Support Website Contact Support Blog

FLIR Duo Pro R

Dear Reto!

This is the link to the pix4d files (p4d, log and report pdf) https://we.tl/yB45ynXJH3 . Unfortunatelly pix4d fails to upload it to the cloud - we reported that already to support.

In essence we import the R_jpg  pictures, what is split into rgb (jpg) and thermal (tif) by pix4d, we then disable the jpgs and only enable the “grayscale” tifs. Furthermore we only allow pictures taken above 400 m, so pictures near the ground are disabled.

We then select the thermal template and select Tau2 13mm, where we manually reset distortion parameters to be 0.

I had some success with applying alternative processing vs. standard, but I have no conclusive opinion about the step mentioned by Ina (“After changing the pixel size run step 1 and then if the difference between optimized and initial internal camera parameters is high then opt for All Prior.”).

Any of your insight would be greatly appreciated!

If you need more raw data I would be happy to share - we have a 37 km x 400 m corridor captured!

Best Regards,

Gabor and Balint

Hi Gabor

I can’t see your latest post here, strange …
It might be easier to use e-mail, for subsequent conversation, my e-mail address is reto.weiss (at) pixmap.ch.

Regards
Reto

 

Hi Gabor,

Did you have a chance to check the other post related to cloud upload?

The R_jpeg and the RGB_jpeg should be processed individually and after step 1 has been completed merge them. See the instructions here.  

Regarding the difference in optimisation; to asset it, you will have to check the quality report (see example below). If the camera optimisation is higher than 5% then apply All Prior. 

Cheers,

Ina

Hi there Gabor, 

Is it possible to plan a grid flight plan and take both images (RGB/thermal) at the same time and location? I noticed you are using a plane drone from the flight plan so it is possible that you’re not using pix4dcapture… how did you dealt with the flight planning? how are you managing sidelaps and front laps?

 Does pix4dcapture support the communication with the two sensors at the same time? 

We are planning on buying a FLIR DUO PRO Rl for solar pv inspections. 

Regards,

MSc. Pedro Penedos

1 Like

Hi Pedro!

We are designing our flight plans using a custom script developed in house. But it is quite easy to plan a grid flight plan even in Mission Planner or in ugcs. I think you want to set the Flir sensor to take pictures at every 1 sec or every 5 m (let’s say) and than do the flight independent from triggering the camera.

Best Regards,

Gabor and Balint

Dear Ina!

If I opt for optimise All Prior, what do I loose in terms of quality, if I am only interested in ortho and not in the DEM.

Thanks,

Gabor and Balint

Hi Ina / Pix4D

I ordered a FLIR Duo Pro R 640 as well, in a few weeks I will start to process RGB+IR images using Pix4Dmapper.

Pix4D, please, provide a proper calibration model for the camera rig, not only the two separate sensors. This would make a big difference, I believe.

So, how can I vote for this to happen as soon as possible?

Regards
Reto

Thanks for the reply,

I am going to order one as well, a FLIR duo pro r, in the next weeks.

Ok so if i understood Gabor, if I want to plan a grid flight plan with defined over/sidelaps for the thermal sensor, triggering both cameras Pix4d capture can’t help me? how would you recommend building these grid flight plans? I am planning on using it on a m600 from dji.

I am also trying to process your latest dataset (thanks) to explore the workflow, but i am having trouble doing what pix4d recommends (merging option), I was capable of merging the two datasets after processing step1 but when i try to process the 3 step it gives an error. Now i and trying to process a second time but doing the step 2 first and only than, what i am interested, in the thermal and rgb ortomosaic (step3). 

Cheers,

ppenedos@icloud.com - Please feel free to begin a conversation

 

Hi all

I’m also trying to process Balint/Gabor’s dataset posted May 1st. However, in the zip archive referenced I only find 1738 _R.JPG images, no RGB at all. 

@Pedro: You refer to two datasets processed, where did you find the RGB images?

Best
Reto

Hey, From what i tried, the photos are converted by pix4d and it splits the two. check once you add the photos, go to project, Image properties editor. 

 

Cheers

I am ordering one as well, for thermal and rgb ortomosaicking.

Yes, of course, you are right. It’s a little bit like magic.
Which initial processing parameters did you use, anything special or one of the templates Pix4D provides?

Reto

Hey, no problem, all i know so far is here. it sais to use a thermal template in the step1. I am lacking information what to do after merging the two, first the report from the merging operation show red sign on calibration. And I am now processing step 2 and 3 but i am only interested in the results from 3. My objective is a thermal ortomosaic and a rgb ortomosaic.

Pedro

Thanks Pedro for the hint. I will try do “the same”. I’m yielding at exactly the same deliverables, a thermal and a rgb orthomosaic, both properly aligned and georeferenced. More to come.

1 Like

I now created two projects,

  • one with rgb images enabled only and
  • one with ir/grayscale images only

The setup for step 1 is almost the same, the only difference I see is calibration method (Standard for RGB, Alternative for IR).
I will process the two data sets now, I guess it takes a while since the sets are pretty big.

Thanks everybody for your valuable input.
Reto

1 Like

I procssed both projects, RGB and IR, the results of step 1 do not look very promising, expecially on RGB side:

  • IR:     Dataset 1676 out of 1738 images calibrated (96%), 1738 images disabled, 12 blocks
  • RGB: Dataset 705 out of 1738 images calibrated (40%), 1738 images disabled, 16 blocks

@Pedro: Are these the same results you had, Pedro?
@PIx4D: Any explanations for the poor calibration results, especially on RGB side?

Attaching the processing results.

The reports after step 1 cam be found here: https://drive.google.com/drive/folders/1dDRJ4gshrl8QdFv6frAqSOZHLb7xXLwV?usp=sharing

My results were also not satisfatory. not the same but close. https://drive.google.com/file/d/1bZCYinG3-U45xaaWtZwXpDwf9qFWDQ4x/view?usp=sharing

https://drive.google.com/file/d/1oREMZ1Q4bwjlLdSnQ71EHDPpGqHfbo3S/view?usp=sharing

 

did you used any template for step 1?

Hi Ina

Can you help us?

Regards
Reto

Dear Reto, Dear Pedro,

We have not tried combining RGB with IR but the pure thermal datasets look better after we processed them:

F1: https://drive.google.com/file/d/1EgAKVouWjJBMkgTcYgza6rpqvB5vKszd/view?usp=sharing (this is the dataset we shared with you)

Another set: https://drive.google.com/file/d/1r-kuljN7lOK3_Eq-AIlJwaeOEesGVlzF/view?usp=sharing (same day, just slightly different location)

 

Best Regards,

Gabor

I found a few error on the sensor width on both sensors the value is wrong. i am processing again from 0. thanks for the new data set. have you been trying to improve overlap and side lap by planning a grid flight plan?