Hello Fellow Yuneec Pilot!
Join our free Yuneec community and remove this annoying banner!
Sign up

Typhoon H Camera 100 mbps?

Nicely done, I'm pleased to see you fly nice and low, so many people shoot their videos as if they're map making and seldom have a dominant element in the shot. Color looked pretty good too.
 
I have recently started editing with Premiere Pro CC and just starting to get to grips with it. I agree that you will need to sharpen in RAW format, as you do with stills, out of interest which sharpen technique is your preferred. In Photoshop I tend to use High Pass. In Premiere I have been using Unsharp mask?
Hey there, Unsharp mask is definitely the way to go when you want to sharpen something in Premiere Pro CC. That being said, I actually use the regular Sharpen tool instead only because Unsharp mask really increases my render time. See I also use a Denoiser plug in from Red Giant, and when I combine this with Unsharp mask the render times are kind of crazy. I once had a 4 minute video take close to 18 hours to render, and that’s with a good computer and fast graphics card. I heard though that the new Premiere Pro update helped to shorten render times when using Unsharp mask, so I’ll have to test that out sometime soon.
 
If you have good footage in raw it should look bad. A flat profile is what you want and that should look bleached and washed out. It is supposed to as it is capturing as much dynamic range as possible. This does not look good in its raw form so to speak.

Good footage always requires post processing. This is how you edit video.

If you get great footage out of the camera it is being artificially sharpened and color corrected in camera.

To me, this is not good. I want it to look bleached out.
Take that bleached out footage in Premiere, color correct it and it should be a big difference. That extra bleached out data gives you more room to work in post, especially in shadows etc...


EXACTLY my friend! max latitud/dinamic range.....The pro-side of filmmaking we allways try to stay in.

Of course, taking in mind that here we have (at the moment)
shooting 4k RAW (24fps) - (4096x2160):
H264
MPEG-4 AVC (avc1)
4:2:0 YUV 1:1
 
Last edited:
Only the highest bit rate I've seen out if the camera is 50Mbps


100 MB !!! may be its hidden arround when shooting full HD 1080p120, BUT ONLY if the final resulting GOP its in 24fps. (who knows?)
We must investigate it.
But if this is achieved (100MB over 4:2:0 color sampling in 4k in this little camera.....then could be a revolutionary mojon)
 
Last edited:
I think people are mixing up MB and Mbps. DJI uses Mbps and states the Phantom 4 is 60 Mbps. Yuneec uses MB and states the Typhoon is 100 MB, which is actually more like 80 Mbps (if my math is correct). The noticeable difference should be in shots like panning. If you pan up too quickly with the Phantom 4, you may get small artifacts beginning to appear in the video. You should see less of these on the Typhoon H. Unfortunately to achieve a better picture or video than a Phantom 4 you also need a quality lens, processor and software. These are the three areas it appears the Typhoon is a bit weak on even though it has a high Mbps rate.

FIRST....Very Bad your Maths! :)
1 Byte = 8 bits
1 MB = 8 Mbits
100 MB = 800 Mbits
This is BASIC computer knowledge

100 MB /s = 800 Mbps !!
only DJI X5R can deliver video at 2400 Mbps! in SSD that flies too, inside the gimbal

SECOND, Yuneec mocks the people with the bitrates, MB and Mbps, When Yuneec say MB it means Mbps... Just go to the end of this article and you will understand what I am saying!

As for video compression bitrates (is the ability to compress the uncompressed footage, the bigest the rate the less the compression the better the quality) depending of the quality of the hardware and software, in conjunction with frames per second.
DVD quality is 6Mbps
Blue Ray HD is 20Mbps

The best cameras can deliver uncompressed video with huge frame rate (RED cameras, over 6K at 240 or 300 FPS)
but costs tens of thousands of dollars.

So, when Yuneec and DJI says 60Mbps, this is 60/8 = 7,5 MB/s
Unfortunately Yuneec's footage looks blurish because it slows down the frame rate and looses detail of the captured image, especially in RAW, (29Mbps = 3,6 MB/s)
Maybe it is software problem and can make it with a firmware update or hardware's capability.

