All--

Don't buy a P4 Multispectral drone unless DJI starts selling it with or makes available a reflectance panel specifically designed for their sensor. Without a reflectance panel there's no way to calibrate the sensor ranges to the atmoshperic conditions during the collection periods. Hence, there's no way to accurately compare the spectral ranges and indices of the same area or vegataion at different periods of time. We spent $6,000 for a piece of junk. DJI needs to rectify this issue immediately!

Unhappily yours,
Keith VanDerSys

Hello, fansd2c5cb68. We have forwarded your advice to our engineers and bring the newest answer for you.
1. Generally, we can use the fusion value of several sensors (camera and the spectral sunlight sensor on top of the aircraft) to obtain the reflectivity,  so as to calculate the NDVI and other indices.
While performing reconstruction with P4 Multispectral photos, DJI Terra will obtain the reflectivity based on the camera’s DN value (reflection intensity) and the incident intensity obtained by the spectral sunlight sensor, which will then provide the reflectivity so as to calculate the NDVI and other indices. This can be used to make a real-time radiometric calibration. Use DJI Terra to build index diagrams without using reflectivity. With this method, relatively uniform reflectivity values can be obtained at different times to output a relatively correct index diagram.
2.  On 11th  December 2020, DJI released DJI Terra v.2.3.1, adding the Radiometric Correction function.  As long as acknowledging reflectivity, users can choose any reflectance panel, which can be sourced from some third-party manufacturers, such as Labsphere, Zenith, etc.
If you have any questions, please feel free to talk!

Thank you for the response.

DJI's advice that owners “collect the radiation calibration data with third-party radiometric calibration board” is a completely unacceptable solution. Calibration panels are made specifically for particular sensors. Hence, the sensor ranges for other cameras will not match those of the DJI camera. The Parrot Sequoia+, for instance, has different sensor ranges for the red, red edge, and nir sensors; It also does not have a blue band. Additionally, MicaSense will not sell calibration reflectance panels to buyers without owning a MicaSense camera. I was told by the MicaSense representatives that they stopped selling panels because DJI customers were depleting their panel stock. While MicaSense may have the closest acceptable alternative, DJI P4 owners cannot purchase panel substitutes.

Again, it is completely unacceptable that DJI would develop and sell a product that is unusable!

Very unhappily yours,
Keith VanDerSys

Hi any updates on this?

Hi, I am using the procedure described by the DJI moderator above, on Metashape, there is no problem. No need for a calibration panel. It gives satisfactory results, quickly, for $ 6000 I find it good. If you need more precision I can only recommend that you buy a hyperspectral camera. The price is not the same!
regards

Hi everyone,
I need to calculate spectral radiance and irradiance separately for each band of phantom 4 multispectral. Currently, I am able to calibrate the images using a reference panel and get the reflectance image (which seems to be reasonable). But when I use the light sensor data, the values I get do not make any sense. For example, in the image of a standard panel, I got 0.26 for blue, 0.20 for green, 0.15 for red, 0.11 for red edge, and 0.12 for nir bands, while I was expecting to get 0.26 for all bands since the panel is standard fixed 0.26 for all bands.   So I need to get radiance and irradiance separately to see what the problem is. In the image processing guide of the DJI for phantom 4 they have explained how to calculate NDVI, but no mention on how we can get radiance and irradiance for each band- they have mentioned similar notations but it doesn't work as I mentioned above.  Any help is highly appreciated.

Any news???? I bought a P4 Multispektral and I dont find any reflectance Panel designed for it!

DJI needs to make and sell/give free to current owners a calibration panel for the P4 MS.

hamid39 Posted at 10-8 20:17
Hi everyone,
I need to calculate spectral radiance and irradiance separately for each band of phantom 4 multispectral. Currently, I am able to calibrate the images using a reference panel and get the reflectance image (which seems to be reasonable). But when I use the light sensor data, the values I get do not make any sense. For example, in the image of a standard panel, I got 0.26 for blue, 0.20 for green, 0.15 for red, 0.11 for red edge, and 0.12 for nir bands, while I was expecting to get 0.26 for all bands since the panel is standard fixed 0.26 for all bands.   So I need to get radiance and irradiance separately to see what the problem is. In the image processing guide of the DJI for phantom 4 they have explained how to calculate NDVI, but no mention on how we can get radiance and irradiance for each band- they have mentioned similar notations but it doesn't work as I mentioned above.  Any help is highly appreciated.

