeing someone race by every five minutes), there is a great deal that is the same. Using the "EIS on" route, the first 50 seconds and the curves starting at about 2:00, followed by the two white patches in the road, show the bike and camera moving around.

It's been suggested that there's a lens defect in the camera, or the lens is dirty. I cleaned the lens very carefully with my standard optical cleaner: half 91% pure isopropyl alcohol (never, ever rubbing alcohol, which has oils in it), half disstilled water, and a couple of drops of Dawn detergent for help with dirt and oils. I cleaned the case's lens port the same way. It's been suggested that somehow EIS itself was causing the "fuzzy spot" problem. Since the fuzzy spot is present with and without EIS, I don't see that as a cause.

After the second run, I swapped SD cards. Each card had either V1.20 or V1.15 firmware already on it. Swap SD's, load the appropriate firmware, check the settings, and it's time to shoot more video.

The final video, shot after the two V1.20 runs, was shot with fimware V1.15 and EIS on. There's no fuzzy spot in the images, and the overall image is far more stable. Compare the V1.15 video with the V1.20 video in, again, the opening 50 seconds and the sequence from about 2:00 forward.

The V1.15 video:

At this point, the big problem (image corrupted by the out of focus zone) and one lesser problem (poor EIS performance) have, I hope, been demonstrated.

The remaining problem is heat. My camera, shooting 1080p@60fps video for about 30 minutes, reached 135F on the front and about 130F on the sides. That's hot, hot enough to keep from holding the camera for any length of time. By comparison, using V1.15 in the same situation, the camera reached 125-130F. While that's hot, too, it's, for me, only uncomfortably hot. Any electrical device, whether it's a simple window fan or a digital camera, is going to produce heat while it's operating. The SJ7's processing does a lot of work, and it's happening in a small place. For that reason, a hot case isn't a surprise. But too hot to handle? That, I think, is too much. (Temperatures were taken with an IR thermometer gun)

The V1.15 "EIS off" run was lost when the camera firmware crashed. There is about 600Mb of data, but it's unreadable. Another result of the crash: the camera wouldn't shut down. When I tried to pull the battery, to turn off the camera, it was stuck to the point that I needed to use forceps to pull it out. The battery had swollen and was forced against the side of the battery well. Summing up: It's my opinion that the V1.20 firmware is a step backward, with a significant image flaw, poor EIS performance, and a heat issue. Again, in my opinion, until the image, stability, and heat issues are resolved, I recommend staying with V1.15.

ADDED: Anyone interested in seeing other V1.15 video, take a look at my channel. Any of the Alps videos and "Hawk Mountain Revisited" were shot with my SJ7 and V1.15. The rest of the vidoes were shot with another brand of camera, shooting 1080@30 fps. Subscribe or Like if you want to.

The bike in the US videos is an '03 BMW K1200RS, the Alps videos were taken on a BMW K1600GT.
 
Last edited:
The video in 1.20/1080p60fps/EIS off is numbah 10 for about 22 seconds and then very clear,why?
Additionally my Star's 1080p120 video has never been presentable as a hallmark of the SJ7!
I hope they can surpass themselves with the SJ8 that the rumor mill is calling the "The great" or "The greatest"
Are they alluding, tongue in cheek, to Cassious Clay?
 
Last edited by a moderator:
"You know what they say in sports, Bob: 'the tape don't lie'. OK, Bob, let's go to the tape!"

First, even if your internet connection isn't fast enough to get smooth playback, use the 4K view. The instruments give an excellent sense what's sharp and what's not. It's sharp enough to read the trip meter at the bottom of the speedo, the time of day on the digital clock, and the gear indicator. The speed markings (every 20 mph) are smaller than the tach numbers (every 1k rpm). Both covers are clear (not fogged or scratched). Let's roll the tape.

For the first few seconds everything looks normal. It easy to read "the clocks" and even see the engine redline at 8000 rpm, with 9, 10, and 11k in red. The tach starts to get fuzzy up to around 0:20 and that seems to go away. Jump ahead to 1:00 (slide the cursor at the bottom of the YT window).

