Zoom camera vs 4/3ds video crop.

[Bussty]

Dumbed down Kuula stitch ... 75 Images &342mp image. With practice it get's better

Gustavus Adolphus College, St. Peter Minnesota - 02/13/2022....

Gustavus Adolphus College, St. Peter Minnesota - 02/13/2022. Mavic 3 7x Zoom panorama reduced to 38%

kuula.co

100% Crop ....

Gustavus Adolphus College, St. Peter Minnesota - 02/13/2022....

Gustavus Adolphus College, St. Peter Minnesota - 02/13/2022. Mavic 3 7x Zoom panorama 100% crop ... approx 30 images stitched.

kuula.co

Cheers, Jon

Jon!

That is impressive! Thanks for posting. So on both images from a horizontal perspective how much horizontal space would be covered by the main lens in comparison in a single frame?

Do you feel you have more resolution than what the main camera lens could have provided from the same shooting position?

You have stitched these really nicely. What are you using?

Thanks so much!

Andrew

 

[adcimagery]

Can we talk about this some more? Your 7x photo vs Main Camera shot. Have you not just proved that if you take a grid of images with the 7x that matched the view of a single main camera frame you would absolutely produce a stitched image of way higher resolution than what the main camera could do? Have you tried this? If so would love to see a comparison so single frame from main camera vs grid stitch of 7x camera to match?

This is my last point to prove and if I can I will buy the M3 over the Autel Lite+ as if that tele can be stitched to provide a higher resolution image from say effectively 1x down to 2 or 3 x then wow! As a stills photographer that offers huge potential.

Hopefully you are keen to try and show the result?? Many thanks in advance

I can guarantee that the 7x stitched would have more resolution, but it's a question of usability. The 7x offers less control over post processing, as you don't get raw files. Also, you'll have to manually shoot that pano, and alignment there may be tricky.

 

[GadgetGuy]

I can guarantee that the 7x stitched would have more resolution, but it's a question of usability. The 7x offers less control over post processing, as you don't get raw files. Also, you'll have to manually shoot that pano, and alignment there may be tricky.

I had no trouble with alignment, following your great suggestion of using the 9 box grid overlay, to end up with a roughly 50% overlap, and stitching in Panorama Studio 3 Pro, and telling it the shooting order of the images. Total of 47 images for 360° with a 15° FOV per image. The main camera has an 84° FOV and uses 8 images for 360°.
360/84° = 4.28 if no overlap
They use 8 images per row for a 50% overlap in the automated spherical pano
360/15° = 24 if no overlap
I used 47 images for a 50% overlap in my manually shot, single row 360° pano
I've successfully shot 3 rows of 47 images and stitched them.
Here is a comparison of the results in low light indoors:
Main camera on the left, compared to the 7x telephoto on the right, both shot from the same distance, in the same lighting.

 

[Meta4]

The main camera has an 84° FOV and uses 8 images for 360°.
360/84° = 4.28 if no overlap

Field of view for lenses is conventionally expressed as the diagonal angle.
For the main camera the diagonal is 84° and the horizontal angle would be 73.7°

 

[adcimagery]

I had no trouble with alignment, following your great suggestion of using the 9 box grid overlay, to end up with a roughly 50% overlap, and stitching in Panorama Studio 3 Pro, and telling it the shooting order of the images. Total of 47 images for 360° with a 15° FOV per image. The main camera has an 84° FOV and uses 8 images for 360°.
360/84° = 4.28 if no overlap
They use 8 images per row for a 50% overlap in the automated spherical pano
360/15° = 24 if no overlap
I used 47 images for a 50% overlap in my manually shot, single row 360° pano
I've successfully shot 3 rows of 47 images and stitched them.
Here is a comparison of the results in low light indoors:
Main camera on the left, compared to the 7x telephoto on the right, both shot from the same distance, in the same lighting.
View attachment 145301

Cool, I remember we were discussing that. Happy to hear it works out!

 

[GadgetGuy]

Field of view for lenses is conventionally expressed as the diagonal angle.
For the main camera the diagonal is 84° and the horizontal angle would be 73.7°

Thank you. Any idea what the horizontal angle is on the telephoto lens, which is 15° diagonally?

 

[GadgetGuy]

Cool, I remember we were discussing that. Happy to hear it works out!

It does, but it is a lot of manual work, compared to the automated spherical pano on the main lens, which is already stitched for you in HiRes now! Unless that increased level of detail is critical, the main camera does a very good job with a full 360° x 180°, if you include the cloned in ceiling, and only takes 75 seconds from the shooting the first of 26 total images to the final 60MB stitch, and saves all the original files, even in DNG, if desired, for further creations!

