Ask my guy to do some further work today, compare the different brands on the market for the dual lens camera.
Good deal Annke 180 degree dual 1/1.8 for $220?
- Thread starter S0daBs
- Start date
I have to say, I really like the IPC-Color4K-B180 in the turret format. With (2) 1/1.8 sensors it should make for a killer overview camera. Andy, please let everyone know when this camera goes on sale on the Amazon store (I am sure you will!!).
Thanks.
Still eagerly awaiting the new mini PTZ, too. Even if it is a 1/2.8 sensor.
Thanks.
Still eagerly awaiting the new mini PTZ, too. Even if it is a 1/2.8 sensor.
Yes coordinating it , hope can be ready at thanksgiving day sales, if not have to wait to early Dec
CCTVCam
Known around here
- Sep 25, 2017
- 2,830
- 3,708
Andy, as we're having suggestions, you just gave me an idea for a new camera.
Excuse my photoshopping. I'm cack handed in that direction!!
Idea is for a new Boobie Cam as a turret using a vertically stacked version of the 180 cam.
Speaker on the bottom, 2 cameras above it. The lens here wouldn't be designed to get 180 degrees but rather the bottom camera a 2.8mm pointing down and out so it captures packages etc. The Camera above facing straight out maybe another 2.8 or a 3.6mm to capture faces etc. Both cameras to use the 1/1.8" 4k sensors so you effectively have the same as on the 180 camera but in a narrower lens format designed to mind parcels and front door. Essentially it would be a Boobie cam clone but using turret format and the 1/1.8" 4k sensors.
Excuse my photoshopping. I'm cack handed in that direction!!
Idea is for a new Boobie Cam as a turret using a vertically stacked version of the 180 cam.
Speaker on the bottom, 2 cameras above it. The lens here wouldn't be designed to get 180 degrees but rather the bottom camera a 2.8mm pointing down and out so it captures packages etc. The Camera above facing straight out maybe another 2.8 or a 3.6mm to capture faces etc. Both cameras to use the 1/1.8" 4k sensors so you effectively have the same as on the 180 camera but in a narrower lens format designed to mind parcels and front door. Essentially it would be a Boobie cam clone but using turret format and the 1/1.8" 4k sensors.
I like the physical dimensions of the Dahua turret compared to the Hik/Annke.
But it has an unexplained problem. The video it produces is not nearly wide enough for the field of view, and it shows in the form of fairly severe aspect ratio distortion.
It is pretty obvious that this will be a problem just looking at the specs. The video's aspect ratio is barely any wider than a standard 16:9 camera, yet the field of view is about double the width.
Here are some samples from Andy that he shared a few pages ago. I've cropped a few pieces.
For a baseline, this is the "4K-T 1/1.2CMOS" (a standard 16:9 camera):
These are from the "4K 180" daylight pic:
See the problem? Things appear unnaturally tall. It is not just a display choice as is weirdly common in the CCTV industry. This is actually a problem in the source video. A little simple math done on the specs shows why this is happening.
Based on the specified field of view, 180° horizontal and 48° vertical, the height is about 27% of the width. (
The video resolution actually produced by the camera is not even close to matching this. The video resolution is 4096x1800, so the height of the image is about 44% of its width. (
27% is a long way from 44%, and that is why the distortion is so obvious. This camera should be producing a much wider image, given its field of view. I wonder if this is one of the "bugs" noted by @Wildcat_1 when he was preparing to review the Dahua 180° bullet? Anyway now that I've noticed it, I cannot un-see it.
Compare to the Annke/Hikvision dual-4MP cam. Its advertised FOV is 180° horizontal and 44° vertical, so the height is 24% of the width, and the video resolution is 5120x1440 so the height is 28% of the width. These numbers are much closer together, and this is why the Annke images shared earlier in this thread don't have a noticeable amount of stretching compared to the Dahua cam.
Some distortion is to be expected because they have to stitch together video from two different sensors and make it look good. But right now it is looking like Hikvision did a much better job of it. Hikvision's video is actually 32:9 which is exactly what you would expect from combining two 16:9 sensors. Dahua's aspect ratio is 20.5:9.
But it has an unexplained problem. The video it produces is not nearly wide enough for the field of view, and it shows in the form of fairly severe aspect ratio distortion.
It is pretty obvious that this will be a problem just looking at the specs. The video's aspect ratio is barely any wider than a standard 16:9 camera, yet the field of view is about double the width.
Here are some samples from Andy that he shared a few pages ago. I've cropped a few pieces.
For a baseline, this is the "4K-T 1/1.2CMOS" (a standard 16:9 camera):
These are from the "4K 180" daylight pic:
See the problem? Things appear unnaturally tall. It is not just a display choice as is weirdly common in the CCTV industry. This is actually a problem in the source video. A little simple math done on the specs shows why this is happening.
Based on the specified field of view, 180° horizontal and 48° vertical, the height is about 27% of the width. (
48 / 180 = 0.267 )The video resolution actually produced by the camera is not even close to matching this. The video resolution is 4096x1800, so the height of the image is about 44% of its width. (
1800 / 4096 = 0.439)27% is a long way from 44%, and that is why the distortion is so obvious. This camera should be producing a much wider image, given its field of view. I wonder if this is one of the "bugs" noted by @Wildcat_1 when he was preparing to review the Dahua 180° bullet? Anyway now that I've noticed it, I cannot un-see it.
