DronePan

What is the best program for stitching of the pics? Thanks!!!
Free and easy is Microsoft ICE, but you need to create an account first to download the Windows only ICE stitching program at this link.
Create your own Photosynths - Photosynth
Then you can upload the finished ICE spherical panorama to Photosynth, where you can keep it unlisted, or publish it. The URL will be ridiculously long, so just go to bit.ly to shorten it for sharing!:cool:
 
  • Like
Reactions: scholls22
1) nothing. it won't work.

2) As I stated above, "60 degrees from each other"

3) we can't. we are testing / adjusting different intervals on a near daily basis.

4) sounds great, thanks!
First, let me apologize for my prior adversarial tone. I'm really on your side and trying to help, but your answers haven't been making it any easier.

1. Something must happen when you initiate Dronepan in my scenario above on the P3P. It may not be what you want, but what does happen is very helpful to solving the problem. Will the first shot be initiated? When will it fail and how?

Heck, even when using the Inspire1 App Store Dronepan version on the P3P, photos are actually taken in my scenario setup!

I even managed to get a 6 shot 360° horizontal 90° elevation panorama out of the Inspire1 Dronepan app from the App Store while connected to my P3P tonight during further testing. That's 1/3 of all the needed images. I even got the gimbal elevation change, but no rotation, as each shot tried to move but shook, rather than rotated from the prior one, and then the gimbal rotated down but repeated the same shot. It's working mostly, even with the Inspire1 code controlling the P3P! I did have to switch to JPG only and close out DJI GO, but it got 1/3 taken!

2. Since the 0° downward images add nothing new to the stitch, it's now a moot point. (I just confirmed this. Your power user is correct!)

3. If we adopt a specific predictable set of camera settings that the user must first enter into DJI GO, we would then have a constant set of parameters that we could predictably time around. Another unknown variable would then be the speed of the mini SD card, which could also eliminated by specifying that a specific brand and speed of mini SD card speed be used, such as the 633x Lexar card included with the P3P. Slower cards write the images slower to the card, and until the camera buffer has cleared enough room to take the next shot by writing to the card, no additional shots can be taken, so shots will be missed whenever the buffer fills up with a slow micro SD card. Do we know the size of the P3P camera buffer? That's an essential element of this equation.

As a baseline, what timed intervals are you currently using for the shutter, the yaw, and the gimbal elevation in the most recent beta that you have distributed for testing?

20 shots in 2 minutes is only 6 seconds in between shots, which likely isn't long enough to complete all necessary changes in between, repeatedly, across multiple users, unless the user has a very unique setup.

There are two additional variables that will wreak major havoc, and they are the user settings in DJI GO for both yaw speed and gimbal elevation speed. If these are changed from the defaults and set to reasonably slow speeds for smooth videography, it would take longer than 6 seconds to yaw 60° or to elevate the gimbal down 30°.

4. Let's start with requiring the user to begin setting up their Dronepan shoot in DJI GO with these settings: Gimbal Elevation set at Horizontal, Single shot mode, JPG only, Auto Exposure, Auto WB, as these settings produce almost instantaneous captures and image saving, and will minimize the total time for testing. (Also, most using your app won't be using DNG or AE bracketing, because they can't be bothered with DNG conversion. If they are willing to do so, they will likely shoot these spherical panoramas manually, as I do, as it really is not all that complicated!)

We just need to automate a known working process!

As I have previously explained, I use 45° intervals of rotation (instead of your 60°), because I can do those easily with the red compass arrow orientation acting like a clock face. More overlap is better for stitching, especially if I miss my mark on any shot, since I am only estimating yaw changes manually by eyeballing them. A shorter yaw of 45° instead of 60° also is more easily completed at a slow yaw speed setting in a limited time in between shots. That gives me three rows of 8 images at each elevation of 90° (Horizontal), 60°, and 30° (I just confirmed that eliminating all 0° images has no adverse effect, so I am eliminating them, too--Thank you for that!). That gives me 24 images in total, with a 22.5° overlap between adjacent images for stitching. With only 6 shots at 60° intervals, you currently only have 15° of overlap for stitching programs, which may not be enough, depending upon the stitching program used, and a 60° yaw may very well take longer than 6 seconds at a slow user yaw speed setting.

