3d mapping training for yuneec h520 & starting in 3d mapping

[Tuna]

Best not give them the oxygen by re-posting that video. Inaccurate and biased at best, it seems.

 

[dsandson]

Flawed, how? Remember he also criticised the bigger X5s and X7s cameras from DJI for mapping. He actually likes the drone and had been praising it until he did this like for like test. I'd still have an H520 for mapping but best know the positives and negatives first.

 

[Tuna]

1. The P4P is not a global shutter camera - it's mechanical, which means it has exactly the same rolling shutter effect as the E90. In fact there are complaints about it on the Pix4D forums, and they issued a bug fix for the P4P to make it fly slower for surveys. DronePilot actually has a mode where it stops for photos if rolling shutter is an issue for you.

2. He takes no account of (and doesn't seem to understand) the fact that the E90 has a wider field of view than the P4P. That means it captures less detail at the same height. He flew both missions at the same height and then is surprised when the E90 has lower ground detail.

3. He makes no mention of ISO and exposure settings on each machine. This is basic stuff and will hugely affect the quality of the result.

4. The fact that he 'needed' Pix4D to run a survey mission on the H520 shows how little understanding he has of mapping. If he's only capable of creating a mission in Pix4D, how can he realistically compare platforms? He's clueless.

5. The rotor thing shows he really has no experience of Yuneec drones at all. Five rotor mode has been a thing for nearly three years now, with plenty of demo videos online to show how it behaves.

6. He doesn't say what firmware he's running on. That's quite important as the camera has undergone some significant changes over the last few months. If he's not upgraded, then he won't be getting the same results as those who have.

The damning with faint praise thing stinks.This is a hatchet job on the H520, based on flawed premises.

 

[FasterPastor]

1. The P4P is not a global shutter camera - it's mechanical, which means it has exactly the same rolling shutter effect as the E90. In fact there are complaints about it on the Pix4D forums, and they issued a bug fix for the P4P to make it fly slower for surveys. DronePilot actually has a mode where it stops for photos if rolling shutter is an issue for you.

2. He takes no account of (and doesn't seem to understand) the fact that the E90 has a wider field of view than the P4P. That means it captures less detail at the same height. He flew both missions at the same height and then is surprised when the E90 has lower ground detail.

3. He makes no mention of ISO and exposure settings on each machine. This is basic stuff and will hugely affect the quality of the result.

4. The fact that he 'needed' Pix4D to run a survey mission on the H520 shows how little understanding he has of mapping. If he's only capable of creating a mission in Pix4D, how can he realistically compare platforms? He's clueless.

5. The rotor thing shows he really has no experience of Yuneec drones at all. Five rotor mode has been a thing for nearly three years now, with plenty of demo videos online to show how it behaves.

6. He doesn't say what firmware he's running on. That's quite important as the camera has undergone some significant changes over the last few months. If he's not upgraded, then he won't be getting the same results as those who have.

The damning with faint praise thing stinks.This is a hatchet job on the H520, based on flawed premises.

It does appear he made a mistake, but I am not sure that he is "clueless" - I do think your point about trying to be an expert on everything, could be an issue here. At the same time I do think they have helped a lot of folks get started and gain confidence in the industry. ( I am not a drone U member or have any affiliation with them )

I have seen a few other lapses, he said there was a thermal camera for the Inspire 2 - there is not, but they have done over 500 podcasts (maybe 800, not sure??)

I did an informal test in mapping to see if I could tell the difference in mapping between the I2, the P4P and the H520, but it was strictly a NADAR map (not 3d)

But overall, despite the lapse on the tech side of how things work on the video, I came to the same conclusion, the P4P has turned out to be a much more flexible for me and making maps. Where the software limitations of the H520, let me to let it go.

I was really hoping for more, and some of the basic features that are missing, were quite surprising. - some features I loved, but it still stayed home most of the time.