 

[Meta4]

Thank you. Any idea what the horizontal angle is on the telephoto lens, which is 15° diagonally?

A 162mm lens in 35mm terms has a diagonal VoV of 15° and a horizontal FoV of 12.7°

 

[Bussty]

I can guarantee that the 7x stitched would have more resolution, but it's a question of usability. The 7x offers less control over post processing, as you don't get raw files. Also, you'll have to manually shoot that pano, and alignment there may be tricky.

Yeah part of my long term master plan is that an SDK is released one day and Litchi let's you make custom Pano's using that lens which would be perfect. Also it seems really odd DJI aren't offering RAW files I wonder if that will ever be a thing? I just can't see why to boost sales they wouldn't build in both those things as feel like relatively simple Firmware updates then they could boast super high resolution 2-3x tele equivalents if you want to stitch the images...

 

[Bussty]

It does, but it is a lot of manual work, compared to the automated spherical pano on the main lens, which is already stitched for you in HiRes now! Unless that increased level of detail is critical, the main camera does a very good job with a full 360° x 180°, if you include the cloned in ceiling, and only takes 75 seconds from the shooting the first of 26 total images to the final 60MB stitch, and saves all the original files, even in DNG, if desired, for further creations!

I think the real potential for that Tele lens is grids up to the area of the main camera lens seems like a massive job to use it for 360x180 but wow what resolution you would have. Thing I like about Tele lenses too is they stitch really easily and have less lens distortion. I would really love to see with my own eyes a single frame from main camera with exact same view created from grid of stitched 7x images. Imagine if you could do that with a one button push and it took 30 secs

 

[adcimagery]

Yeah part of my long term master plan is that an SDK is released one day and Litchi let's you make custom Pano's using that lens which would be perfect. Also it seems really odd DJI aren't offering RAW files I wonder if that will ever be a thing? I just can't see why to boost sales they wouldn't build in both those things as feel like relatively simple Firmware updates then they could boast super high resolution 2-3x tele equivalents if you want to stitch the images...

Eh, seeing how limited the firmware was at release for the Mavic 3, it doesn't surprise me that this wouldn't make the cut. As for raw files, I'm guessing the files are so heavily processed (noise reduction, lens corrections) that they'd be opening themselves up to criticism if the raws were available. Or maybe they just don't see tele as much of a feature beyond marketing copy.

 

[GadgetGuy]

I think the real potential for that Tele lens is grids up to the area of the main camera lens seems like a massive job to use it for 360x180 but wow what resolution you would have. Thing I like about Tele lenses too is they stitch really easily and have less lens distortion. I would really love to see with my own eyes a single frame from main camera with exact same view created from grid of stitched 7x images. Imagine if you could do that with a one button push and it took 30 secs

See post #23 above. It's the exact same thing. Also, it can never take 30 seconds. It requires roughly 50x the number of images, and will never be able to be stitched in camera. 26 images alone take 45 seconds to shoot and 30 seconds to stitch. It likely won't even be able to be shot in a single battery.

 

[Bussty]

See post #23 above. It's the exact same thing. Also, it can never take 30 seconds. It requires roughly 50x the number of images, and will never be able to be stitched in camera. 26 images alone take 45 seconds to shoot and 30 seconds to stitch. It likely won't even be able to be shot in a single battery.

Hi there, yes you are right but before I spend the dosh would just love to see the real world comparison on a whole scene. This functionality is the main point I would buy the M3 over the Lite+ and Yes wasn't really serious about the 30 secs.

 

[Bussty]

Eh, seeing how limited the firmware was at release for the Mavic 3, it doesn't surprise me that this wouldn't make the cut. As for raw files, I'm guessing the files are so heavily processed (noise reduction, lens corrections) that they'd be opening themselves up to criticism if the raws were available. Or maybe they just don't see tele as much of a feature beyond marketing copy.

What I don't understand is DJI have other 12mp cameras with RAW so sensor can't be the issue is it the lens? I still think the below is reasonably good image quality if taken with that 7x camera especially if only a small part of the main image in a grid stitch. I do wonder if DJI has just missed the potential they have here? It borders on a new way of capturing high quality images across a range of magnifications. Let say they went in boots and all and had a high quality say 10x Tele and designed from the ground up with overlays in viewfinder indicating differing areas you could capture and automating the capture process. The area of the Tele would become a unit of high quality and you effectively select the area you want that to cover and just take one photo with the camera/drone doing the rest with option to self stitch of course. Pipe dream of course but I think kind of where drone manufacturers need to start looking as provides capability never before seen in a consumer drones and the thing that has bugged me in the past is turning up to an awesome distant shot but only having the wide angle of a drone lens so would provide so much more flexibility. Being a drone the drone can do all the hard work of lining up the various shots which on a DSLR you have to do yourself. Just feels an avenue open to them that should pursue especially for still photographers.

