Camera Resolution?

[BobW55]

Ok we all know the CGO3+ is not the toughted 100MBS, but is actually closer to 50-60 MBS.
so I have 2 questions.

1) "IF" the camera did operate at 100MBS, would the difference be noticeable to the untrained eye?
2) "IF" we reduced the FOV to something closer to 80-85 degrees would we have better resolution?

To me, reducing the FOV should help with quality due to the simple fact the camera has less to process, but I am no expert in this area. I do know if you go too narrow, it will look like you are in a tunnel. If you go too wide, it looks like a fish eye.

PLEASE DO NOT post with whining about the camera being less than advertised, we all know it is and have beat the subject to death on other threads.
PLEASE DO NOT post about what replacement lens is best. This is more for a technical discussion on what the current camera CAN do, and what advantage/Disadvantage there is to changing the FOV.
Thanks

 

[Steve Carr]

It is my understanding the 100mbs is the max speed of the processor. Most recordings of video at max resolution are around 40-60 mbs. At night that number drops considerably. The reason is because the processor only records changes from one frame to the next. If most of the frame is black then there is no change for most of the image.
The FOV has nothing to do with resolution. A 12 Megapixel image at 80 degrees or 120 degrees is still 12 Megapixels. I'm just a hack so I'm sure someone will provide better information.

 

[DerStig]

1) yes. EVERYONE would notice it

2) no has nothing to do with the physical resolution of the imager.

 

[THoff]

1) Few people would notice an increase in quality at a higher bitrate, and then only under specific circumstances such as when panning rapidly where much of the scene changes. In order for this to be visible, you'd have to freeze or slow down the video or the detail becomes hidden in motion blur, however. DerStig is telling you otherwise, but my point is easy to prove by taking high-quality source video typical of your use and encoding it using AVC/H.254/MPEG4 at 100 Mbps and 50 Mbps VBR and comparing the output. As bitrates go up further, increases in quality become harder and harder to notice -- doubling the bitrate again to 200 Mbps VBR will result in an image that will be indistinguishable unless you are pixel-peeping on a frozen image.

2) No.

 

[McLongShaft]

For point number 1: I don't think that the untrained eye would notice much difference. With a higher bit rate you'd get more information which would help with post processing, but it's like the old saying (well a saying that I've just made up), 100Mb/s vs 50Mb/s still looks the same when shot through a potato - the potato being the camera in this case as its not the best. That being said I'm pleased with the videos I've taken with the H and I don't find it as bad as some have said.

Point 2: I think I see what you mean, if you reduce the viewing angle you're capturing a smaller area, which means that those megapixels are being used to capture a smaller scene. Which in effect would mean that you would have a higher quality/better ratio or pixels to area. You would still have the same megapixel count but over a smaller area. So, we wouldn't have a better resolution but we'd have a better image at the end of the day.

Sorry if that was a bit of a disorganised statement, I hope you get my jist.

 

[BobW55]

Mclong,
On point #2,
That is where my head is with this. If I could reduce the FOV, the sensor had less data to mess with an should give a bit better image. I am sure at some point, you could only get the FOV to improve things before optical distortion comes into play. I also think that less FOV, less getting the props in scenes and less glow from the LEDs at night.
I also think that with less FOV, the image would look better when digitally zoomed in on.
Wish I had a handful of lenses to play with.

 

[Rayray]

I'm happy with the CGO3+ resolution and picture quality, not knowing enough like DS to find fault. There actually are folks whose computers, etc. won't play 4K videos at 50 let alone 100 (I have one that will and 2 that won't). Most cards max at 70-90.

I do think I'd like a 90 degree lens better, mine appears to be 115, but the grass is always greener...

 

[McLongShaft]

I'm completely agree Bob. If we imagine that with the standard optics 1 pixel takes up 2cm of ground space or of an object then by that same means, a narrower FOV would produce something like 1 pixel to 1.5cm. Of course the number in this aren't correct, but the principal would stand true.

I may look into changing to a narrower FOV.

 

[DerStig]

You will notice a higher bitrate as a sharper image with less motion artifacts and you will notice better color rendition and better dynamic range (more held details in the shadows and highlights)

Re-encoding something as Thoff claims won't show you the difference once compressed always compressed.

 

[THoff]

You will notice a higher bitrate as a sharper image with less motion artifacts and you will notice better color rendition and better dynamic range (more held details in the shadows and highlights)

Maybe with the cameras you use day-to-day, but we are talking about 4:2:0 video here that is blotchy by nature. A higher bitrate isn't going to bring back detail that is thrown away by the chroma sampling.

Re-encoding something as Thoff claims won't show you the difference once compressed always compressed.

I'm not talking about taking CGO3+ footage and re-encoding it at a higher bitrate. There is plenty of 4K 4:4:4 sample footage that can be found on the Internet -- use that as a source.

