MAVIC 3 Azimut? Panoramic angle??

[Remimoty]

Hey, I just want to now if you know a technique to see the azimut angle of the Mavic 3 like the Course Lock on the Phantom4 or DJI M300... I use this for the panoramic photos...
Thanks!

 

[GadgetGuy]

Hey, I just want to now if you know a technique to see the azimut angle of the Mavic 3 like the Course Lock on the Phantom4 or DJI M300... I use this for the panoramic photos...
Thanks!

I'm not familiar with how the azimuth angle is visible in Course Lock on the P4 and M300. Can you please clarify?

I, too, would love to have a better way of reliably making small, repeatable yaw changes for panos, other than just visually approximating, by looking at the aircraft orientation icon on the map view, which is how I have done them manually in the past.

I can easily envision 45° changes by starting out at 12 o'clock, and picking halfway between 12 and 3 for 8 shots per row. DJI has now automated those easier panos. However, to use 2x on the main camera or 7x on the telephoto camera of the M3 for gigapano results, requires much smaller yaw changes in degrees!

Eagerly following all other replies!

 

[adcimagery]

Given the lower degree of control over nodal point location, I think you'd want to err on the side of shooting with greater overlap anyway. I think you can probably just eyeball a 30% overlap between shots, perhaps by using the grid overlay. Am I missing a particular use case for the azimuth angle?

 

[GadgetGuy]

Given the lower degree of control over nodal point location, I think you'd want to err on the side of shooting with greater overlap anyway. I think you can probably just eyeball a 30% overlap between shots, perhaps by using the grid overlay. Am I missing a particular use case for the azimuth angle?

Eyeballing works fine for 45° angle changes at 1x, but when you want a 7x telephoto with 6.5° angle changes, it becomes more complicated. However, your suggestion of using the 9 box grid overlay seems like it might work. It would work independently of the telephoto level, guaranteeing the appropriate 30% overlap. I'll have to give it a try, using the 2x digital zoom on the main camera initially for 100 shots, and graduating to the 7x telephoto camera for 1250 shots, hopefully within 33 minutes!

 

[adcimagery]

Eyeballing works fine for 45° angle changes at 1x, but when you want a 7x telephoto with 6.5° angle changes, it becomes more complicated. However, your suggestion of using the 9 box grid overlay seems like it might work. It would work independently of the telephoto level, guaranteeing the appropriate 30% overlap. I'll have to give it a try, using the 2x digital zoom on the main camera initially for 100 shots, and graduating to the 7x telephoto camera for 1250 shots, hopefully within 33 minutes!

You’re planning on a tele-lens pano with 1200+ shots?

 

[GadgetGuy]

You’re planning on a tele-lens pano with 1200+ shots?

That's the ultimate goal! If GigaPanos can do it, why not the M3?
Need to check the specs of PTGui and PanoramaStudio Pro to see if there is a max number of images they can stitch. In any event, starting with the 2x digital zoom of the M3 or the 2x optical zoom of the M2Z should double the resolution, requiring only 100 shots. The 1200+ shots is based upon a complete Spherical Pano using the 7x telephoto. However, a 180° pano using the 7x telephoto would cut that in half, and use a more practical 625 shots. Any thoughts?

 

[GadgetGuy]

Still interested in the answer to the OP's original question, as using the grid overlay won't help without any identifying objects in the field of view from which to measure a 30% movement within the frame.

 

[adcimagery]

That's the ultimate goal! If GigaPanos can do it, why not the M3?
Need to check the specs of PTGui and PanoramaStudio Pro to see if there is a max number of images they can stitch. In any event, starting with the 2x digital zoom of the M3 or the 2x optical zoom of the M2Z should double the resolution, requiring only 100 shots. The 1200+ shots is based upon a complete Spherical Pano using the 7x telephoto. However, a 180° pano using the 7x telephoto would cut that in half, and use a more practical 625 shots. Any thoughts?

I don’t think you’ll really benefit from shooting a pano at 2x digital, as digital zoom is just a digitally uprezzed and cropped 1x. You could get the same effect by just upscaling a 1x pano.

The 7x will be possible, but you’re also going to face challenges with parallax errors, shifting lighting because it’ll take so long to shoot, and stitching errors if the craft drifts over the course of shooting. I’m not sure it’s worth the immense effort, but good luck!

 

[GadgetGuy]

I don’t think you’ll really benefit from shooting a pano at 2x digital, as digital zoom is just a digitally uprezzed and cropped 1x. You could get the same effect by just upscaling a 1x pano.

The 7x will be possible, but you’re also going to face challenges with parallax errors, shifting lighting because it’ll take so long to shoot, and stitching errors if the craft drifts over the course of shooting. I’m not sure it’s worth the immense effort, but good luck!