Look across the frame at about the same row, in the frame, as the speedo and tach. The left miror is clear and shows a yellow ""road turns" sign and trees. The phone is running the SJCam app and most of the icons are at least recognizable. The gas gauge is looking good, and the speedo says I'm doing a bit over 40 mph, with 22 miles on the trip meter. The digital clock says it 12:33 and I'm in 5th gear. The temperature is below the middle mark, the right control pod is clear and the GPS model "Zumo" is readable. That's my right elbow in the mirror and a sign and trees in the background.

The top of the GPS is a little foggy, the top of tach is worse, and there's a big fuzzy blob above that. It's... The Fuzz Zone! Try this experiment: put a PostIt note next to the fuzzy area. I cut a notch in a PostIt and put the notch about where the fuzzy part is at 1:00. Then I went back to 0:00 and played the video. The fuzzy area will stay fairly close to the notch in the PostIt. As it should. The flaw's position is fixed relative to the edges of the image frame, not what's on the bike. Try the same thing with EIS on.

Let me know how it works out for you.

"And that's it for today, Bob! See ya next time!" LOL
 
"You know what they say in sports, Bob: 'the tape don't lie'. OK, Bob, let's go to the tape!"

First, even if your internet connection isn't fast enough to get smooth playback, use the 4K view. The instruments give an excellent sense what's sharp and what's not. It's sharp enough to read the trip meter at the bottom of the speedo, the time of day on the digital clock, and the gear indicator. The speed markings (every 20 mph) are smaller than the tach numbers (every 1k rpm). Both covers are clear (not fogged or scratched). Let's roll the tape.

For the first few seconds everything looks normal. It easy to read "the clocks" and even see the engine redline at 8000 rpm, with 9, 10, and 11k in red. The tach starts to get fuzzy up to around 0:20 and that seems to go away. Jump ahead to 1:00 (slide the cursor at the bottom of the YT window).

Look across the frame at about the same row, in the frame, as the speedo and tach. The left miror is clear and shows a yellow ""road turns" sign and trees. The phone is running the SJCam app and most of the icons are at least recognizable. The gas gauge is looking good, and the speedo says I'm doing a bit over 40 mph, with 22 miles on the trip meter. The digital clock says it 12:33 and I'm in 5th gear. The temperature is below the middle mark, the right control pod is clear and the GPS model "Zumo" is readable. That's my right elbow in the mirror and a sign and trees in the background.

The top of the GPS is a little foggy, the top of tach is worse, and there's a big fuzzy blob above that. It's... The Fuzz Zone! Try this experiment: put a PostIt note next to the fuzzy area. I cut a notch in a PostIt and put the notch about where the fuzzy part is at 1:00. Then I went back to 0:00 and played the video. The fuzzy area will stay fairly close to the notch in the PostIt. As it should. The flaw's position is fixed relative to the edges of the image frame, not what's on the bike. Try the same thing with EIS on.

Let me know how it works out for you.

"And that's it for today, Bob! See ya next time!" LOL
I think your fuzzyzone is actually a defective sensor or lens. What did our support say in your ticket? What is your support ticket number?
 
Um - I thought the comments would be passed forward to SJCam developers. I don't know how to include all of the comments and video links in a ticket. I suppose I could write "please refer to URL to see a write up of the problem. It's too hard to put in a simple ticket."

The V1.15 test was shot after the V1.20 tests. That shows the problem is not the sensor or lens. If they were the source of the problem, it would be in the V1.15 video, too. It's not present.

For further demos of my camera's and sensor, see any video on my channel starting with

Including this video, which YT doesn't include in my list of videos for some reason.

The other videos were shot with the Sena Prism shooting 1080p@30 fps and uploaded at 1080p. They show how bad the Prism's output is.
 
Refer to tickets 23619 (mainly video issues) and 23620 (overheating).
 
All the SJ7 needs is 4K/24p at 16:9 and a flat color profile to be on the level of the GoPro Hero 5 Black. This could be added in a firmware upgrade (I think the 4K/24p at 16:9 is coming soon). 4K/60p is interesting, but many people don't need it, but 4K/24p at 16:9 and a flat color profile would turn the SJ7 into the true GoPro killer it can be. They have the image quality down.