If we are honest really good hardware WITH functional software, is not an easy thing to accomplish at this stage, we have very high expectations at this point, everything must work, every time, and pretty well.

I am wondering at this point, if Yuneec is better off introducing mapping features the the H+ so they have one drone that can do many things at a better price point. To spend $3,500 on a drone setup that will not take one panoramic shot, is disappointing.

Another idea they could do, is come out with a thermal camera with improved resolution that is radiometric calibrated. Right now, you have to spend a lot of $$$ to get a dji system that does this.

 

[Haydn]

It does appear he made a mistake, but I am not sure that he is "clueless" - I do think your point about trying to be an expert on everything, could be an issue here. At the same time I do think they have helped a lot of folks get started and gain confidence in the industry. ( I am not a drone U member or have any affiliation with them )

I have seen a few other lapses, he said there was a thermal camera for the Inspire 2 - there is not, but they have done over 500 podcasts (maybe 800, not sure??)

I did an informal test in mapping to see if I could tell the difference in mapping between the I2, the P4P and the H520, but it was strictly a NADAR map (not 3d)

But overall, despite the lapse on the tech side of how things work on the video, I came to the same conclusion, the P4P has turned out to be a much more flexible for me and making maps. Where the software limitations of the H520, let me to let it go.

I was really hoping for more, and some of the basic features that are missing, were quite surprising. - some features I loved, but it still stayed home most of the time.

If we are honest really good hardware WITH functional software, is not an easy thing to accomplish at this stage, we have very high expectations at this point, everything must work, every time, and pretty well.

I am wondering at this point, if Yuneec is better off introducing mapping features the the H+ so they have one drone that can do many things at a better price point. To spend $3,500 on a drone setup that will not take one panoramic shot, is disappointing.

Another idea they could do, is come out with a thermal camera with improved resolution that is radiometric calibrated. Right now, you have to spend a lot of $$$ to get a dji system that does this.

The Radio metric thermal is coming it is called the E10T. It is the Boson 320 sensor initially... Then the option of the 640 will come in.
It should have an option of lens but unsure to what extent. And these will be facture tuned prior to the customer.
They will both be Radiometric with the lower light RGB camera as well.
Haydn

 

[dsandson]

Well look, I better say that i am a Drone U member. However I'm not the sort of person to always take what I hear for gospel. I'm a Civil Engineer and former pilot, and my interest in drones comes from merging these interests into 3D mapping by drone. I have recently gained my UK PfCO for my employer, but I've been researching and working on non-work mapping for the last year. Drone U has been helpful for me in that one of their interests is mapping, and found I was learning more through them than through general forums. I suppose the people who are regularly mapping with drone aren't always the ones on internet forums.

I have so say e have a very active facebook group. Now I genuinely hate FB, so it was very begrudgingly that I started using it. But i have to say I'm still to this day very impressed how civil, hugely active and helpful the group of people on there are. Don't think I've seen any discussion descent into arguments, or have to be moderated. That said I'm really not a cheerleader, and wont jump onto any and all forums to argue or defend this lot. And I'm here as I have an interest in Yuneec drones, and specifically the H520 for mapping. (Another disclaimer, I own a P3P... don't hold it against me... it was cheap and has helped me to cheaply validate the mapping and imaging process to my employer who wouldn't have jumped on their own). Right that wasnt an advert, just wanted to explain where I'm coming from.

1. The P4P is not a global shutter camera - it's mechanical, which means it has exactly the same rolling shutter effect as the E90. In fact there are complaints about it on the Pix4D forums, and they issued a bug fix for the P4P to make it fly slower for surveys. DronePilot actually has a mode where it stops for photos if rolling shutter is an issue for you.

