I have 6 Wyze Cam v3s and about a dozen Wyze bulbs (half color and half white). I am probably going to get rid of all of it, terminate my Wyze Cam Plus subscription and start over with a different ecosystem.
Primary reason is the removal of RTSP support. This was a buying decision for me. I understand that 24x7 recording to the cloud isn’t feasible, but the Wyze cams miss enough events to make it necessary to have a secondary continuous source of recording. Recently I had a theft and the Wyze Cam Plus recording ended before the actual theft occurred. Without having my RTSP backup, there wouldn’t have been evidence to submit.
Is there really no planned support for bringing RTSP back? This is a DEAL BREAKER when even with Wyze Cam Plus my recordings end before capturing an actual crime occurring.
Cam+ recording is limited to about 25 seconds. If you have an SD card installed, it would capture the entire event. You can see this if you press the “Playback” button at the bottom of the display.
RTSP is an industry-standard video streaming protocol that allows video equipment from manufacturers to work together. Wyze doesn’t want to implement it. Use any web search engine for more details.
No, there is no limit. It will record 24/7 if there is sufficient motion in view. The only limitation is 5 minutes per event with no cooldown, which means that if motion is non-stop for 17 minutes, it would break up the 17 minute event into 4 videos. 3 that are 5 minutes long and the last one would be 2 minutes long, but all 17 minutes would be recorded. I have a camera watching my pool and if it’s windy, I have cameras that will record constantly for several hours straight, especially at night where it’s constantly reflecting the light on the ripples of the water.
That is the intent of the MicroSD card
I really wish Wyze would do more to offer RTSP. I don’t care too much about saving video with it (I’ve found cloud and continuous SD card recordings to be reasonably sufficient for that for my needs), but more so I can easily use the streams for automations and custom projects. Right now we have to rely on 3rd party solutions for this. I think the most popular of these include Docker Wyze Bridge, Scrypted, and running Tiny Cam Pro as a server. Those will all turn most or all off the camera streams into RTSP without needing different firmware. That’s what I do currently. I wish Wyze would start making it standard on future cameras, at least their pro line which already often has local AI and stuff for people who don’t want a subscription and are willing to pay a little more.
I don’t use the SD card in events mode because it might miss something too. I use cloud for events and leave the SD card in continuous recording so nothing can ever get missed.
Having said that, the SD card records in 1 min intervals. If there is motion starting at 1:01:59 and ending at 1:02:01, it will record 2 whole minutes even though the motion/event was only 3 seconds long. It will back record everything back to 01:01:00 for the first 59 seconds where nothing happened, and it will record everything all the way up to 01;02:59 even though nothing happened for last 58 seconds. So, a 3 second event turns into 2 minutes.
The Cloud events won’t do that though. Some newer cams will pre-roll a few seconds, and all cams will usually post-record a few seconds after motion ends, but they don’t record in full minute intervals like the SD card.
This is normal for cloud events and SD card events to be different lengths of time. They’re setup differently.
Maybe I misunderstand then. Can you provide a video example? The criteria should be:
Cloud events will record as long as there is motion (motion is determined by the sensitivity level which requires a minimum number of pixels changing at the same time) and often a few seconds afterward. (A couple of newer cams will preroll a few seconds before adequate motion is detected). These events don’t have to be in 1 minute increments. If motion stops for a few seconds, it will end the event then, and if motion starts back up 10 seconds later, it will start a new event. This may happen if a person holds still for a few seconds, then starts to move again.
Note that just because we may visually see what we think is adequate “motion” does not necessarily mean that number of pixels were changing adequately between frames.
Note that I am speaking of Cam Plus specifically. There are nuances to the other options such as Cam Plus Lite which may have a cooldown taken into account.
SD card events are determined by the same motion standards, except that will post in 1 minute segments, even if 98% of that minute had no motion in it, it will still record the entire minute to the SD card.
So, in the example I mentioned in #1 about how cloud recording works, let’s say there is a little bit of motion at 1:01:25-1:01:40 then no motion, then motion starts again 10 seconds later (1:01:50) and goes for 2 minutes (1:03:50), then stops for 5 seconds (1:03:55) then goes again for 20 seconds(1:04:15), ending at 1:04:15. Here is how those would be recorded:
Cloud
It would record a 15-second event, then a 2-minute event then another 20-second event
SD Card:
It would record 4 full minutes straight from 1:01:00 - 1:04:59.
So cloud would record 2 min and 35 seconds total split up in 3 events, and the SD card will record 4 full minutes straight because there was motion during each minute and it only asks “Was there ANY motion during this minute at all? If yes, record the entire minute onto the SD card.”
But what counts as motion itself is the same for both the cloud events and the SD card, the only difference is how long they record for (pre-roll and post-roll recording when there is any detected motion.
So, let’s say that someone walks away from the camera and it detects motion at first, but then the person gets to a certain distance and is no longer causing a sufficient number of pixels to change…the cloud may stop recording them walking away, while the SD card continues to record them walking away for a whole minute longer. That doesn’t mean the camera detected that continued motion, it just means the SD card continues to record to fill the whole minute sector. If a tree branch or something else moves to cause motion, then it may even extend recording longer. If a person has cloud events toggeled to “Smart detection” then the SD card will show more events than the Cloud events tab because the “Smart Detection” toggle means that the server deletes ALL events that don’t have a detected AI object in them. In this case, a waving plant or shadows or lights or whatever else might cause the SD card to continue recording the person walking away indefinitely even though they aren’t actually changing sufficient pixels to count as motion anymore, and the Cloud events won’t show anything in new event recordings because the person is not using enough pixels in the view to be detected as a person, and since there is no AI object in the event, the server wouldn’t save the event video because said person selected to have it ONLY save “Smart Detections” and not motion recordings. But the SD card still saves all motion.