-Nate
 
Last edited:
24 fps is great from a tripod or drone, but from a moving vehicle, not so much. This is where 60 fps shines. Movement streaking is minimized. It may look sexy in a still frame, but in a video it's usually a distraction. I've very rarely shot 4K for this reason, plus 4K is a quick way to fill even a big chip in a hurry. 60 fps sure isn't going to help on that score. Until the Tb SD's come out... [/GRIN]
 
24 fps is great from a tripod or drone, but from a moving vehicle, not so much. This is where 60 fps shines. Movement streaking is minimized. It may look sexy in a still frame, but in a video it's usually a distraction. I've very rarely shot 4K for this reason, plus 4K is a quick way to fill even a big chip in a hurry. 60 fps sure isn't going to help on that score. Until the Tb SD's come out... [/GRIN]

But for professional use 24p is essential. This is what sets GoPro apart. There are professional functions that GoPro includes and I would like to see SJCam implement at this price point, which would attract indy filmmakers and minimalist filmmakers. Even the low light on the SJ7 is better than the SJ5000x so this is the issue I have. There's a whole audience of people out there would be willing to, buy the SJ7 (or two) over there GoPro, for these 2 features. It's literally 2 features that can be added in a firmware upgrade.

-Nate
 
The SJ8, at least on paper, is probably closer to what you're looking for. I think the SJ7 is probably closer to being an action cam in the full sense of the term. It's not that it can't produce good imagery so much as I don't think it was designed to be an indie minimalist production camera. Obviously, if that's all someone has, it's no reason to give up. But if I were gearing up, had a $500 budget, and time to wait for the release, I'd look closely at the SJ8 as the core of my gear.

As I said offline to someone, "beware of the 500 lb Swiss Army knife". "Just a couple of features" here and "just a couple of features" there and pretty soon it's the 500 lb Swiss Army knife, loaded with every possible feature and too big to be useful for anyone. IMNSHO that's exactly what's bitten V1.20. Lots of nifty features but sucky images.
 
As I said offline to someone, "beware of the 500 lb Swiss Army knife". "Just a couple of features" here and "just a couple of features" there and pretty soon it's the 500 lb Swiss Army knife, loaded with every possible feature and too big to be useful for anyone. IMNSHO that's exactly what's bitten V1.20. Lots of nifty features but sucky images.

Thing about it is some of the "nifty features" could be done away with. A few major frame rates, (including 4K @ 24p) a flat color profile and the basic action camera stuff would make people sit up and take notice. The YI Mijia Mini offer F-LOG. It has only some basic features and a tripod hole in the bottom. Their major feature is the 4K and now the F-Log that was added in a firmware update. I think the SJ7 is a better camera in terms of quality, but imagine being able to buy two SJ7 Stars at the price of one Gopro Hero 5 black and getting 4K @ 24p and a flat color profile. You're looking at the true GoPro Killer! Fact is GoPro has features people like me (and many others) don't need or use. But those two pro features are extremely useful.

Also I don't get the lack of 24p on many major action cameras. They have 24p in their interpolated cameras, so they know people want it, but they throw 24p out when they make a real 4K camera. Hawkeye, Thieye, and others had 24p interpolated cameras, but not with their real 4K cameras. When the YI Mijia Mini added F-LOG, there's still no 24p. Why offer a pro feature minus the frame rate? A basic real UHD 4K action camera can do with 4K/30p/24p/25p, 2K/60p/30p/24p 1080/120p/60p/30p/24p and finally 720/240p/120p/60p/30p and frankly most people are only using 720p for high frame rates. Anything beyond this is extra. If you need 4:3 it should be 4:3/30p and 1080p at 30p and doesn't have to be added for frame rate. You can crop 4:3 in post if needed and still get a quality image. The 24p frame rate is a major feature.

As for the SJ8, I'm more impressed with the look of the camera and the 4K/60p, but honestly I don't need 4K/60p nor voice command and such. Those are cool, but what I think users will be more interested in is quality footage/images, the ability to customize the look of the image and things like that. 4K/60 is great, but 4k/60p and no 4k/24p would be a head scratcher.