As for the micro SD cards, the best card out there is SanDisk Extreme Pro micro SDXC
SANDISK EXTREME PRO SDSDQXP-064G-G46A 64GB
It can read video at 95MB/s, which means, 760 Mbps
and WRITE at 90MB/s, 720 Mbps,12 times faster than the faster write speed of both Phantom 4 and Typhoon H and costs around 40 euros.

Yuneec is cunning with their specifications because they addressing to people without technological background ...after all, they change them every a couple of months!!!!!

LOOK HOW THEY ARE CUNNING...
Below are the Bitrates from Panasonic website, for the Panasonic GH4, one of their best models..
Panasonic's bigest bitrate in 4K is 100 Mbps = 100 Mbits per second, 12,5 MB/s and 200Mbps in HD, 25MB/s.
4k HD Cameras - DMC-GH4KBody with 4K Video - Panasonic US

YUNEEC uses Panasonic GH4 body for the Tornado CGO4 Drone, the professional one, and in the specifications they say... INTEGRATED GH4 CAMERA WITH ZOOM LENS
And advertising it as 100 MB 30FPS!!!!!!!!! (800Mbps!!!)
Tornado CGO4
So If we believe the Yuneec, their panasonic camera is 8 times faster and bigger in quality than Panasonic itself!!!

JUST VISIT THE WEBSITE OF YUNEEC AND PANASONIC

[4K] 3840x2160 29.97p, 100Mbps (IPB) (LPCM) 23.98p, 100Mbps (IPB) (LPCM) [Full HD] 1920x1080 59.94p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) 29.97p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) 23.98p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) [4K] 3840x2160 25.00p, 100Mbps (IPB) (LPCM) [Full HD] 1920x1080 50.00p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) 25.00p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) [C4K] 4096x2160 24.00p, 100Mbps (IPB) (LPCM) [4K] 3840x2160 24.00p, 100Mbps (IPB) (LPCM) [Full HD] 1920x1080 24.00p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) [4K] 3840x2160 29.97p, 100Mbps (IPB) (LPCM) / 100Mbps (IPB) (AAC) 23.98p, 100Mbps (IPB) (LPCM) [Full HD] 1920x1080 59.94p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) 29.97p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) 23.98p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) 59.94p, 28Mbps (IPB) (AAC) 29.97p, 20Mbps (IPB) (AAC) [HD] 1280x720 29.97p, 10Mbps (IPB) (AAC) [VGA] 640x480 29.97p, 4Mbps (IPB) (AAC) [4K] 3840x2160 25.00p, 100Mbps (IPB) (LPCM) / 100Mbps (IPB) (AAC) [Full HD] 1920x1080 50.00p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) 25.00p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM) 50.00p, 28Mbps (IPB) (AAC) 25.00p, 20Mbps (IPB) (AAC) [HD] 1280x720 25.00p, 10Mbps (IPB) (AAC) [VGA] 640x480 25.00p, 4Mbps (IPB) (AAC) [C4K] 4096x2160 24.00p, 100Mbps (IPB) (LPCM) [4K] 3840x2160 24.00p, 100Mbps (IPB) (LPCM) [Full HD] 1920x1080 24.00p, 200Mbps (ALL-Intra) (LPCM) / 100Mbps (IPB) (LPCM) / 50Mbps (IPB) (LPCM)

More about Video Bitrates...
Understanding bitrates in video files - Encoding.com Help
 
Last edited:
So, when Yuneec and DJI says 60Mbps, this is 60/8 = 7,5 MB/s
Unfortunately Yuneec's footage looks blurish because it slows down the frame rate and looses detail of the captured image, especially in RAW, (29Mbps = 3,6 MB/s)
Maybe it is software problem and can make it with a firmware update or hardware's capability
With the latest firmware I did some testing that shows total bitrate for Gorgeous at 50074kbps, Natural at 49933kbps and RAW at 47590kbps. In another thread on this forum that was confirmed as well. Though not great not as bad as 29Mbps or 29696kbps.
 

Attachments

  • MB.png
    MB.png
    117.2 KB · Views: 16
So, when Yuneec and DJI says 60Mbps, this is 60/8 = 7,5 MB/s
Unfortunately Yuneec's footage looks blurish because it slows down the frame rate and looses detail of the captured image, especially in RAW, (29Mbps = 3,6 MB/s)
Maybe it is software problem and can make it with a firmware update or hardware's capability.