The above is how the cameras are set up to function. If I misunderstand you, or if you have a different example, then please explain or share an example. If you post a video example, some of us can probably help explain what is going on in your particular case.
The same event was recorded by both the cloud and the SD card. The SD recording is longer. Why is that? If the SD card recording is supposed to stop when there’s no more motion. The inference is that the cloud determined that the motion stopped but the camera was still detecting motion. Why the discrepancy?
To be clear, the moving objects were a person and a car. Both were being tracked.
That is not correct. When the camera records to the uSD card, the recording will ALWAYS be one minute long if there is detected motion at any point within that minute. That one minute recording will ALWAYS be from the top of the minute until the top of the next minute. It makes no difference if that detected motion was 1 second long or the entire minute or anything in between.
That is not what it is supposed to do. (I explain this above and will explain it below in this comment too)
It was not still detecting motion.
Cloud recordings only upload motion (with a few seconds leeway). The SD card records all 1 minute segments that contained ANY motion within them. If motion is 1 second long, it will record the whole minute. If motion is 2 seconds long, but 1 second started in 1 minute and ended in the other minute, then the SD card records for 2 minutes even though motion is only 2 seconds long.
Personally, I would mostly suggest you verify it with testing it on your cameras and see the proof yourself. Documentation can change or be outdated (in this it isn’t, but in general, I find when people are skeptical of an explanation, they typically only ever believe it when they have tested and verified it themselves. I recommend pulling out an SD card, put it in a computer and see that every single file is 1 minute long (with the possible of exception of a partial length file from something like when the camera first turns on and doesn’t have a full minute to record).
But since you asked courteously: besides employees saying this in the forums &/or AMA’s in the past, including Wyze Devs telling me this is exactly how they set it up to work, and me testing it myself years ago (2020) in a small room without any other accidental motion, and pulling SD card files out to analyze them and verifying it for myself…here is some documentation for you:
This is the way it currently works on all Wyze cameras with an SD card set to “Events Only” recording.
I hope that helps. As I said, it’s really easy to verify yourself so you don’t have to take anyone’s word for it. That’s the nice thing about it. There is no need to take my word for it. Countless people have tested and proved it themselves, including me doing it for myself to verify it actually works the way they said it does.
I know it’s OT but I’ll use this space to say that 1-minute design isn’t something I’d recommend. Any developer can tell you; opening and closing a file is a very expensive proposition. so much that in some system designs, developers hold that opened file structure for re-use by different cooperating processes.
If you have a 3-minute video split into 3, you just multiplied your overhead by 3. I wonder if all the slowdown in SD card access reported in other threads is somehow related to this design.
And yes, based on this explanation, the criteria for determining no motion is different. Whether they did that for a specific advantage, I don’t know. I hope it isn’t because coding is easier.
While the file ending protocol may differ, I think “Motion” or “No Motion” is technically the same. The way Wyze has explained it is that the cameras are programmed to keep 1 minute of video in the RAM. A separate process determines when/if there is motion. So for the SD card, the Motion subroutine will mark when there is motion, and then if there was motion during that period, it just transfers that whole 1 minute segment to the SD card. It is not necessarily saying that motion continued until the end of the 1 minute segment, but that the motion subroutine indicated there was motion at some point within that stretch of 1 minute video that was held in the RAM.
The same motion subroutine also tells the camera when to start uploading video to the cloud and stop uploading the video. So the motion itself is still the same on both, but they are both programmed differently about when it is proper to close the file. Cloud will close a file early because there are ongoing costs involved to store it, so if they were needlessly storing up to an extra minute at the start and end of each motion event, that would be a lot of waste, and time that most users wouldn’t want (who wants to sit for up to a whole minute wondering when the motion starts, for example, or stay an extra minute wondering if something happens again at the end?).
Regardless of the minutia involved and the semantical difference of whether it’s motion or file-ending protocols that are different, it amounts to the same thing, in that the cloud and the SD card record different video lengths and the disparity can be confusing to some people as is demonstrated in this thread where it may be confused that the SD card recordings are more sensitive to motion, when it’s just the way the file ending protocol is handled. So your point of them functioning differently in the end is still valid.
TL;DR: Within the bounds of economic and hardware resources, real-time processing of continuous live data stream takes precedence over efficiency of historical data storage.
I understand your frustration because every time I want to see a particular thing exactly that activity doesn’t record. Plus, some of my windows are fogged up all the time and I can’t see anything out of those cameras. Sometimes also I get recordings from the day before or two days before from the ones that didn’t record on the right day. It’s totally messed up and I’m getting very frustrated with the whole system.
Yep. When Wyze stopped using RTSP then started “urging” Wyze Cam Plus, that’s when i started having second thoughts. It’s all about the money it seems. Not quality. The older cams without Cam Plus are what i’d keep. I’m gonna start looking for something like Wyze used to be.