Recommended Posts

Very well said, Kyle.

I couldn't agree more with what you've said. I also feel your pain with dealing with TBC and the UX5. When i first started using drones, I was trained on the first gen UX5 and was tasked with flying cut blocks up in northern Alberta, Canada. As you said, the processing is a critical point that cant be overlooked. I havent worked with the photogrammetry module in TBC since then, but what I remember about it was that the software just wasn't up to snuff. What Trimble did was purchase a company out of Belgium i believe called Inpho. As the big players usually do, they bought out a little company and incorporated it into their software package. The problem that I ran into with TBC was that the Inpho photogrammetry module was designed in such a way that it was not capable to threading all available processors. This slowed down the processing huge. In the end, I was never able to produce any useful data using the UX5 and TBC. 

Since then I have become very acquainted with Pix4d. They have become my processing software of choice for the type of flying I do. 

Do you use any other processing software other than TBC?

Share this post


Link to post
Share on other sites

Hello All,

I am new to this thread but am interested if anyone has done a comparison of the pro's and con's of the different photogrammetry software packages available?  To be honest I haven't read this thread in it's entirety so I apologize if I missed it here.  

My criteria are fast and accurate.  I am using it for deformation monitoring of earthen dams.  I may use Structure from Motion processing as well as LiDAR data sets.

I've looked at Trimble Inpho, Agisoft Photoscan, Pix4D, Context Capture (Bentley), and SimActive's 3D Correlator.

Thanks!

Brett

Share this post


Link to post
Share on other sites
8 minutes ago, Brett Baker said:

Hello All,

I am new to this thread but am interested if anyone has done a comparison of the pro's and con's of the different photogrammetry software packages available?  To be honest I haven't read this thread in it's entirety so I apologize if I missed it here.  

My criteria are fast and accurate.  I am using it for deformation monitoring of earthen dams.  I may use Structure from Motion processing as well as LiDAR data sets.

I've looked at Trimble Inpho, Agisoft Photoscan, Pix4D, Context Capture (Bentley), and SimActive's 3D Correlator.

Thanks!

Brett

There are more threads with the topic embedded but this will get you started. ESRI's Drone2Map was based off of the Pix4Dmapper software but they are going native and the next version is supposed to be all their own. simActive's Correlator3D has some unique hardware requirements but it is some slick software if you have a GPU that will run it. The GPU needs to be OpenGL and OpenCL compliant, and not that many high-end GPUs fit that bill.

Edited by R Martin
additions and some text. We all like text to read.
  • Like 3

Share this post


Link to post
Share on other sites

Thanks!  I can buy the PC to fit the requirements of the software.  I've heard good things about programs that were old school photogrammetry solutions scaled to drone use.  I'll be laying out GCP's and a GPS base station to try to nail the accuracy.  I have ESRI GIS so their solution may be a good way to go. -Brett

Share this post


Link to post
Share on other sites
42 minutes ago, Brett Baker said:

Thanks!  I can buy the PC to fit the requirements of the software.  I've heard good things about programs that were old school photogrammetry solutions scaled to drone use.  I'll be laying out GCP's and a GPS base station to try to nail the accuracy.  I have ESRI GIS so their solution may be a good way to go. -Brett

Last I heard Drone2Map was running $1400.00 per year but not sure if that is a complete license or per seat. They do have an eval period you can sign up for and trial the software. I have seen it but not used it. It does have a few features that are unique that I wish the other products had. I am waiting until the ESRI native version is released before I go any further with it. I have a Pix4D license so buying additional software that is based on my current software just does not make sense.

  • Like 1

Share this post


Link to post
Share on other sites
1 hour ago, Brett Baker said:

Thanks!  I can buy the PC to fit the requirements of the software.  I've heard good things about programs that were old school photogrammetry solutions scaled to drone use.  I'll be laying out GCP's and a GPS base station to try to nail the accuracy.  I have ESRI GIS so their solution may be a good way to go. -Brett

Some things to keep in mind and that you might want to check out.  Take a look at this drone: https://www.microdrones.com/en/mdaircraft/md4-1000/  This drone costs approximately $65K  I'm suggesting that you buy it but that's what it costs to get the level of accuracy required to certify aerial data.  In part, because it uses instrumentation like https://www.applanix.com/products/dms-uavs.htm, a $15K IMU and more expensive L1 and L2 GPS receivers etc.

If your career or your customer's career depends on your ability to certify aerial data then you need to seriously consider whether a drone with off the shelf 5hz GPS, an IMU that was developed to be used in a cell phone and a sub $2K "RTK" base station will provide you with the data you can trust.