I'm no photographer, so had to go do some research. Whilst your right about the mechanical shutter, in doing some independent reading most of what i saw was that this configuration does solve the majority of the rolling shutter effect. Drone U seem to have been getting a lot of their info from meeting with Pix4D reps. I think Pix4D's logic was that corrections can be applied through software for rolling shutter, but post-imaging corrections can only go so far. However Pix4D only has 2 options, Rolling Shutter or Global Shutter. I'd wonder if its better knowing a correction had been applied versus not being able to quantify the residual rolling shutter effect on the mechanical shutter.

2. He takes no account of (and doesn't seem to understand) the fact that the E90 has a wider field of view than the P4P. That means it captures less detail at the same height. He flew both missions at the same height and then is surprised when the E90 has lower ground detail.

Thats a good point. He does allude to it but does seem to get waylaid. Explains the lower point density.

3. He makes no mention of ISO and exposure settings on each machine. This is basic stuff and will hugely affect the quality of the result.

I assume he's using the 'sunny' option. Pix4D Capture, like most mapping apps, doesn't offer fine grain control of these settings.

4. The fact that he 'needed' Pix4D to run a survey mission on the H520 shows how little understanding he has of mapping. If he's only capable of creating a mission in Pix4D, how can he realistically compare platforms? He's clueless.

He needed Pix4D Capture to do as close a comparison as possible. Hence the idea of trying to be scientific. Its a straight comparison, not a university thesis. He's had the drone I reckon since at least last November, ands has had enough success with it to be recommending it for pure mapping up to this point.

5. The rotor thing shows he really has no experience of Yuneec drones at all. Five rotor mode has been a thing for nearly three years now, with plenty of demo videos online to show how it behaves.

Yeah, I've seen that Yuneec UK Youtube clip, and other accounts from TH owners here. Dont know where he got that from, but I assume that since Yuneec isn't famed for giving out review samples, then he owns it and he hasnt been inclined to test it out.

6. He doesn't say what firmware he's running on. That's quite important as the camera has undergone some significant changes over the last few months. If he's not upgraded, then he won't be getting the same results as those who have.

Thats something I've been thinking about... how does the mapping software manage the image distortion removal when the in camera adjustments keep changing with each firmware update? Was there not a minimum firmware level that enabled Pix4D Capture to function?

 

[arruntus]

Like any video that is published on the Internet, the first thing to do is to thank the effort that person or people have made to publish it. Then we can have a say in what is being said and then we can all learn.

I'm going to continue with the points that the partner @dsandson has made.

1. The mechanical shutter is not the same as the global shutter but it helps a lot to avoid the rolling shutter effect. Therefore, it is certainly better to have it than not to have it as is the case with the E90.
2. Whether you have a camera with more or less field of view is determined by each manufacturer. So when you compare two cameras (of two different drones) you cannot enter variables that only correspond to the manufacturers. It would be a question of asking Yuneec or DJI why they have set that field of view if it is better or worse for photogrammetry. But it is not relevant to what is discussed in the video as it is a parameter that the user can not vary so you take what you have and compare the final results.
3. Not to mention the ISO used and all the other parameters, I think it's a small glitch. It is recommended to use them fixed, not in automatic, so that the data does not suffer variations for this reason.
4. The fact that I use Pix4D capture, I think he mentions it or I think I understand it (my English is still very bad) to try to make the comparison as fair as possible. Doesn't it say so? Therefore, two drones are compared with the same application to avoid that different applications can distort the results and include even more variables. I think that's the right way to go when you compare two different products. As close as possible and get the most accurate results possible.
5. The issue of the 5 motors in a hexacopterus, I think it is confused or not well explained and when he says that they turn without control refers to a quadcopter. We can all be confused (I am the first )
6. The subject of distortion, I don't know whether to talk about it........... But that you have to correct something by software when it should come in handy for hardware is bleeding enough to mention in a comparison with another drone.