That leaves the 2x optical from the M2Z, but its still images are only 12MP instead of 20MP. Is the 2x digital on the M3 main camera only a 14MP image or an upsized 20MP image? Is it a dynamic zoom or strictly digital? If the former, it might improve the quality and resolution of the end result over the 1x version. Hmmm. If the 7x can be shot at the same rate as the 1x done by DJI (40 secs for 26 images), it could be completed in 33 minutes. It will likely require Litchi for the automation. Need to wait for the SDK. In the mean time, the 1x M3 spherical panos are still the best DJI has created to date. The HiRes stitch provides almost the same 100% quality I get from PTGui and PanoramaStudio Pro, with an included ceiling already cloned in.

 

[adcimagery]

That leaves the 2x optical from the M2Z, but its still images are only 12MP instead of 20MP. Is the 2x digital on the M3 main camera only a 14MP image or an upsized 20MP image? Is it a dynamic zoom or strictly digital? If the former, it might improve the quality and resolution of the end result over the 1x version. Hmmm. If the 7x can be shot at the same rate as the 1x done by DJI (40 secs for 26 images), it could be completed in 33 minutes. It will likely require Litchi for the automation. Need to wait for the SDK. In the mean time, the 1x M3 spherical panos are still the best DJI has created to date. The HiRes stitch provides almost the same 100% quality I get from PTGui and PanoramaStudio Pro, with an included ceiling already cloned in.

On the M3, it’s my understanding that only the 1x and 7x are “real” focal lengths. At 2x, it’s upscaling and cropping the 1x, so you could get better results from doing that on the computer with the more effective upscaling algorithms.

I’ll have to check the drone’s stitch, but the big thing for me is not having the DNG’s level of adjustability for things like projection, blending, color and luminosity adjustments, etc.

 

[GadgetGuy]

On the M3, it’s my understanding that only the 1x and 7x are “real” focal lengths. At 2x, it’s upscaling and cropping the 1x, so you could get better results from doing that on the computer with the more effective upscaling algorithms.

I’ll have to check the drone’s stitch, but the big thing for me is not having the DNG’s level of adjustability for things like projection, blending, color and luminosity adjustments, etc.

I suspect you are correct, and it is way too much hassle to do it manually. Trying it on the Mavic 2 Zoom optical 2x loses the benefit of increased resolution when the stills are only 12MP instead of 20MP, as on the Mavic 2 Pro, and the much bigger M3 sensor.

You can have both! Not every pano works well. For a mere extra 35 seconds you get a 50MB+ HiRes stitch, as a proof of concept test print, and you still have all 26 DNG images to work with later. They are not mutually exclusive. Just choose "keep originals" and while in the Pano mode, select DNG under the camera settings. The camera settings while shooting panos are separate from the regular default still camera settings. For example, you can't choose both JPG and DNG for the pano originals, like you can for stills. Just choose DNG.

 

[adcimagery]

I suspect you are correct, and it is way too much hassle to do it manually. Trying it on the Mavic 2 Zoom optical 2x loses the benefit of increased resolution when the stills are only 12MP instead of 20MP, as on the Mavic 2 Pro, and the much bigger M3 sensor.

You can have both! Not every pano works well. For a mere extra 35 seconds you get a 50MB+ HiRes stitch, as a proof of concept test print, and you still have all 26 DNG images to work with later. They are not mutually exclusive. Just choose "keep originals" and while in the Pano mode, select DNG under the camera settings. The camera settings while shooting panos are separate from the regular default still camera settings. For example, you can't choose both JPG and DNG for the pano originals, like you can for stills. Just choose DNG.

Yeah, I’d have to do the math, but 48mm at 12mp might still be more angular resolution than the m3’s 24mm at 20mp, but there’s a lot of variables there.

Yeah, my point was that the stitched pano doesn’t have the same adjustability as the raw files. Since I will always plan on creating a pano from the raws, the stitched file is just taking up time and storage space, however slight. I’m happy to pick DNG, but I should also be able to pick “just shoot the pattern and save the files”.

 

[GadgetGuy]

Yeah, I’d have to do the math, but 48mm at 12mp might still be more angular resolution than the m3’s 24mm at 20mp, but there’s a lot of variables there.

Yeah, my point was that the stitched pano doesn’t have the same adjustability as the raw files. Since I will always plan on creating a pano from the raws, the stitched file is just taking up time and storage space, however slight. I’m happy to pick DNG, but I should also be able to pick “just shoot the pattern and save the files”.

I'd be very interested in the math, if you get a chance.

From everything I have understood about how the M3 pano algorithm works, the first 50% shoots the individual panos and saves them, if you have selected that option, and the last 50% processes the stitch, which can easily be cancelled by just moving the sticks at any time after the first 40 seconds needed to reach 50%. That should be a decent workaround for you, if you don't like test prints! Experiment with it. I haven't tried deliberately abandoning the stitch, since I like the test prints, but play with moving the sticks after 50% to say 60% to make sure the files are written. Should work.