Hello, there. Sorry for the late response. The ISO value captured by the light sensor of Phantom 4 Multispectral varies from each band, utilizing which to directly output the reflectivity is inaccessible. The consequent value calculated will not render users a tenable reference. If you want to get the reflectivity of each band, we would recommend you use reflect panel, refer to the procedure of radiometric calibration. That is, calculate the reflectivity by using the DN value of photos.

DJI Wanda Posted at 11-2 04:27
Hello, there. Sorry for the late response. The ISO value captured by the light sensor of Phantom 4 Multispectral varies from each band, utilizing which to directly output the reflectivity is inaccessible. The consequent value calculated will not render users a tenable reference. If you want to get the reflectivity of each band, we would recommend you use reflect panel, refer to the procedure of radiometric calibration. That is, calculate the reflectivity by using the DN value of photos.

Have you a link to radiometric calibration procedure with reflectance panel  ?

DJI P4M is a piece of ..., $6000 for a box of waste, nothing can be done without Terra, even if there is a reflectance panel, no documentation. The image processing guidelines only calculate indices like NDVI and don't mention reflectance at all. Does DJI think customers don't need reflectance? Don't buy DJI P4M.

I still underestimated Terra and DJI. It turns out that Terra cannot perform radiometric correction on a single image. Even if the picture of the entire task is corrected, it will not provide you with the generated reflectance, but will give you indices such as NDVI. I even have reason to think that this radiometric correction is a perfunctory function. Really regret buying DJI products

Forgot about TERRA, try with QGIS. P4M images are very good for 6000$ drone included.

when starting the drone Phantom 4 multispectral, error application:unable to take off. mission cannot be performed in your area, height limited

Maulit.Almas Posted at 12-22 21:54
when starting the drone Phantom 4 multispectral, error application:unable to take off. mission cannot be performed in your area, height limited

https://www.dji.com/lv/flysafe/geo-map

LV_Forestry Posted at 12-22 22:24
https://www.dji.com/lv/flysafe/geo-map

Мы находимся нормальной зоне, даже я активировал синий и желтую зону все равно выходить этот ошибка

Hi everyone.

Which is the order of bands in saved images?
Example I have for the same shot 6 images: DJI_1160 - DJI_1161 - DJI_1162 - DJI_1163 - DJI_1164 - DJI_1165
The first one is in jpg and corresponds to RGB sensor. What about the other? Are they stored in order of wavelength or following other storaging rules?
I need to know the bands for each image

Thanks

@djiuser_T6ZJTMQc4TXW--

Correct. The images are in order of the spectrum wavelengths: Blue, Green, Red, REG, NIR--see attached pic.
k.

Hi there, so, I just bought mi ph4m please don't tell me there is no way to get good results without micasense panel!!!
I've read the main issue is to compare different campains (different flights).

@ Capsula NT--

Correct. The reflectance values can't be compared between flights without a calibration panel. Micasense won't help you solve the issue since they won't sell panels without the purchase of their camera, however. We instead purchased a 5" Permaflect target with a 50% absolute reflectance for about $800 from Labsphere. At present, it's the only solution since DJI didn't (and won't) provide a calibration panel in the purchase. Here's a Labsphere link: https://www.labsphere.com/product/permaflect-targets-rigid-lidar-commercial-and-custom/.

Capsula NT Posted at 3-7 23:19
Hi there, so, I just bought mi ph4m please don't tell me there is no way to get good results without micasense panel!!!
I've read the main issue is to compare different campains (different flights).

Hi,
I don't want to get into a reflectance target user vs solar sensor user conflict, I have stopped using reflectance targets for crop monitoring with P4M. Agisoft Metashape manages the P4M solar sensor very well. That's my point of view, do what you want with that. But $800 for a LiDAR target seems a bit excessive to me. I have a Resonon camera that also uses this type of reflectance target, and quite honestly the results are comparable to the P4M. $800... buys batteries.