Comrades, don't take it as an offense. I have an H520, I will never buy a DJI but I have no problem recognizing that today is a better drone because reality says so. I'd love to say otherwise, I've been waiting 10 months to say it and I still can't

 

[Graham]

We know that there is variability in the quality of the images from the E90. Arruntus has a camera that does not suffer from the distortion that mine does and apparently there are other cameras similar to his (Arruntus, if you decide to sell, let me know). Since the last update we have the option to ameliorate the distortion and I believe this has improved the number of Automatic Tie Points seen in my Pix4D analyses.

We do not know if the E90 used in the video comparison was using the distortion fix and nor do we know what firmware was in use. Furthermore, a sample of one may not be representative of the range of cameras.

I always use the option to halt the aircraft when doing surveys as I prefer to avoid blur.

It seems the C23 is producing images that are extremely similar to the E90.

 

[dsandson]

2. Whether you have a camera with more or less field of view is determined by each manufacturer. So when you compare two cameras (of two different drones) you cannot enter variables that only correspond to the manufacturers. It would be a question of asking Yuneec or DJI why they have set that field of view if it is better or worse for photogrammetry. But it is not relevant to what is discussed in the video as it is a parameter that the user can not vary so you take what you have and compare the final results.

Well increased FOV gives less detail, but will help improve photogrammetry tie points as it helps improve overlap. I've heard recommendations for ground based close-up photogrammetry is to use a fish eye camera such as the GoPros. I guess its not bad, just different.

It seems the C23 is producing images that are extremely similar to the E90.

Have to say I've read a lot of negativity about the E90 camera on here. I would hope the new firmware update has taken the lessons learnt from the C23 (as it appears to be getting great reviews on this board) and applied it here. The datapilot firmware of 1.4 sounds like a major release. Maybe something interesting there. Anyway I'll watch and when the time comes for an upgrade hopefully the H520 cuts fits the bill.

 

[arruntus]

Well increased FOV gives less detail, but will help improve photogrammetry tie points as it helps improve overlap. I've heard recommendations for ground based close-up photogrammetry is to use a fish eye camera such as the GoPros. I guess its not bad, just different.

In my opinion I think you're wrong, in fact, what you're looking for most in a photogrammetry camera is a straight line lens. With as little distortion as possible. The fisheye in a camera must be corrected by software to use the images, which would be counterproductive.

The fov is relatively important but the important thing to sew well the images is the overlap. That's what the pilot determines when he schedules the mission. If you have a lot of fov you obviously have less detail and it can be more expensive for the program to sew easily.

 

[Tuna]

Whether you have a camera with more or less field of view is determined by each manufacturer. So when you compare two cameras (of two different drones) you cannot enter variables that only correspond to the manufacturers. It would be a question of asking Yuneec or DJI why they have set that field of view if it is better or worse for photogrammetry. But it is not relevant to what is discussed in the video as it is a parameter that the user can not vary so you take what you have and compare the final results.

The user absolutely can vary a parameter to account for field of view - they simply change the height of the drone. Most mapping software allows you to specify either a 'ground resolution' or height when setting up a flight. If you choose ground resolution, the height is calculated back from the FOV of the camera. That in turn will decide the 'scan width', and along with your overlap will define how many passes are needed to cover a given area.

The guy in the video did none of that - he flew both missions at a fixed height of 90m. Choosing to fix your height means that the ground resolution is completely determined by the FOV.

Hence it's completely relevant to the video. He either doesn't understand how those flight parameters work, or deliberately chose to fly in a way that results in lower ground resolution on the E90.

 

[arruntus]

The user absolutely can vary a parameter to account for field of view - they simply change the height of the drone. Most mapping software allows you to specify either a 'ground resolution' or height when setting up a flight. If you choose ground resolution, the height is calculated back from the FOV of the camera. That in turn will decide the 'scan width', and along with your overlap will define how many passes are needed to cover a given area.

The guy in the video did none of that - he flew both missions at a fixed height of 90m. Choosing to fix your height means that the ground resolution is completely determined by the FOV.