Also, the majority of the Photogrammetry software is based on on the same white papers from about 15 years ago which is why most of its all the same.  There are primarily two types, Pix4D which is predominantly used for NADIR and ContextCapture which is used for Oblique scanning.  One is better for farming the other is better for engineering.   

Here's a scan we did of Glendale http://realitymodeling-pw.bentley.com/a3D/Construct Timelapse/models/4-12/App/?scene=http://realitymodeling-pw.bentley.com/a3D/Glendale/Production_1/Scene/Production_1.3mx#%2F  using ContextCapture.  This is a project where a dignitary was going to speak and law enforcement wanted to determine snipper angles, parade routes, evacuation routes, etc., so certification of the data wasn't required.  For the amount of space and the vertical nature of the scan, NADIR would not have been nearly as effective as Oblique. 

I'm not a civil engineer so I know very little about most of the terrestrial based tools, but we've been working with Bentley Systems and a few large engineering firms and have tested a ton of "solutions."  As much as people want to believe that things like SiteScan, DroneDeploy, and DJI have they answer if you want certifiable data, they aren't.   

Depending on what problem you're trying to solve do you really want to potentially jeopardize your career using a 16MP camera, carried on a toy and controlled by a cell phone app?  As much as the idea of cloud computing and having the model rendered before your back to your desk sounds sexy, there are way too many holes in that value chain that you should not trust. 

 

  • Like 2

Share this post


Link to post
Share on other sites

Thanks for the great info!  Microdrones have already been out to visit us and collected some test data.  They are going to return soon and do it again.  We are also looking at PulseAero Vapor55.  Yeah, the money adds up fast.  Our driving force is safety, deformation monitoring.  We have several earthen dams that don't meet today's standards for seismic stability.  One straddles a creeping fault (14mm/yr).  I spoke with a gentleman from southern Utah who told me he prefers ContextCapture over Pix4d.  I think he said he did some comparisons between the two and found he got better results with the Bentley product.   We will probably base most of our work on lidar data but we also have projects where the imaging is required.  For the LiDAR we are looking at the Reigl VUX-1.

Share this post


Link to post
Share on other sites
3 hours ago, Brett Baker said:

Thanks!  I can buy the PC to fit the requirements of the software.  I've heard good things about programs that were old school photogrammetry solutions scaled to drone use.  I'll be laying out GCP's and a GPS base station to try to nail the accuracy.  I have ESRI GIS so their solution may be a good way to go. -Brett

Whatever solution you choose, don't be cheap on the workstation. Buy the best you can possibly get and save yourself a lot of hours waiting on results. 

  • Like 1

Share this post


Link to post
Share on other sites
1 hour ago, Av8Chuck said:

Some things to keep in mind and that you might want to check out.  Take a look at this drone: https://www.microdrones.com/en/mdaircraft/md4-1000/  This drone costs approximately $65K  I'm suggesting that you buy it but that's what it costs to get the level of accuracy required to certify aerial data.  In part, because it uses instrumentation like https://www.applanix.com/products/dms-uavs.htm, a $15K IMU and more expensive L1 and L2 GPS receivers etc.

I'd love to have a budget that could afford either that or a PulseAero solution. But we work with what we can afford and make the best of it.

  • Like 1

Share this post


Link to post
Share on other sites

Here is a recommendation I got for the workstation-

12 core CPU (Fast)

16 or 32 GB of system RAM (DDR 4 or higher)

NVIDIA Graphics card with at least 4 to 8 GB of RAM on the card

At least 500GB to 1TB SSD as your primary drive and then a secondary 2 to 4TB SATA drive to dump to for backup, etc.

Share this post


Link to post
Share on other sites
2 hours ago, Av8Chuck said:

Some things to keep in mind and that you might want to check out.  Take a look at this drone: https://www.microdrones.com/en/mdaircraft/md4-1000/  This drone costs approximately $65K  I'm suggesting that you buy it but that's what it costs to get the level of accuracy required to certify aerial data.  In part, because it uses instrumentation like https://www.applanix.com/products/dms-uavs.htm, a $15K IMU and more expensive L1 and L2 GPS receivers etc.

If your career or your customer's career depends on your ability to certify aerial data then you need to seriously consider whether a drone with off the shelf 5hz GPS, an IMU that was developed to be used in a cell phone and a sub $2K "RTK" base station will provide you with the data you can trust.