If someone wants clarification or just to discuss it, there is no problem, but no conflict. I have deep respect for reflectance target users I am already convinced that they bring something in the results, but not enough.
Test with a piece of teflon before spending 800$ ;)

Hi everyone,
I'm using the information in the P4 Multispectral Image Processing Guide to calculate approximate reflectances. The guide mentions that "All bands are calibrated against the standard NIR band". There is a calibration parameter "Rho NIR" that apparently must be applied to all the bands to obtain reflectances. I'm not sure how to obtain this parameter from the image metadata. Can anyone give me a hint on this?

Thank you!

LV_Forestry Posted at 3-13 12:50
Hi,
I don't want to get into a reflectance target user vs solar sensor user conflict, I have stopped using reflectance targets for crop monitoring with P4M. Agisoft Metashape manages the P4M solar sensor very well. That's my point of view, do what you want with that. But $800 for a LiDAR target seems a bit excessive to me. I have a Resonon camera that also uses this type of reflectance target, and quite honestly the results are comparable to the P4M. $800... buys batteries.

Hi,
good to hear at least someone is happy with his purchase. I have the P4M as well. What if I don't want to compare fields at all?
I use my P4M to
1. check the results of tests in fields (ex. one part of the field has received a fertilizer, the other has not)
2. generate prescription maps for precision spraying/spreading with the T30 (which i'm about to test very soon, the process of generating prescription maps and importing them on the T30 is a bit tedious)

So my question is: if I use the P4M to only gather differential data, do I need to calibrate? or use radiometric correction?

Arnaud LRC Posted at 6-9 23:16
Hi,
good to hear at least someone is happy with his purchase. I have the P4M as well. What if I don't want to compare fields at all?
I use my P4M to

Hi,
I do the same kind of work. On the same plot there is a control area which did not receive any fertilizer or other product. Depending on the action of the sprayed product the difference should be visible. Even without calibration. Multispectral imaging is very subjective, in the laboratory it is certainly possible to obtain very precise data with good quality cameras and controlled light sources, for a single leaf. With the drone it's different, the light is changing, it may not be evenly distributed over a single image. I strongly encourage using filters, for example height to remove bare ground. But still discrimination by spectral index on neighboring trunks or crops. If you need experience sharing, don't hesitate to ask. I have some skills in French language. It can also benefit me, sometimes that's how I realize that I'm going in the wrong direction.

LV_Forestry Posted at 6-10 12:59
Hi,
I do the same kind of work. On the same plot there is a control area which did not receive any fertilizer or other product. Depending on the action of the sprayed product the difference should be visible. Even without calibration. Multispectral imaging is very subjective, in the laboratory it is certainly possible to obtain very precise data with good quality cameras and controlled light sources, for a single leaf. With the drone it's different, the light is changing, it may not be evenly distributed over a single image. I strongly encourage using filters, for example height to remove bare ground. But still discrimination by spectral index on neighboring trunks or crops. If you need experience sharing, don't hesitate to ask. I have some skills in French language. It can also benefit me, sometimes that's how I realize that I'm going in the wrong direction.

Hi,

I didn't realize you had replied. Would you be interested in having a chat via telephone (or whatsapp as a cheaper alternative)? I would be interested to hear what you are doing exactly. Maybe we can indeed help each other out. I'll send you my phone number in a private message.

Hope to hear you soon.
Regards, Arnaud L

Hi, yes we can arrange that. See you in private message.

Hi,
We have been using a MicaSense RedEdge - M underslung on a Phantom 4 RTK for environmental. We use a sunshine sensor (irradiance sensor) and a reflectance panel to calibrate our imagery, and try and keep it consistent. We use it to identify algae in interdal reef, and weeds in native vegetation.

We can't get the MicaSense RedEdge - M anymore, so we're looking at the P4M as an alternative. I am concerned we won't get calibrated bands with the P4M as it doesn't' have the sunshine sensor, even if we use the calibration panels we have from micasense.

this thread seems to suggest that the P4M somehow uses the ISO of the RGB image to calibrate irradiance. Has anyone tried this with a calibration plate, and have you had any success?

Thanks

ballan Posted at 7-25 17:09
Hi,
We have been using a MicaSense RedEdge - M underslung on a Phantom 4 RTK for environmental. We use a sunshine sensor (irradiance sensor) and a reflectance panel to calibrate our imagery, and try and keep it consistent. We use it to identify algae in interdal reef, and weeds in native vegetation.

Hi, sun sensor is mounted on top of the P4M. Each picture, each band have itself calibration value.  

