Help needed to verify and fine tune a GSD Calculator

[Airshots.uk]

Hi fellow pilots,

I would appreciate a little help from any pilot who can spare a few minutes to help me verify and test a GSD Calculator I have built on my website;

GSD Calculator

If you find any errors or omissions either drop me an email or update this thread.

Be aware that I sourced the data needed (see below if you are adding or correcting) from a variety of sources, including asking ChatGPT to fill in the ones I struggled to find elsewhere. If any search spat out dubious or ambiguous information I simply didn't create an entry for that drone. It grew from a small handful of drones that I was personally interested in to a list of 74 drones including Autel, DJI, Dranganfly, Skydio, Trimble, Wingtra and Yuneec.

The base data needed for the calculations are as follows:

1. Sensor Width (mm)
2. Sensor Height (mm)
3. Focal Length (mm)
4. Image Width (px)
5. Image Height (px)
6. Horizontal Field of View (deg)
7. Vertical Field of View (deg)

I look forward to hearing from many of you out there and will thank you all in advance. I may be a bit tardy in responding as I am a working pilot and I'm in the UK, so time differences will apply to most of you. Many thanks,

Paul

 

[DJ Wes]

Hi fellow pilots,

I would appreciate a little help from any pilot who can spare a few minutes to help me verify and test a GSD Calculator I have built on my website;

GSD Calculator

If you find any errors or omissions either drop me an email or update this thread.

Be aware that I sourced the data needed (see below if you are adding or correcting) from a variety of sources, including asking ChatGPT to fill in the ones I struggled to find elsewhere. If any search spat out dubious or ambiguous information I simply didn't create an entry for that drone. It grew from a small handful of drones that I was personally interested in to a list of 74 drones including Autel, DJI, Dranganfly, Skydio, Trimble, Wingtra and Yuneec.

The base data needed for the calculations are as follows:

1. Sensor Width (mm)
2. Sensor Height (mm)
3. Focal Length (mm)
4. Image Width (px)
5. Image Height (px)
6. Horizontal Field of View (deg)
7. Vertical Field of View (deg)

I look forward to hearing from many of you out there and will thank you all in advance. I may be a bit tardy in responding as I am a working pilot and I'm in the UK, so time differences will apply to most of you. Many thanks,

Paul

I have spent some time gathering sensor sizes for DJI drones for use in my own GSD calculator. While there is no definitive source, I have found several different values provided for sensor size:

• Nominal (this value is sometimes advertised as the sensor size)
• Actual (this value can be computed based on the sensor area and aspect ratio)
• Active (this is the active area of the sensor which is often less than the calculated size)

While there are a number of sources, the best I have found is:

DroneModels/droneModels.json at main · Theta-Limited/DroneModels

Contribute to Theta-Limited/DroneModels development by creating an account on GitHub.

github.com

Your values for the DJI Air 2S do not match any of the values I have found for that drone. I'm curious what your source is for that drone and are you using the nominal/advertised sensor size, actual sensor size, or measured sensor size?

The dimensions you list for the DJI Air 3S imply that it captures photos with an aspect ratio of 3x2. Correct me if I am wrong, but I thought that the Air 3S captures 4x3 images. The image height you list does not match that from the DJI specs.

In your application, the flight altitude is set in feet while the calculated footprint is displayed in meters.

 

[Theta]

@DJ Wes I believe that for the air 2s, we went off the advertised (nominal) size and used a formula to approximate its actual size. We do this for certain models that we can't find much data on.

We may be wrong on the sensor size for the air 2s. The air 2 s does not output altitude metadata in any absolute reference (e.g. above msl), so it doesn't produce accurate results with OpenAthena regardless. Our camera calculation for this drone therefore might not have been very rigorous. The altitude metadata issue is common with older DJI models but is slowly being phased out.

Typically we go off the full pixel width and height of the sensor. We assume that top and bottom may be cropped by the firmware but that the width remains intact (except in photos with digital zoom applied). The Air 2s might have been an odd one with atypical 3:2 ratio.

 

[Eko]

The base data needed for the calculations are as follows

Now I have a third different set of values for the Mini 4 Pro. It is a real headache for many I believe. The best is to be found by trial and error, I guess.

I don't know why all these values and other necessary details are not officially published by the vendor, in the first place.

 

[Airshots.uk]

I have spent some time gathering sensor sizes for DJI drones for use in my own GSD calculator. While there is no definitive source, I have found several different values provided for sensor size:

• Nominal (this value is sometimes advertised as the sensor size)
• Actual (this value can be computed based on the sensor area and aspect ratio)
• Active (this is the active area of the sensor which is often less than the calculated size)

While there are a number of sources, the best I have found is:

DroneModels/droneModels.json at main · Theta-Limited/DroneModels

Contribute to Theta-Limited/DroneModels development by creating an account on GitHub.

github.com

Your values for the DJI Air 2S do not match any of the values I have found for that drone. I'm curious what your source is for that drone and are you using the nominal/advertised sensor size, actual sensor size, or measured sensor size?