1. So, let's set the shooting interval at 15 seconds, starting at time 0, and ending time after 24 shots, at time 0 + 5 minutes + 45 seconds. Set the yaw interval at 15 seconds starting at time 0 + 3 seconds and ending at time 0 + 5 minutes + 48 seconds. Set the gimbal elevation interval (to lower by 30° each time, from the starting specified horizontal elevation) to 2 minutes, starting at time 0 + 2 minutes + 6 seconds, so yaw and gimbal elevation commands are at different times, and ending after time 0 + 5 minutes + 6 seconds. I can guarantee 100% success with this 24 shot algorithm using the exact time intervals I specified! Let's use it as our working model, from which to make changes, to optimize the process to shorten the time intervals and/or the number of images needed.

So, using this as our baseline, let's try two more scenarios:

2. Let's change to your 18 images, using the 60° yaw, while still keeping to the 15 seconds in between images. It will take 90 seconds each to complete the row and prepare for the next, for a total time of 3 minutes and 45 seconds. Begin image 1 of 18 at time 0 and take a new shot every 15 seconds, ending with image 18 at time 0 + 3 minutes and 45 seconds. Set the yaw time interval to 15 seconds beginning at time 0 + 3 seconds, and ending at time 0 + 3 minutes and 33 seconds. Set the gimbal elevation interval (to lower 30° each time, from the original specified horizontal elevation) to 1 minute and 15 seconds, beginning at time 0 + 1 minute and 21 seconds, and ending at time 0 + 2 minutes and 36 seconds.

3. Let's now change to your 18 images at 10 second intervals, using the 60° yaw.
It will take 60 seconds to complete each row and process and prepare for the next image, for a total time of 3 minutes, completing the last capture after 2 minutes and 50 seconds. Begin image 1 of 18 at time 0 and take a new shot every 10 seconds, ending with image 18 shot at 2 minutes and 50 seconds. Set the yaw time interval to 10 seconds beginning at time 0 + 3 seconds and ending at time 2 minutes and 43 seconds. Set the gimbal elevation interval ( to lower 30° each time, from the original specified horizontal elevation) to 1 minute, beginning at time 0 + 53 seconds (or 0+ 55 seconds if yaw and gimbal elevation commands cannot be simultaneously executed) and ending at time 0+ 1 minute 53 seconds ( or 0 + 1 minute 55 seconds if yaw and gimbal elevation commands cannot be simultaneously executed.

So, 3 different scenarios:
1. 24 images, taking 6 minutes to complete, allowing 12 seconds for a 45° yaw, and 9 seconds for a 30° gimbal elevation change
2. 18 images, taking 4 minutes to complete, allowing 12 seconds for a 60° yaw, and 9 seconds for a 30° gimbal elevation change
3. 18 images, taking 3 minutes to complete, allowing 7 seconds for a 60° yaw, and 7 or 4 seconds for a 30° gimbal elevation change, depending upon the ability to yaw and change the gimbal elevation at the same time.

Clearly, scenario 1 should be problem free. 12 seconds for a 45° yaw and 9 seconds for a 30° gimbal elevation change even at very slow user settings for yaw speed and gimbal elevation speed should be fine.

Scenario 2 should also be fine even though 12 seconds for the 60° yaw is 1/3 more movement than a 45° yaw, while the 9 seconds for the 30° gimbal elevation change remains the same.

Scenario 3 is really pushing the limits. 7 seconds for a 60° yaw may be too fast at a slow yaw setting. Nothing rips video frames faster than a yaw speed that is too fast, so most users lower that speed significantly. 7 seconds for a 30° gimbal elevation change is tight, but if the yaw and gimbal elevation changes must be initiated separately, 4 seconds for a 30° gimbal elevation change is likely way too fast at a slow gimbal elevation setting. Extra time could be programmed only in between the consecutive shots in the sequence at the two times when the the gimbal elevation needs to be changed, but that would require extra programming beyond the simple steps outlined here, which may or may not be possible.

Finally, let's look at a Scenario 4: The advertised 20 shots in 2 minutes with a 60° yaw in between each shot, and 3 separate 30° gimbal elevation changes during the sequence. That requires a 6 second shot interval, including a 60° yaw and a 30° gimbal elevation change. There is no way that will work reliably with a slow micro SD card and slow yaw setting and a slow gimbal elevation change setting, and, never, if DNG or JPG+DNG is selected, or any form of AE bracketing or multi-shot setting is chosen. A JPG+DNG 5 bracket AE sequence takes 30+ seconds to complete before the buffer will allow any additional shots to be taken. All shutter commands during that time are completely ignored and trashed, even though it appears they are being executed, complete with corresponding shutter noises, but no images will end up on the card from those "too soon" shutter releases.