Also, the majority of the Photogrammetry software is based on on the same white papers from about 15 years ago which is why most of its all the same.  There are primarily two types, Pix4D which is predominantly used for NADIR and ContextCapture which is used for Oblique scanning.  One is better for farming the other is better for engineering.   

Here's a scan we did of Glendale http://realitymodeling-pw.bentley.com/a3D/Construct Timelapse/models/4-12/App/?scene=http://realitymodeling-pw.bentley.com/a3D/Glendale/Production_1/Scene/Production_1.3mx#%2F  using ContextCapture.  This is a project where a dignitary was going to speak and law enforcement wanted to determine snipper angles, parade routes, evacuation routes, etc., so certification of the data wasn't required.  For the amount of space and the vertical nature of the scan, NADIR would not have been nearly as effective as Oblique. 

I'm not a civil engineer so I know very little about most of the terrestrial based tools, but we've been working with Bentley Systems and a few large engineering firms and have tested a ton of "solutions."  As much as people want to believe that things like SiteScan, DroneDeploy, and DJI have they answer if you want certifiable data, they aren't.   

Depending on what problem you're trying to solve do you really want to potentially jeopardize your career using a 16MP camera, carried on a toy and controlled by a cell phone app?  As much as the idea of cloud computing and having the model rendered before your back to your desk sounds sexy, there are way too many holes in that value chain that you should not trust. 

 

Impressive model!  Do you know if there is any downside to using ContextCapture in nadir mode of data collection?  I am looking more for a both hands on wheel type of a software solution as opposed to a black box that does some of the decision making on its own.

Share this post


Link to post
Share on other sites
34 minutes ago, Brett Baker said:

Here is a recommendation I got for the workstation-

12 core CPU (Fast)

16 or 32 GB of system RAM (DDR 4 or higher)

NVIDIA Graphics card with at least 4 to 8 GB of RAM on the card

At least 500GB to 1TB SSD as your primary drive and then a secondary 2 to 4TB SATA drive to dump to for backup, etc.

We have a 12 core, 64GB RAM, NVIDIA and fiber channel attached RAID.  The Glendale scan was approximately 1700 images and it took 3-days to render low res and about 5-days for the final render.  We tried twice to use Pix4D and it failed each time at 61%.  I'm guessing the version we were using had a memory leak.

51 minutes ago, Brett Baker said:

Thanks for the great info!  Microdrones have already been out to visit us and collected some test data.  They are going to return soon and do it again.  We are also looking at PulseAero Vapor55.  Yeah, the money adds up fast.  Our driving force is safety, deformation monitoring.  We have several earthen dams that don't meet today's standards for seismic stability.  One straddles a creeping fault (14mm/yr).  I spoke with a gentleman from southern Utah who told me he prefers ContextCapture over Pix4d.  I think he said he did some comparisons between the two and found he got better results with the Bentley product.   We will probably base most of our work on lidar data but we also have projects where the imaging is required.  For the LiDAR we are looking at the Reigl VUX-1.

1

What analytical software are you using?  If you use Bentley products the real benefit with ContextCapture is the integration and how you'll be able to use the data in the future.  We've been working on projects with Bentley to be able to blend NAIDIR with Oblique data and Oblique data with LiDAR.    

I was very skeptical of the MD1000, but it's just fascinating to watch when it's on a mission.  Have you seen any of the data?  It's remarkable at how accurate it is without ground control points.  However, there is a downside to that drone, you're putting $64K into the air and its a quad.  There is no hardware redundancy so any problem whatsoever with the power train and it's worthless.

1 minute ago, Brett Baker said:

Impressive model!  Do you know if there is any downside to using ContextCapture in nadir mode of data collection?  I am looking more for a both hands on wheel type of a software solution as opposed to a black box that does some of the decision making on its own.

We just completed scanning Union Station, I can't share that yet but it was for construction purposes so it doesn't cover quite as much area but it's much sharper and more accurate.

You're on the pointy end of the stick and clearly, you understand that whatever solution you choose it won't be based on a hobby grade solution.  Even thought eh MD1000 is a quad, when you see the data and how it collects it you understand why it costs so much.  With enough battery, the Riegl exceeds the 55-pound limit.  I believe they've come out with a smaller version but also you evaluating this technology just as Flash-LiDAR is starting to become available.  

 

  • Like 1

Share this post


Link to post
Share on other sites
14 hours ago, Brett Baker said:

Here is a recommendation I got for the workstation-

12 core CPU (Fast)

16 or 32 GB of system RAM (DDR 4 or higher)

NVIDIA Graphics card with at least 4 to 8 GB of RAM on the card

At least 500GB to 1TB SSD as your primary drive and then a secondary 2 to 4TB SATA drive to dump to for backup, etc.