Just to be clear, RAW mode changes the image *less* than any of the others (that's a good thing!). All small sensor cameras produce slightly 'blurred' images out of the sensor - then usually in-camera software puts in sharpening to make it look more detailed. This happens with all action cameras - from GoPro though to DJI and 'no-name' brands. RAW switches off the artificial sharpening, it doesn't 'loose' detail.

The point here is that the little processor in the camera only has so much power. It uses the 'simplest' compression algorithm that does the job and might do basic sharpening, but that's never as good as you can do in professional software on a desktop PC. For the best quality output, you want to make sure the camera does as little as possible, then do it properly afterwards. RAW may look 'blurry' but it has all the information you need for the best possible final image.

Not all compression is the same. I can compress video much more on my PC in post processing than a little GoPro can whilst it's filming and still get better looking results. Making the bitrate higher by turning up sharpening in the camera doesn't give you more information, it just makes the video harder to edit later.

So, if you don't want to edit the video, feel free to turn up sharpening and use Natural or Gorgeous modes to boost the saturation. It'll also give you higher bitrates, but that doesn't mean much.

If you do plan to edit the video and want the best possible image quality, shoot in RAW and expect to add a sharpening effect (unsharp mask or similar) and colour grading in post processing.
 
Last edited:
  • Like
Reactions: banned user
People should spend their time worrying about how it looks instead of how you count it. How you count it should be based on how it looks. :)
 
  • Like
Reactions: cokeaddict
Just to be clear, RAW mode changes the image *less* than any of the others (that's a good thing!). All small sensor cameras produce slightly 'blurred' images out of the sensor - then usually in-camera software puts in sharpening to make it look more detailed. This happens with all action cameras - from GoPro though to DJI and 'no-name' brands. RAW switches off the artificial sharpening, it doesn't 'loose' detail.

The point here is that the little processor in the camera only has so much power. It uses the 'simplest' compression algorithm that does the job and might do basic sharpening, but that's never as good as you can do in professional software on a desktop PC. For the best quality output, you want to make sure the camera does as little as possible, then do it properly afterwards. RAW may look 'blurry' but it has all the information you need for the best possible final image.

Not all compression is the same. I can compress video much more on my PC in post processing than a little GoPro can whilst it's filming and still get better looking results. Making the bitrate higher by turning up sharpening in the camera doesn't give you more information, it just makes the video harder to edit later.

So, if you don't want to edit the video, feel free to turn up sharpening and use Natural or Gorgeous modes to boost the saturation. It'll also give you higher bitrates, but that doesn't mean much.

If you do plan to edit the video and want the best possible image quality, shoot in RAW and expect to add a sharpening effect (unsharp mask or similar) and colour grading in post processing.


Good to see that you still don't understand how a camera works and how bitrate relates. Small sensor cameras don't "produce slightly 'blurred' images out of the sensor" thats not the case at all. the size of the sensor has nothing to do with image sharpness. Yes camera manufactures use varying amounts of sharpening in any of their "look" modes. Sharpening is not the reason that files are larger or smaller. In fact in most every other camera on the market from small action cams to DSLR's to Professional Production cameras have the largest file size when using RAW. Yuneec for some odd reason choose not to do that. Bitrate is a sign of compression level the lower the bitrate the more compressed the signal and on a 4:2:0 camera that is not a good thing

From a previous post of mine that you obviously haven't read.

Let me address the parts of the puzzle. lets look a the image chain from front to back. If you have a high resolution lens more information will get to the sensor, if you have a lower resolution lens it will lower the amount of information (data) that then sensor sees. Same goes for focus. Being in focus increases contrast, higher contrast is more information and that makes for more data. Being lower res reduces the processor overhead. If we were to record each and every bit that comes off the sensor and we are shooting 4K (4096x2160) at 24fps, there are three color channels (YUV) and each one is 8bit's deep (10bit would be better and 14bit is what is used for feature and TV production) and that gives a frame file size of 26.5MB. One second is 637MB or about 300Gigabytes/hour.... That's a lot of data that the processor has to wade through. If the image has a bit lower resolution you will limit the about of data since your GOP I frames have less difference across the frame. Now you have to compress that and that's getting rid of a ton of data and that's where codec's come into play. Some are good most are bad but in the end your grouping similar tones in the image and writing a key frame that describes the entire frame then your using those groups to only update the areas that have a change (GOP compression) I could get into I Frames P frames and B frames but it all has to do with motion and there where the variable bitrate comes in that is confusing Tuna.

