Are there any plans to document the client CPU/GPU usage between the different fisheye cameras your testing?
Axis 12MP Stereographic Camera Tested (M3058-LVE)
This model only supports H.264. Do you know if Axis plans to support H.265 on this model?
Would H.265 be a significant improvement in bandwidth/latency/etc over Zipstream?
John, Have you all tested the video clarity between H.265 and H.264 at the same bitrate? H.265 marketing often misses the point that H.265 offers superior clarity than H.264, all else equal. I've done my own tests and notice quite a difference in reduction of smearing and pixelation especially when digitally zooming in with 4K resolution. Like you said, when you can add intelligent codecs and change i-frame and FPS settings dynamically, H.265 may not be needed to lower bandwidth but may be important in offering better evidence.
Have you all tested the video clarity between H.265 and H.264 at the same bitrate? H.265 marketing often misses the point that H.265 offers superior clarity than H.264, all else equal. I've done my own tests and notice quite a difference in reduction of smearing and pixelation especially when digitally zooming in with 4K resolution.
Can you share your tests (images, settings, etc.)?
At the same bitrate and same resolution/fps/etc., H.265 is typically using lower compression levels than H.265. To that end, there might be some visual benefits but it would depend on what quantization levels were being used, at what bit rates, etc. One of the things we have found over the years of testing is that there is a diminishing marginal benefit with lower compression levels in terms of improved image quality but this depends if you are starting with a quantization level of 10 or 30 or 40, etc. If you can share your test details, we can review what you have done and provide feedback.
Make a comparison between the 3 tested models of 12 mp.
Axis, Vivotek and Avigilon.
That's coming in a couple of weeks. We have some models still on order, but this series will end with an 8+ camera shootout, with more image quality testing in more varied scenes.
Do you think this is OEMed from Brickstream (Flir)? http://www.brickstream.com/
#1, does it even look like anything from Brickstream?
Outside of Brickstream offering stereo vision cameras, what connection do you see between the two?
What Milestone VMS and Smart Client versions was used in this test with Axis dewarping plugin?
Is the chart supposed to be in Mb/s or Kb/s? 3,200 Mbs? Just to nitpick....
As to the ImmerVision Dewarping. ExacQ supports a lot of fisheyes with this plug in, but it never works as well as the native. We had a lot of problems with the dewarping of the Axis M3047 with ExacQ and the ImmerVision codec. Once they updated the software version to support it, it worked leaps and bounds better.
Ethan, did you by chance open a ticket with Nx about the lack of a sub stream? Is it not possible to setup a sub stream of the warped view, say around VGA resolution, in the camera web interface?
We didn't open a ticket, but I'll ask them about it now. It's a limitation of the camera, I believe: the substreams are all used for dewarped views and can't be reconfigured to other types. I'll confirm that it can't be done in the 30 FPS mode, as well, which would be more applicable to Nx, since they do their own client side dewarping.
Would like to see vs Panasonic WV-X4171, please compare.
How high is this camera (and all comparable fisheyes) mounted for testing?
Thanks... I'm think of these cams being used in an environment where they are ceiling hung at about 20'. The Axis camera calculator doesn't even have this camera yet. How do I determine the viewable square footage it would cover?
To calculate camera coverage you simply draw a circle around the camera. It will cover from wall to wall since it's 181ºx181º viewing angle. Once you have the actual coverage, measure the square footage and divide pixel by that, you'll get your pixel density (more or less, since it has more pixels on the edges than in the center due to the panomorph lens)
It will see everything below the cameras mounting height, out to infinity. Now, as you get further from the camera, the lower the pixel density will be. Most fisheye type cameras will cover about a 20’ diameter with reasonable quality. Once you get beyond 10’ away, quality drops very fast. So, mounting it 20’ above your subjects would be a bad idea. You will need to use a pole drop about 10’, if that is possible.
If you aren’t able to drop it 10’ and mounting at 20’ is your only option, I would recommend a non-fisheye camera instead.
As noted above, the sensor is a 12MP sensor but the maximum resolution is 8.9MP. This is based on a rectangular chip but essential a square (circle inside) image. Are all 12MP sensors rectangular and therefore all 360 cameras marketed as 12MP about 8.9MP usable resolution? Are any 12MP sensors a square chip, allowing use of all the pixels?
Are any 12MP sensors a square chip, allowing use of all the pixels?
In a 180/360 degeee camera? No.
Even if the sensor was perfectly square, the best a single round lens could cover while maintaining 180 degree coverage would be 78.5% of the square area.
The Axis utilizes 74%, which is not bad considering.
The Bosch that Roger references above does utilize 100%, however it a box camera with a rectangular FOV.
I'm sorry that I wasn't more clear. I'm specifically speaking about 360 degree cameras which use a round fisheye lens. For a traditional lens, I don't see an issue. This question is meant specifically for 360 cameras. If I recall correctly a small group of 360 cameras use an oblong lens (I think Sony) but most use a round lens.
Thanks for the response!