Decide on your software first and then build a system to run it (just a thought). There are a lot of GPUs on the market that will handle the job. You need to know which side of the hardware end the software is going to run on for most of its heavy-duty processing. Pix4D leans more towards the CPU while Correlator3D requires the GPU to carry most of the load. Some balance the act between the CPU and GPU. So the software is ultimately going to drive your decision on what you need for computing power.

My work desktop will process a medium sized TIF in about 1 1/2 hours in a graphics conversion. The workstation will do it in about 5 minutes. For the image processing end of the game (where you are going to spend most of your time other than paperwork documenting your flight, managing your program, and trying to beg for more money for better toys), as Chuck said in a post large jobs are going to run for days. Even our 300-500 image jobs run for 4-5 hours. That is time you could be flying and generating more data or income. Fast is good. Faster is better. Lightspeed is where you want to be. Time is money.

  • Like 1

Share this post


Link to post
Share on other sites
On 8/24/2017 at 0:53 PM, Av8Chuck said:

 

Here's a scan we did of Glendale http://realitymodeling-pw.bentley.com/a3D/Construct Timelapse/models/4-12/App/?scene=http://realitymodeling-pw.bentley.com/a3D/Glendale/Production_1/Scene/Production_1.3mx#%2F  using ContextCapture.  This is a project where a dignitary was going to speak and law enforcement wanted to determine snipper angles, parade routes, evacuation routes, etc., so certification of the data wasn't required.  For the amount of space and the vertical nature of the scan, NADIR would not have been nearly as effective as Oblique. 

 

Chuck,  you say 1700 images for this model.  What elevation and area scanned?

Share this post


Link to post
Share on other sites
2 hours ago, Dave Pitman said:

Chuck,  you say 1700 images for this model.  What elevation and area scanned?

I believe we were at 400' AGL.  We did concentric circles from a small park with a 10 story white building next to it in the middle and worked our way out from there.

Share this post


Link to post
Share on other sites
2 hours ago, Av8Chuck said:

I believe we were at 400' AGL.  We did concentric circles from a small park with a 10 story white building next to it in the middle and worked our way out from there.

Ok, thanks.  It looks like a nice model for it's intended purpose.

Share this post


Link to post
Share on other sites
5 hours ago, Brett Baker said:

Chuck could you give me your thought process for choosing ContextCapture?  I've had another gentleman recommend it to me.

Thanks!

Brett  

Hi, Brett, we didn't actually choose it, it chose us.  We're working with Bentley Systems to help figure out workflow and accuracy issues using aerial photogrammetry.  We're a developer, not a service provider, this project wasn't a part of our development, we did it on short notice to help out.  

Along with Acute 3D [ContextCapture], we've tested Pix4D and Autodesk ReCap [and a couple of others] and ContextCapture produces the best 3D Mesh.  Not sure if the best 3D mesh makes it a better solution?

2 hours ago, Dave Pitman said:

Ok, thanks.  It looks like a nice model for it's intended purpose.

We just finished scanning Union Station which doesn't cover as much area, it's about 7x7 miles but much more accurate.  It took about four times the amount of time to get the accuracy.

Share this post


Link to post
Share on other sites
On 4/2/2017 at 5:06 AM, Av8Chuck said:

That's certainly been our experience.  

I just get a chuckle when you use "good quality" in the same sentence as DJI and 3DR...

The biggest grumble I hear about surveyors is their perceived (by their customers) reticence to adopt new technology and deliver the same old same old, justifying this by legislative acts.  I am a surveyor, an hydrographic surveyor, we have had to adopt technology for remote sensing the bottom of the ocean and remotely / autonomously piloted underwater vehicles to meet our clients needs.

Anyway credentials established.

The DJI consumer drones (Phantom 4 series)  represent a remarkably capable and highly integrated platform for aerial mapping at high resolution. I think we can all agree with that. At 60m having a 2cm pixel size is awesome.  L1 RTK GNSS systems such as the Emlid Reach have massively reduced the cost of GCP setting, but still need care and knowledge to do it right (auditably and verifiably), but you can do it without GCP's if you mount the Emlid Reach unit on the drone and post process the kinematic position. This reduces position uncertainty to well below the metre mark.  In fact surveys with the Reach on the drone have shown up an interesting feature of the DJI altimeter, as the platform warms up the drone drops two to three metres whilst reporting that it is maintaining altitude. Something you would not know without a separate sensor! This explains those not quite right sloping DEMS that people produce, but also points to a way to use a $1000 drone for 'professional' work.