Lets look at why Bitrate is an important number it gives and indication of compression ratio. Yuneec says that their cameras output 50Mbps or about 102/1 compression ratio. When they first advertised the capabilities of the CGO3+ they said it was 100Mbps which would have been 51/1 compression ratio. The Yuneec Raw is only giving you about 29Mbps which is 176/1 compression ratio. Compare those numbers to the P4 and the GPH4B which are running 85/1 compression and you start to see a performance metric.

In the end it's about color accuracy, contrast, and focus which is collectively called image quality and all of those are affected by Bitrate. Saying that Bitrate isn't a performance indication is a sign of not understanding the chain. One other thing to consider, if Bitrate wasn't important Yuneec wouldn't have advertised 100Mbps then changed over all their material to show 50Mbps...

And I posted this comparison

Did a comparison of the Bitrates Data wrappers, codecs and color space tonight between the DJI X3, the CGO3+ in each of its daylight modes and the GoPro Hero 4 Black
video-codec-camparisons-jpg.1985


a few things the Yuneec frame rate will require transcoding to use in anything broadcast because its not at the correct 23.976fps and it seems that Yuneec is capping the variable bitrate to 50Mbps so no it will not go higher as some have suggested. they are also using an old codec that does have some decoding issues. The choice of format profile is also puzzling since its not optimized for 4K. It looks like one of the reasons that the exposures steps so bad is their choice in GOP frames, they are reading inter frames ever second where go pro is every half second and DJI is every 1/3rd second. Yuneec is also compressing their image more than DJI and GoPro
 
DerStig, I hadn't seen the comparison of the different camera codes before, thanks for posting that. I think you have some of your information backwards, though. The thing that stands out to me is that Yuneec uses a much longer GOP than DJI does (24 b-frames vs. 8 b-frames between i-frames). That makes it more efficient at the same bitrate than DJI's implementation. As you undoubtedly know, encoding a full i-frame (independent frame) takes much more bandwidth than encoding b-frames (differential frames). The DJI codec has three times as many independent frames as the Yuneec codec, which takes fewer resources to encode and decode, but requires more bandwidth. On the downside, every differential frame has to trade off fidelity for efficiency, so having a GOP that is too long will result in loss of quality between independent frames.
 
DerStig, I hadn't seen the comparison of the different camera codes before, thanks for posting that. I think you have some of your information backwards, though. The thing that stands out to me is that Yuneec uses a much longer GOP than DJI does (24 b-frames vs. 8 b-frames between i-frames). That makes it more efficient at the same bitrate than DJI's implementation. As you undoubtedly know, encoding a full i-frame (independent frame) takes much more bandwidth than encoding b-frames (differential frames). The DJI codec has three times as many independent frames as the Yuneec codec, which takes fewer resources to encode and decode, but requires more bandwidth. On the downside, every differential frame has to trade off fidelity for efficiency, so having a GOP that is too long will result in loss of quality between independent frames.

Is he still posting? Yes, he gets a lot of his information back to front I'm afraid. He's an expert though, so we're not allowed to contradict him. :) The DJI and Yuneec codecs have very different characteristics, but I've not seen any evidence yet of quality issues that can be attributed to the codec of either platform. Unfortunately the fixation on bitrates is a bit like the fixation on pixel counts that dominated camera marketing a decade ago - it's a convenient number that consumers can latch on to. I'd be more concerned about sensor size and Nyquist Limit but these are things that marketing guys can't sell and that show up the hard limits of the sort of cameras you get with this class of drone.
 
I totally agree that the obsession with the bitrate is misguided. If the codec can encode the video at 50Mbps instead of 60Mbps (as used by DJI) with no discernable degradation, then there is nothing at all to be gained from wasting more space on the SD card.

Another inaccuracy is the statement that "they are also using an old codec" -- both DJI and Yuneec use the same codec and profile, the only difference is the GOP length.
 