LV_Forestry Posted at 7-26 01:01
Hi, sun sensor is mounted on top of the P4M. Each picture, each band have itself calibration value.

Thanks for the clarification!

We use Pix4D for processing, and I have found instances of people using the Micasense or Sentera calibration plates with the P4M.
https://sentera.shop/collections/accessories-parts/products/reflectance-panel

https://micasense.com/shop/Calibrated-Reflectance-Panel-2-p467113700

However, I can't find any records of the radiometric correction values for the different bands. Is anyone using one of these correction plates, and if so, what known values are you using for each band?

Thank you

ballan Posted at 7-26 18:44
Thanks for the clarification!

We use Pix4D for processing, and I have found instances of people using the Micasense or Sentera calibration plates with the P4M.

I will surely create a conflict by writing this. Each multispectral camera has its own calibration file. It is not enough to buy a reflectance target from a third-party manufacturer. I have a Tetracam ADC micro multispectral camera, a Resonon PIKA-L hyperspectral camera. And for having done the comparative test, the results are very good using the P4M solar sensor. Of course not totaly identical, btw also not identical between ADC micro and PIKA-L. The advantage of the P4M is that I don't need to use this white plate. Keep in mind that taking multispectral images from a drone is not the same as taking them in a lab. The crops move with the wind, the cloud cover if there is any, vary the contrasts...
The objective of the P4M is modest (according to its price) it is to do crop monitoring. If it is to do science, it is better to raise the budget and buy a more powerful camera. Otherwise you have to be content with P4M and interpret the results which I recall are very good whatever people say.You will find the correction values in the meta data of the images.

LV_Forestry Posted at 3-13 12:50
Hi,
I don't want to get into a reflectance target user vs solar sensor user conflict, I have stopped using reflectance targets for crop monitoring with P4M. Agisoft Metashape manages the P4M solar sensor very well. That's my point of view, do what you want with that. But $800 for a LiDAR target seems a bit excessive to me. I have a Resonon camera that also uses this type of reflectance target, and quite honestly the results are comparable to the P4M. $800... buys batteries.

Hi LV_Forestry
First of all, thanks for sharing your knowledge. Your ideas, comments and knowledge have helped me a lot. I've also been doing drone mapping for a few years but only ever in RGB. Multispectral and the PH4M are new to me. I have currently taken pictures of a forest. My problem now is that when I do the Calibrate Reflectance with "use sun sensor" in Metashape, all the images are just black afterwards. Do you have any idea what this could be and can I still process the ortho. Or can I simply brighten the images again after the Calibration Reflectance by "Set Brightness". Does that make sense and does that give meaningful results.  Thank you very much and best regards.
Markus

Staudi Posted at 7-27 06:53
Hi LV_Forestry
First of all, thanks for sharing your knowledge. Your ideas, comments and knowledge have helped me a lot. I've also been doing drone mapping for a few years but only ever in RGB. Multispectral and the PH4M are new to me. I have currently taken pictures of a forest. My problem now is that when I do the Calibrate Reflectance with "use sun sensor" in Metashape, all the images are just black afterwards. Do you have any idea what this could be and can I still process the ortho. Or can I simply brighten the images again after the Calibration Reflectance by "Set Brightness". Does that make sense and does that give meaningful results.  Thank you very much and best regards.
Markus

Hi,
After choosing the solar sensor as the reflectance calibration, you can brighten your images by clicking as you wrote on Set Brightness -> estimate -> OK.

This is purely visual and does not affect the data. But I do it every time to avoid having a completely black point cloud. This helps to understand its orientation for the manual dense point cloud cleaning step before generating the DEM. This is an important step to refine the quality of the orthophoto which is based on the DEM.

You are welcome, if I can help, I do, but be careful, I don't have the exact science. I invite you to document yourself with different sources, compare data, and do what best suits your use.

Is it possible to process P4 Multispectral datasets in Pix4DMapper without a reflectance calibration target?

djiuser_6PTmSM6fLeUu Posted at 8-2 16:51
Is it possible to process P4 Multispectral datasets in Pix4DMapper without a reflectance calibration target?

Yes, you can tell Pix to just use the DLS data. I can't speak to whether that is as good as having calibration target images as well, as I have not tested that with this sensor myself (yet).