 

RTK_alt_vs_DJI_alt.png

Share this post


Link to post
Share on other sites
12 hours ago, Spatial Analytics said:

The biggest grumble I hear about surveyors is their perceived (by their customers) reticence to adopt new technology and deliver the same old same old, justifying this by legislative acts.  I am a surveyor, an hydrographic surveyor, we have had to adopt technology for remote sensing the bottom of the ocean and remotely / autonomously piloted underwater vehicles to meet our clients needs.

 

Most of the local shops are on-board with adapting new technology. Being legally liable for their products though, they do thoroughly test the concepts before leaping blindly...makes more fiscal sense.

Share this post


Link to post
Share on other sites

Interesting how this thread seemingly started out about the legality of surveying done by pilots that aren't licensed surveyors to contrasting the capabilities of various hardware and software solutions.

It would seem to me as long as the pilot's work products are not being used for legal purposes (drawings being filed with the County) the pilot and the client are free to agree on whatever level of accuracy they want to.  The pilot better then be prepared to verify they achieved the desired accuracy before sending their work product(s) to the client.  

I think there's a sizeable market of potential clients that want a work product for various in-house purposes that isn't providing "survey grade" precision but might be used for estimated measurement purposes that has a level of precision better than Google Maps for example.  I would agree however that probably 99% of the pilots out there do not have the skills and/or equipment and/or knowledge to provide centimeter accuracy for engineering purposes.  

You could consider drones another tool in the surveyor's bag of tools but I think it's a much more "high maintenance" tool than surveyors might be used to using.  Wait until you have a firmware update on the drone or software update on the pilot app that you have to spend lots of time troubleshooting, not to mention the mechanical maintenance/repair of the drone, the FAA regulations you have to deal with, weather considerations, etc.  I think aerial imaging for whatever purpose (including surveying) is quite a different discipline than land surveying.  Does the electrician also have a plumber's wrench in his bag so he can also handle the plumbing?  Does a construction company save money by paying for a full-time employee to learn/maintain/troubleshoot/fly an in-house drone versus just paying an outside pilot on an as-needed basis?  I'd venture to say NO in the majority of situations.

I had a meeting with an Engineering firm that when I asked them if they had done anything in-house with drones, all their faces turned red and there was some mumbling about a crash and the police.  Suffice to say they no longer had an in-house drone.

 

 

Share this post


Link to post
Share on other sites

There are several factors being discussed.  At this point in time the debate is more about the standards than the laws.  Plus most of us aren't out to prove this can be done at any costs but can we build a business that can provide these services profitably.  

So the level of accuracy is important.  If you start a business where the quality of your data is less than it can be for a reasonable cost then you and your customer are at a competitive disadvantage. If there's an established standard and you can meet that standard regardless of time and costs then there's real value in the data and the customer is less likely to be at a competitive disadvantage.  

 

Share this post


Link to post
Share on other sites
On 7/21/2016 at 4:38 PM, Beacon Sky Survey said:

Anyone ran into problems for not having a land surveyor state license? 

Providing aerial surveying can sometimes replace boots-on-the-ground topographic surveying. I expect there will (already been) push back that UAV data are not as accurate as ground measurements. For example, would a client prefer state licensed ground crew over a newbie UAV surveyor? How do you justify UAV other than lower cost? 

Thanks, 

Kam

There is a difference between surveying and mapping. Surveys are performed and stamped by a registered public land surveyor (RPLS). Mapping, regardless of the degree of accuracy, is not. There is more to surveying than flying a tract of land or job site. We perform mapping services while our current equipment falls somewhere between survey grade and map grade. I cannot pass that off as survey grade because I do not have the background, the education, or the state stamp to do so. So the data I generate will not stand up in court. It is basically as our disclaimer states, "for informational purposes only..." and should not be taken as gospel for design and construction. Even though it is accurate to within about 1/2".

So in answer to your question, a client should go to a state RPLS if they need survey grade data period. If they just want to generate time-lapse documentation of a job site or something similar where accuracy is not an issue, then anyone who can provide accurate map grade data will do. If you do not have a state issued license to act as a RPLS, then you are not a surveyor; you are a mapping technician.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now




  • Welcome to UCCF.

    The UAV Coach Community Forum is actively moderated by the UAV Coach team and offered to help serve those in the UAV industry. Use this space to meet and greet, to ask and answer questions, to share what you're working on, etc. Have fun, play nice, and fly safe! :)

  • Forum Statistics

    • Total Topics
      2,618
    • Total Posts
      13,104