DerStig, I hadn't seen the comparison of the different camera codes before, thanks for posting that. I think you have some of your information backwards, though. The thing that stands out to me is that Yuneec uses a much longer GOP than DJI does (24 b-frames vs. 8 b-frames between i-frames). That makes it more efficient at the same bitrate than DJI's implementation. As you undoubtedly know, encoding a full i-frame (independent frame) takes much more bandwidth than encoding b-frames (differential frames). The DJI codec has three times as many independent frames as the Yuneec codec, which takes fewer resources to encode and decode, but requires more bandwidth. On the downside, every differential frame has to trade off fidelity for efficiency, so having a GOP that is too long will result in loss of quality between independent frames.

None of the information is backwards, this is a camera on a moving platform you want more I frames not less. The less you have the lower resolution your image is due to temporal artifacting
 
  • Like
Reactions: omnius
Is he still posting? Yes, he gets a lot of his information back to front I'm afraid. He's an expert though, so we're not allowed to contradict him. :) The DJI and Yuneec codecs have very different characteristics, but I've not seen any evidence yet of quality issues that can be attributed to the codec of either platform. Unfortunately the fixation on bitrates is a bit like the fixation on pixel counts that dominated camera marketing a decade ago - it's a convenient number that consumers can latch on to. I'd be more concerned about sensor size and Nyquist Limit but these are things that marketing guys can't sell and that show up the hard limits of the sort of cameras you get with this class of drone.

EXCELLENT Googling there Tuna, now if you only understood what you cut'n'pasted. My information is NOT back to front and you still don't understand what is going on with the cameras as exampled by your use of a GoPro Hero Session camera as your comparison baseline. We get it you have a vested interest in the Yuneec products as shown by your ad in your sig.
 
only there is a discernible degradation due to temporal error and yes MP42 is an older codec.
The implementation may be different, but they use the same ISO-standard video format and profile. The implementations are constrained by the same specifications.
 
it's Ford vs Chevy, the both use V-8's they both are rear wheel drive therefore they are the same. No but hey if it makes you feel better go on and think that
Your analogy doesn't make any sense. We are talking about an international standard that guarantees interoperability. There are no doubt small differences in the way the codec implementations handle the detection of inter-frame changes, but these are likely not perceptible by the average user -- realistically you would have to subtract one frame from one implementation from the same frame of another implementation to visualize them. And most of the differences would be hidden by the 4:2:0 chroma subsampling that these codecs use.
 
Your analogy doesn't make any sense. We are talking about an international standard that guarantees interoperability. There are no doubt small differences in the way the codec implementations handle the detection of inter-frame changes, but these are likely not perceptible by the average user -- realistically you would have to subtract one frame from one implementation from the same frame of another implementation to visualize them. And most of the differences would be hidden by the 4:2:0 chroma subsampling that these codecs use.

I worked on the predecessor to this: Media Analysis Solutions for IP/SDI Infrastructure | Tektronix which did all of the analysis for compliance and video quality - and most of the restrictions on GOP rates were based on expected drop out of satellite links. If you're reading from an SD card you don't have those issues and the only real restriction is that it makes it harder to search and play streams backwards. :)

Perhaps the biggest issue with this sort of discussion is that getting good results from a flying camera is a tough challenge and everyone gets very different results. It's easy to blame the tools (and it's true that the tools aren't easy to use - Yuneec have work to do still), but we still haven't seen *any* actual evidence that the codec introduces any problems in video quality.
 
An unstabilized action camera like the GoPro would certainly struggle with long GOPs because the inter-frame differences often are enormous. Eventually you run out of bandwidth to encode them or you just don't have the horsepower to analyze the motion vectors in realtime, and the picture breaks up until you get the next i-frame.

But our drones have stabilized gimbals, and our footage doesn't have snow or water droplets flying across the field of view. We can tolerate long GOPs, and that allows very good quality to be delivered at a relatively low bitrate. There is nothing to be gained from using the 100 Mbps bitrate some people are clamoring for. As you said, these are acquisition and delivery formats, not editing formats.
 
  • Like
Reactions: omnius

New Posts

Members online

Forum statistics

Threads
20,955
Messages
241,595
Members
27,286
Latest member
lahorelaptop