-Nate
 
Having been involved with other cameras development and early testing, i was amazed to see that fixing some problems in firmware would make other pop up, and this could easy be things not really related to what ever was fixed.
So firmware development are not as easy as i thought it was it seem.

The professional features like a flat color profile i am not so sure, okay it sound like a easy thing to implement in the firmware, but having seen what i have i am not so sure.

I think the focus are on the general users, and they are just shooting from the hip and throwing everything on the web, and its most likely the same that cant even figure out to work their phone camera properly.
And sadly those people just ruin everything for us sane people, but they are the wast majority so it make sense to cater to them.
 
@Nathyn Masters I think we're agreed that "the image is the thing". We might (or not) completely agree about how to get that image. But that's what makes for good discussion, eh?

@kamkar1 Most users don't care about most of the features. As long as they can shoot their bit of Now! video, life is good. Fretting about ISO, EV, white balance, and even firmware changes isn't important in their world. There's a markets for dash cams, so that's supported. Some people want a surveillance camera, so that's supported. Stills? It's in there. And on and on. And soon, there's the 500 lb. Swiss Army knife.

The more complex a thing is, the harder it is to test everything, and find unexpected interactions. And add the multiple vendors (sensor, signal processor, etc.) and the chances of a Problem increase exponentially. I'm willing to bet there are bugs in V1.20 (and 1.19, 1.18, 1.15...) that don't show up often, but they're there. And who knows how they got there.

However, what annoys me is that nobody at SJCam seems to have pointed the camera at a resolution grid to verify the basic image is sound. Ditto for testing EIS. Heck, put it on a table, point the camera at the grid, and move the table. What does it take to make the grid dance in the frame? Maybe all of this happened. But my three videos suggest otherwise.

BTW, I really, really hope someone else will attempt the same comparison of firmware rev.s. It's certainly possible my camera has problems. I think I've set up a situation that either shows camera flaws or shows they're not present. A couple of other cameras shooting out a car window, from a drone, or something similar should help answer the question: Is V1.20 flawed?
 
Indeed there are many aspects of the gopro 6 and other existing and coming cameras i would most likely never use myself. ( i call such features facetwitt features, cuz to me at least its things i can easy do without, and they are not necessary free as some people think those social media things are )
And i often wish i would be able to get that product cheaper without those features, my car have electric windows in the front but not the back, i would have liked to save a few bucks and gotten a wind lever on the front doors too.

It is damn hard i think to do some testing, and not least older Vs newer firmware without having 2 cameras.

EIS you could test if you could build a little shake table you could run at different levels of movement, but it would be cheaper to just get 2 cameras and test them at the same time using the same settings or not and same or not firmwares.

For me IQ are the most important thing for the little cameras, and i still think that EIS are a good feature for the uses where a gimbal are out of the question.
 
Nice as the "pick and chose" option would be, I can't see that happening in cameras. And I can see it raising, not lowering prices. Getting the firmware to deal with a changeable feature environment adds to the cost. And that has to be amortized over the entire product - at the least. The easiest solution is to ignore what you don't use. Or wonder why something like adding the SJCam watermark option even exists in the feature menu. [/head scratching]

Testing... I guess that comes down to a management choice. It adds to the R&D costs. It affects the product's reputation. Which costs less? Or maybe the intent is "get it out the door; we'll worry about the details later". Who knows?

I have no idea about how many SJ7's were made (I assume a production run and not continuous production). Shoot 'em out the door, hope nothing goes wrong, and either repair under warranty or "sorry, but you're out of the warranty period". Ditto for firmware. Add "invert image" or "add new fps", see if that part works, and out the door - we'll fix it later. And here we are, and that, IMHO, is how we got where we are.

About EIS, I think I've been clear that lack of a working EIS is deal-breaker or close to it. "The image is everything" - a bouncing image, unless specifically looked for, is a bad image. Look at my earlier videos, shot with a Sena Prism. The EIS is minimal and the video suffers for it. OTOH, if the idea is to convey the idea of a hand carried camera to make something seem "this is as real as it gets", well fine, no EIS is the way to go.
 
Back
Top