The dimensions you list for the DJI Air 3S imply that it captures photos with an aspect ratio of 3x2. Correct me if I am wrong, but I thought that the Air 3S captures 4x3 images. The image height you list does not match that from the DJI specs.

In your application, the flight altitude is set in feet while the calculated footprint is displayed in meters.

Hi DJ Wes,

Many thanks for taking the time to dig deeper and to respond - it is much appreciated.
The Theta Limited JSON file is a useful resource - thanks for that. I have incorporated the data from that and modified my data to reflect the data in that file.

I have no clue where I got the Air 2 data from - I have gathered from so many sources recently I couldn't tell you. But I will be able to answer that in the future as I have now kept a record of the new and updated entries. Have a look again at the AIr 2 data and see if that now matches yours.

Yes the altitide is set in feet - deliberately. But the calculations are all done in meters so there is a hidden calculation that converts feet to meters.

Paul

 

[Airshots.uk]

@DJ Wes I believe that for the air 2s, we went off the advertised (nominal) size and used a formula to approximate its actual size. We do this for certain models that we can't find much data on.

We may be wrong on the sensor size for the air 2s. The air 2 s does not output altitude metadata in any absolute reference (e.g. above msl), so it doesn't produce accurate results with OpenAthena regardless. Our camera calculation for this drone therefore might not have been very rigorous. The altitude metadata issue is common with older DJI models but is slowly being phased out.

Typically we go off the full pixel width and height of the sensor. We assume that top and bottom may be cropped by the firmware but that the width remains intact (except in photos with digital zoom applied). The Air 2s might have been an odd one with atypical 3:2 ratio.

Ah Theta,

Thanks for collecting a ton of data and thanks for allowing me to incorprate the same.

Paul

 

[Airshots.uk]

Now I have a third different set of values for the Mini 4 Pro. It is a real headache for many I believe. The best is to be found by trial and error, I guess.

I don't know why all these values and other necessary details are not officially published by the vendor, in the first place.

Hi Eko,

I feel your pain and agree that if manufacturers published the actual data everything would be simpler. The fact that they don't means data has to be calculated or even estimated in some cases. In those situations I have tried to exclude those drones from my database.

Saying that I recent discoverd that asking ChatGPT can produce data that you can't source elsewhere; for example a request to isolate DJI Mini 4 Pro data results in the following (from Chat GPT)...

Here’s a spec table for the DJI Mini 4 Pro, based on the camera sensor parameters:
Parameter Value
Sensor Width (mm) ≈ 8.96 mm
Sensor Height (mm) ≈ 6.72 mm
Focal Length (actual) 6.72 mm mavicpilots.com+9forum.flylitchi.com+9dji.com+9petapixel.com
Image Width (pixels) 8,064 px
Image Height (pixels) 6,048 px
Horizontal FOV (°) ≈ 67.4°
Vertical FOV (°) ≈ 53.1°


Calculation Details

Sensor size uses data from a community-verified source:

Pixel pitch: 0.00222566 mm, 0.00222138 mm

Pixel dimensions: 4,032 × 3,024 px mavicpilots.com forum.flylitchi.com

Thus: 0.00222566 × 4,032 ≈ 8.96 mm width; 0.00222138 × 3,024 ≈ 6.72 mm height.

Image resolution (8,064 × 6,048 px) is confirmed by DJI specs forum.flylitchi.com+1mavicpilots.com+1
.

Field of view calculated with:
FOV = 2 × arctan(sensor dimension / (2 × focal length))

Horizontal: 2 × atan(8.96 / (2×6.72)) ≈ 67.4°

Vertical: 2 × atan(6.72 / (2×6.72)) ≈ 53.1°

 

[DJ Wes]

I have no clue where I got the Air 2 data from - I have gathered from so many sources recently I couldn't tell you. But I will be able to answer that in the future as I have now kept a record of the new and updated entries. Have a look again at the AIr 2 data and see if that now matches yours.

I can see you changed the sensor size to be the active area of the sensor as @Theta has provided. However, your calculated horizontal and vertical FOV differ from my calculations. I describe how to calculate them here:

Converting Diagonal Field of View (FOV) to Horizontal FOV

Wes Barris

www.litchiutilities.com

Yes the altitide is set in feet - deliberately. But the calculations are all done in meters so there is a hidden calculation that converts feet to meters.

The point I was trying to make is that if someone is entering data using imperial units, they probably want to see all results in imperial units as well.

 

[Eko]

Here’s a spec table for the DJI Mini 4 Pro, based on the camera sensor parameters

I want to send my thanks a thousand times for your valuable work, if that can equate your great contributions. I decided to change all the drone specific camera details on my records, as per your calculations.

Thanks @Airshots.uk