Fewer total images requires greater yaw movement in between shots for coverage, and more time to stabilize the quad gimbal thereafter, and requires more time at a slow user selected yaw speed to complete the yaw movement in the short fixed time period in between shots.

Somewhere between the two extremes above of 6 second shot intervals with 60° yaws squeezed in between, and 15 second shot intervals with only 45° yaws in between is an optimal set up.

At least we have outlined how all the parameters work together so that no surprises pop up when the limits are being pushed at the expense of reliability across multiple users with vastly different yaw and gimbal speed settings.

Let's figure this out together. I've covered all the possible minefields that you may have previously overlooked that may be causing you to pull your hair out! :)
 
Last edited:
Look realistically what you are doing (Dronepan or other) may have been great once, but panos are way to easy now days, whether underwater, on a mountain or using a drone. The software is already there. Anyone can do a pano with a drone using old tools. And they work very well.
 
Last edited:
Look realistically what you are doing (panodrone) may have been great once, but panos are way to easy now days, whether underwater, on a mountain or using a drone. The software is already there. Anyone can do a pano with a drone using old tools. And they work very well.
Not sure where you are coming up with Panodrone when this is a thread about Dronepan. No one has claimed that spherical panoramas are difficult. The Dronepan app is just automating the image taking process. It doesn't create the sperical panorama. Another program is needed to create the spherical panorama, and turn it into an interactive webpage. These are spherical panoramas, which are quite different from normal panoramas such as those on the Panodrone website. Perhaps you misunderstood?
 
I tried the latest beta today and noticed that only 19 shots were taken. 3 at different pitches for each 60 segment and one directly below. I think the 3 on one segment is better than circling completely three times at different pitch as things can change in the time for one rotation. I haven't created a sphere yet so I don't know what affect the single downward shot will have.

The software got confused as it still displayed a 20th shot and even a 21st shot back at the start, but it only stored 19.
 
I tried the latest beta today and noticed that only 19 shots were taken. 3 at different pitches for each 60 segment and one directly below. I think the 3 on one segment is better than circling completely three times at different pitch as things can change in the time for one rotation. I haven't created a sphere yet so I don't know what affect the single downward shot will have.

The software got confused as it still displayed a 20th shot and even a 21st shot back at the start, but it only stored 19.
I shot 5 successful spherical panoramas last night with the latest Dronepan beta. 100% success, even using J + R for each shot! The strategy of minimizing the number of yaw movements and including a slight delay after the yaw movement is brilliant. It allows the yaw to settle in, and then only the gimbal is moved for the 3 elevation shots at 30° intervals, starting out at horizontal. As previously discussed, the straight down shots are completely unnecessary and can lead to ghosting. 18 shots are all that are necessary, consisting of the 3 elevations (0°, -30°, -60°) at 60° intervals around the circle.
I was even able to task switch between DJI GO and Dronepan, although DJI GO crashed once. The new 22 shot sequence consists of 7 sets of 3 elevation changes and one downward facing shot (which is unnecessary). Horizonrtal intervals around the 360° appear to be 51.4° x 7, instead of the prior 6 intervals of 60°, allowing for more overlap.

A couple reminder tips upon entering Dronepan would be helpful before allowing the user to press the play button:
1. Make sure video recording is turned off (errors out if you don't, but it also screws up the capture sequence, resulting in 19 images instead of 22, but oddly still created a successful spherical panorama, even though 7.8° should have been missing due to insufficient overlap).
2. Flip the mode switch from P to F (errors out with a message, telling you to do so, but why not put the message first...).
 
One of the reviews on the DronePan app says they could not reconnect to DJI GO after using DronePan. This is making me nervous to try it. How would I recover the drone floating 400 feet up in the air if this happens to me?

Edited: Okay, I just tried this with the equipment just sitting on a table. I see what he means. I have to restart DJI GO as if I had never run it yet... it does not remember where it last was. This is on iOS, by the way. I think I'll avoid trying this live unless someone can let me know it'll all be okay.
 
One of the reviews on the DronePan app says they could not reconnect to DJI GO after using DronePan. This is making me nervous to try it. How would I recover the drone floating 400 feet up in the air if this happens to me?
Just unplug the lightning cable and plug it back in and kill off DJI GO and relaunch it to recover DJI GO, should that happen. You still have full transmitter control, even without the tablet hooked up. Make sure to switch the mode switch back to P for DJI GO control.
 