 

[DerStig]

Maybe with the cameras you use day-to-day, but we are talking about 4:2:0 video here that is blotchy by nature. A higher bitrate isn't going to bring back detail that is thrown away by the chroma sampling.
I'm not talking about taking CGO3+ footage and re-encoding it at a higher bitrate. There is plenty of 4K 4:4:4 sample footage that can be found on the Internet -- use that as a source.

You need to look at gopro footage, it's not nearly as over compressed as the CGO3+ 4:2:0 isn't "Blotchy by nature" it can look perfectly fine. Re compressing something you find on the net won't show you what the visual difference is between 50 and 100mbps simply because its already been compressed

 

[THoff]

You need to look at gopro footage, it's not nearly as over compressed as the CGO3+ 4:2:0 isn't "Blotchy by nature" it can look perfectly fine.

The GoPro uses a three times shorter GOP at a bitrate not much higher than the GGO3+ -- that actually makes it more compressed because it has more independent frames. If you had only independent frames at the same bitrate, you'd have the most heavily compressed image.

Yes, 4:2:0 is blotchy. Try using it for green-screen work and see if you can get a good key from it.

Re compressing something you find on the net won't show you what the visual difference is between 50 and 100mbps simply because its already been compressed

There is plenty of 4K 4:4:4 RAW footage on the net for downloading.

 

[banned user]

The GoPro uses a three times shorter GOP at a bitrate not much higher than the GGO3+ -- that actually makes it more compressed because it has more independent frames. If you had only independent frames at the same bitrate, you'd have the most heavily compressed image.

Yes, 4:2:0 is blotchy. Try using it for green-screen work and see if you can get a good key from it.
There is plenty of 4K 4:4:4 RAW footage on the net for downloading.

All these conversations I've read here, is it fair to say that if the Yuneec codec was altered to have a shorter GOP, that we would have buttery smooth video similar to the Go-Pro?

 

[THoff]

A shorter GOP requires a higher bitrate because it has proportionally more independent frames (which encode the entire image rather than the differences from the previous frame, at the cost of needing more storage). You won't get better video just by shortening the GOP, you also need to increase the bitrate.

 

[banned user]

A shorter GOP requires a higher bitrate because it has proportionally more independent frames (which encode the entire image rather than the differences from the previous frame, at the cost of needing more storage). You won't get better video just by shortening the GOP, you also need to increase the bitrate.

Ah ha, so a shorter GOP does not result in a higher bit rate from more GOP in the same amount of time of video. I was hoping that if it was only a simple one line of code to be altered that a case could be put to Yuneec to make a change.

 

[THoff]

No, the bitrate is independent of the GOP. If you are asking the codec to put more independent frames into the same amount of space, then the partial frames will have to throw away more information and the image will suffer.

One way to battle bitrate constraints is to use a longer GOP. The downside there is that any fidelity issues are cumulative with each partial frame until you get to the end of the GOP and have another independent frame.

There is no such thing as a free lunch.

 

[banned user]

No, the bitrate is independent of the GOP. If you are asking the codec to put more independent frames into the same amount of space, then the partial frames will have to throw away more information and the image will suffer.

One way to battle bitrate constraints is to use a longer GOP. The downside there is that any fidelity issues are cumulative with each partial frame until you get to the end of the GOP and have another independent frame.

There is no such thing as a free lunch.

Ah thanks for explaining, I'm just trying to understand if there is a simple solution to suggest to Yuneec, would raising the 50Mbs max limit on the codec then allow for more and shorter GOP's to be captured and therefore improving the smoothness of the recorded video? Is it that simple?

 

[BobW55]

I wish we could upgrade the sensor/processor. How ever from what I have read here, and what I can Google, going from 50 to 100 MB most human eyes will not see a big diiference, unless you get pixel picky.
So to my original post all we have left to improve the current camera is a better lens and maybe some reduction in FOV.

Unless of course Yuneec comes out with a better camera all together.

 

[THoff]

I doubt the codec parameters are set in stone, they are likely set by the camera firmware during initialization. Either Yuneec or a determined enthusiast should be able to change the bitrate and GOP length.

 

[banned user]

I wish we could upgrade the sensor/processor. How ever from what I have read here, and what I can Google, going from 50 to 100 MB most human eyes will not see a big diiference, unless you get pixel picky.
So to my original post all we have left to improve the current camera is a better lens and maybe some reduction in FOV.

Unless of course Yuneec comes out with a better camera all together.

If you side by side compare a go-pro 4 black 4k video beside the CGO3P 4k video there is a very noticeable difference in smoothness of video the go-pro seems to have much more information in any given second of video, but the two cameras are built form basically the same parts.
I'm wondering if Yuneec were able to adjust the codec settings, would we be able to achieve smoother video.
I wonder if too if it would ever be possible to flash these processor's ourselves.
The S2 processor specs say that it has a "Flexible GOP configuration" and...as well as a lot of other stuff I don't know about. http://www.ambarella.com/uploads/docs/S2 IPCam Product Brief.pdf