Hence it's completely relevant to the video. He either doesn't understand how those flight parameters work, or deliberately chose to fly in a way that results in lower ground resolution on the E90.

You're right, looking for the same GSD would have been more accurate than using the same height.

 

[FlyaDrone]

Might consider the Agisoft Photo Scan app for affordable mapping experience. If you have a high end computer it has a local site license capability which allows you to run it on your machine. A two level pricing structure permits you the opportunity to gain valuable experience and determine if you wish to buy the high end option. Much tutorial info in YouTube. I hope to hear experienced mappers feedback on if this is an affordable option versus cloud computing from Drone Deploy or Pix4D subscription services.

And if UgCS Ground Station could be supported on H520 you would have terrain following capability. Again I look forward to getting feedback on if that is a viable option.

I am eager for the H520 to live up to and deliver on it's potential. Would be glad to sell my Inspire 2 and fly an open source system. Yuneec, get serious with your optical improvements and create a SDK for increased OEM payload options.

 

[arruntus]

Might consider the Agisoft Photo Scan app for affordable mapping experience. If you have a high end computer it has a local site license capability which allows you to run it on your machine. A two level pricing structure permits you the opportunity to gain valuable experience and determine if you wish to buy the high end option. Much tutorial info in YouTube. I hope to hear experienced mappers feedback on if this is an affordable option versus cloud computing from Drone Deploy or Pix4D subscription services.

And if UgCS Ground Station could be supported on H520 you would have terrain following capability. Again I look forward to getting feedback on if that is a viable option.

I am eager for the H520 to live up to and deliver on it's potential. Would be glad to sell my Inspire 2 and fly an open source system. Yuneec, get serious with your optical improvements and create a SDK for increased OEM payload options.

With Pix4D you can process both in the cloud and locally. In fact, I've never actually processed in the cloud

 

[dsandson]

Might consider the Agisoft Photo Scan app for affordable mapping experience.

PhotoScan actually costs the same for a perpetual license as Pix4D costs for a years subscription. It is apparently very good, and supposedly scales better than Pix4D for very large projects.

And if UgCS Ground Station could be supported on H520 you would have terrain following capability.

UgCS is actually supports the H520. The only issue I suppose is that the free tier doesn't support terrain following. That would be the Pro licence at around €600 which is steep!

I am eager for the H520 to live up to and deliver on it's potential. Would be glad to sell my Inspire 2 and fly an open source system.

Yeah as it stands the mapping functionality just isn't enough for me. However I have time to wait and see if TH+ functionality can be transferred to data pilot so I wouldn't have to keep a DJI backup drone. Fingers crossed it gets delivered as I'd really like the poor weather stability so I can fly more often in the winter time.

 

[sercan04448]

Hi everyone.

We have a h520 . We are using mushrooms one (right) . We are wondering differences of two antennas. Which one is useful for our flights.

 

[h-elsner]

We have a h520 . We are using mushrooms one (right) . We are wondering differences of two antennas. Which one is useful for our flights.

This has absolutely nothing to do with 3D mapping. Why do you reanimate a thread older than 4 years? I will never understand....

The two antennas have different direction patterns (radiation patterns, Radiation pattern - Wikipedia). The mushroom has maximum 360° around the head of the mushroom. It is good for having signal from all direction, also back.
The patch antenna has a narrow antenna pattern in direction of the flat side for wider range but needs to be directed to the drone.

Each of them have advantages and disadvantages (range vs. direction). You have to choose the one you need for the current task.

br HE

 

[DoomMeister]

The mushroom is omnidirectional and should be positioned to be like a mushroom growing from the ground.

The other is a patch antenna and provides for more directional gain, but must have the flat surface pointed at the aircraft all the time.

The patch will retain the video link better as long as the pilot aims it at the aircraft. Maybe more important is the positioning of the two stick antennas that are used in controlling the aircraft. They should be pointed up and out at 45 degrees (90 degrees to each other) for best results.

With mushroom


With patch