  • Like
Reactions: Porc
Just unplug the lightning cable and plug it back in and kill off DJI GO and relaunch it to recover DJI GO, should that happen. You still have full transmitter control, even without the tablet hooked up. Make sure to switch the mode switch back to P for DJI GO control.

Thanks. It probably helps if I properly switch between apps too... I am new to iOS. It looks like if I press the Home button twice, it gives me a list of apps and I can switch without losing DJI GO. At least, that's what appears to happen here on the table. I'll try for real at lunch time.
 
Just unplug the lightning cable and plug it back in and kill off DJI GO and relaunch it to recover DJI GO, should that happen. You still have full transmitter control, even without the tablet hooked up. Make sure to switch the mode switch back to P for DJI GO control.

Unplugging and plugging in the cable again did the trick. However, I wasn't able to get DronePan to do anything... it just gave an error of some sort. I was too focused on the drone to worry about writing down the error message.
 
Unplugging and plugging in the cable again did the trick. However, I wasn't able to get DronePan to do anything... it just gave an error of some sort. I was too focused on the drone to worry about writing down the error message.
You have to make sure you aren't recording video when switching to Dronepan, and you have to change the mode switch from P to F before pressing the play button in Dronepan. Make sure your camera mode setting is set to single shot only, but you can use either JPG or R or JPG +R, and if using JPG+R, you'll end up with 44 images total.
 
  • Like
Reactions: Porc
I'm on the beta for a P3 and the app works really well! Stitched together with PTgui (obviously) the uploaded to a pano website.

https://www.360cities.net/image/huntington-beach
The link repeatedly crashes Safari on my iPad as soon as I try scrolling around. You might want to consider configuring PTGui to reduce the file size to support iPads. It's in the settings.
 
  • Like
Reactions: markdizle
I was specifically speaking of task switching between iOS P3P flying apps. The SDK requires exclusive control of the P3P, and merely "suspending" the DJI GO app won't work. It has to be completely terminated. Therefore, you cannot task switch between P3P flying apps, unless a workaround is found. Besides, would you really want both flying apps running at the same time? If a crash occurs due to one of the apps, you would never know which app caused the crash!:cool:
Updated: Suspending DJI GO while control is transferred to Dronepan by changing the Mode switch from P to F now works!
 
Tried this app today. Pretty amazing. I was flying at 120 meters high and 300 meters away. After the Pano was ready I switched back to the Go app and it said NO CONNECTION. Back to the Pano app and flying on the camera back. Glad I knew the surroundings and how to fly manually.

With PTGui it's just very easy to make a 360.
I had this same thing happen, I figured I did something wrong. Do you know how to get back to the go app with out the connection loss?
 
I gave this a whirl yesterday.. ended up with some not so great results. More testing to come. Not sure if its the app, or the processing afterword that is to blame.
 
My last several Dronepan shots have resulted in missing panorama pieces.This is unacceptable! Taskswitching to DJI GO after completing the Dronepan shot sequence frequently results in a crash of DJI GO. The developer refuses to respond to this thread and has obviously abandoned the app! They have refused to follow my suggestions to guarantee a sucessful shot sequence.

I can no longer recommend the app!

Instead, shoot the panorama manually by replicating the automated process.
It is easier and faster and guarantees full control of your drone in DJI GO throughout the process in P-GPS mode. No app crashing. No loss of FPV. No missing tiles needed for the spherical panorama. 100% success guaranteed every time!

Set gimbal to horizontal and point compass heading due North by rotating the aircraft. From that orientation, take shots 1-3 at horizontal, 30° below and 60° below. Rotate the aircraft 45° to Northeast. Take shots 4-6 starting at the gimbal position of 60° below horizontal, elevating to 30° below, and horizontal. Rotate the aircraft 45° to due East. Begin shots 7-9 at horizontal, and then 30° below and 60° below. Repeat this process at 45° intervals of rotation until the full 360° has been completed, with shots 30-32 at due Northwest at each elevation of horizontal, 30° below, and 60° below. Stitch in PTGui or Panorama Studio Pro or ICE. Enjoy! :cool:
 
  • Like
Reactions: III% Streve

Recent Posts

Members online

No members online now.

Forum statistics

Threads
143,094
Messages
1,467,602
Members
104,980
Latest member
ozmtl