Compare to the Annke/Hikvision dual-4MP cam. Its advertised FOV is 180° horizontal and 44° vertical, so the height is 24% of the width, and the video resolution is 5120x1440 so the height is 28% of the width. These numbers are much closer together, and this is why the Annke images shared earlier in this thread don't have a noticeable amount of stretching compared to the Dahua cam.
Some distortion is to be expected because they have to stitch together video from two different sensors and make it look good. But right now it is looking like Hikvision did a much better job of it. Hikvision's video is actually 32:9 which is exactly what you would expect from combining two 16:9 sensors. Dahua's aspect ratio is 20.5:9.
Last edited:
Just thinking out loud here, but it would be ideal if product discussions about Dahua / Empiretech dual cameras was in a separate discussion. Rather than here, in a Annke product discussion. Mostly because it will be easier to find the info if it's not buried here. Plus would return the OP's subject back to the Annke cam.
BTW, I definitely will be getting the EmpireTech 180 turret if the reviews are good. And CCTVcam's proposal would be ideal for my front porch if the form factor was low profile.
- Thomas
BTW, I definitely will be getting the EmpireTech 180 turret if the reviews are good. And CCTVcam's proposal would be ideal for my front porch if the form factor was low profile.
- Thomas
Here I have rescaled the snapshot Andy sent for the Dahua 4K 180° turret to show what it looks like at 32:9 aspect ratio. Top is the original. Bottom is rescaled to 32:9 aspect ratio.
32:9 might not actually be correct for this camera depending on how much of each frame the camera is cropping off during the image transformation. I think it looks a little bit too wide at 32:9 but it is hard to be sure. I think it is a lot closer to reality than the original, at least.
The proper way to know would be to photograph a square test pattern and measure it in the resulting snapshot to make sure it is square.
32:9 might not actually be correct for this camera depending on how much of each frame the camera is cropping off during the image transformation. I think it looks a little bit too wide at 32:9 but it is hard to be sure. I think it is a lot closer to reality than the original, at least.
The proper way to know would be to photograph a square test pattern and measure it in the resulting snapshot to make sure it is square.
I had to find the control to adjust the aspect ratio on the Hikvision to get it to look decent, Not having the Dahua to test with I would bet it has the same control also so you can get the correct aspect for the image size, Kind of had to hunt for it If I remember correct.
Interesting. That is such a weird set of resolutions. 2880 * 1264 is very nearly the same aspect ratio as 4096 * 1800 and 3840 * 1680. Only 3840 * 1080 is a 32:9 aspect ratio and it is substantially lower res than the Annke/Hikvision cam.
CCTVCam
Known around here
- Sep 25, 2017
- 2,830
- 3,708
Reaistically, IF and that's a big IF, Dahua chose to make one, it's unlikely in format to be anything other than a 180 cam turret with different lenses and the internal turret and internals rotated on their side. If Dahua could make it any smaller they would have done.
The only way I can see they could have done that other than interpolation is to have used a slighly larger sensor in pixel terms and not binned as many pixels. Dahua can establish that by finding out what chip is in it and whether they have an equivolent. That might explain why the Dahua is better at night. More pixels on same size chip = smaller = less light gathering.
Last edited:
I picked up a couple of these cams for two different systems that run Blue Iris. My CPU usage jumped about 25 percent on each system after adding the cameras. The mainstream streams at about 1100kps, about 10 times what my 5442s stream. I use direct-to-disk and the other recommended Blue Iris CPU optimizations. Anyone else had this issue or found a way to reduce the CPU usage on these cams?
Parley
Known around here
An interesting problem that I ran into was having the substream dropping out. It happened again yesterday with a camera I was moving, and this camera was on an NVR that has 7 cameras. It turned out that I had to put the camera on its own power supply instead of using the NVR POE power supply. Problem solved. So, if you are having substream problems look into separate power for the camera.
Correct, yes. This one. These are my first Hikvision OEMs but they appear to run the Hikvision firmware--i.e., remote access uses the Hik-Connect, not AnnkeVision platform.
The substream appears to be working fine in Blue Iris. I'm assuming it's the high mainstream rate. Or maybe, despite have direct-to-disc setup, it's not processing the substream for this particular camera.
CCTVCam
Known around here
- Sep 25, 2017
- 2,830
- 3,708
Are you sure that's not 11,000kbs? 1,100kbs woulld be very low.
As for data rate, don't forget the 5442 is a 4mp camera, the 4kX / t an 8mp. Double the pixels doesn't equal double the data. It does mean a substancial increase though. I believe most people use around 8,000kbs on the 5442 mainstream so at 11,000kbs woud represent a 50% increase which would seem reasonable and expected. I think some people even run the 4k x higher than that.
Last edited:
Depends. If he meant kilobytes per second, which is what Blue Iris shows in status displays, then that is equivalent to 8.8 Mbps. However he also said that is 10x higher than his regular 4MP cams, and 0.88 Mbps for a 4MP cam is concerningly low (bad choice of VBR quality I'm guessing).
It would be expected for one of these cams to consume 2x the CPU of a "5442" 4MP camera. Are you sure the sub stream is working? Check the Blue Iris status window. The "Sub FPS/key" column should have very similar values to the "FPS/key" column. And the Sub bitrate should also be showing a sane value.
Yep, both main and sub seem to be working fine. Mainstream averages about 1150kps and substream 200 kps. Still trying to figure out why the jump in CPU.