 

[GadgetGuy]

Hi there, yes you are right but before I spend the dosh would just love to see the real world comparison on a whole scene. This functionality is the main point I would buy the M3 over the Lite+ and Yes wasn't really serious about the 30 secs.

A 3 row 360° pano is very doable, requiring only 47 images per row. However, when you compare it to the 75 second stitched 360° x 180° pano from the main camera, when you zoom into both images, you won't notice any difference until you start pixelating the main image and pixel peeping, like in the example above, which is far more zoomed in than any website rendering of the panos will ever allow. The main trade off is you lose the entire rest of the scene, which is captured by the main camera's 26 DNG files, with an f/2.8 lens, instead of an f/4.4 lens, and with a sensor with far greater dynamic range. In the 8 minutes it takes to just shoot the a three row telephoto pano, you can shoot 6 automated panos from different elevations or different locations, and you are done, unless you want to work with the DNG's!

As between the M3 and the Lite+, Autel Robotics only has a handful of engineers compared to DJI's 8,000 engineers, according to Randall, the former CEO of Autel Robotics, who resigned, when they didn't keep their promises to him. You will likely never see Litchi for the Lite+. DJI sells a million drones a year, while Autel Robotics sells between 5,000 and 10,000 a year. I'd go with DJI for a happier experience. We have far more resources here on this forum among those million drone purchasers a year. We crowd source our own support! They try over at AutelPilots.com, but they have less than 1% of our forum members, and most are unhappy. I own both an EVO and an EVO 2 Pro 6k. I regret buying both! Neither has held a candle to their DJI equivalents and successors. Took 18 months for Autel Robotics to just add 360° panos to the EVO 2 Pro 6k. They now still take 4.5 minutes each, and they cannot be stitched by the camera after that! Compare that to the 60 seconds on the M2P and the 75 seconds on the M3, including stitching and cloning in a ceiling!

 

[Bussty]

A 3 row 360° pano is very doable, requiring only 47 images per row. However, when you compare it to the 75 second stitched 360° x 180° pano from the main camera, when you zoom into both images, you won't notice any difference until you start pixelating the main image and pixel peeping, like in the example above, which is far more zoomed in than any website rendering of the panos will ever allow. The main trade off is you lose the entire rest of the scene, which is captured by the main camera's 26 DNG files, with an f/2.8 lens, instead of an f/4.4 lens, and with a sensor with far greater dynamic range. In the 8 minutes it takes to just shoot the a three row telephoto pano, you can shoot 6 automated panos from different elevations or different locations, and you are done, unless you want to work with the DNG's!

As between the M3 and the Lite+, Autel Robotics only has a handful of engineers compared to DJI's 8,000 engineers, according to Randall, the former CEO of Autel Robotics, who resigned, when they didn't keep their promises to him. You will likely never see Litchi for the Lite+. DJI sells a million drones a year, while Autel Robotics sells between 5,000 and 10,000 a year. I'd go with DJI for a happier experience. We have far more resources here on this forum among those million drone purchasers a year. We crowd source our own support! They try over at AutelPilots.com, but they have less than 1% of our forum members, and most are unhappy. I own both an EVO and an EVO 2 Pro 6k. I regret buying both! Neither has held a candle to their DJI equivalents and successors. Took 18 months for Autel Robotics to just add 360° panos to the EVO 2 Pro 6k. They now still take 4.5 minutes each, and they cannot be stitched by the camera after that! Compare that to the 60 seconds on the M2P and the 75 seconds on the M3, including stitching and cloning in a ceiling!

Thanks for those insights, the size difference between the two companies has been in the back of my mind. DJI is huge and widespread I actually just saw a Mavic 3 being used by the Ukrainian military as a spotter drone on Youtube!.

My Mavic Mini has never missed a beat including many 360 panos using Litchi so I do have very positive experiences with DJI.

The thing the main camera on the M3 can never do is give you a high quality x2 x3 or x4 shot so that's what I'm wanting the Tele stitch for as so often my main subject is further in the frame. As for higher dynamic range I can combat that by taking two sets at different exposures and hand blend or auto blend in PTGUI or any HDR software.

Thanks for your comments currently have local supplier on notice to let me know when they have at least February manufactured M3's as may as well try to get an Oval tele lens even if it's not conclusive yet that have improved image quality.

 

[GadgetGuy]

Thanks for those insights, the size difference between the two companies has been in the back of my mind. DJI is huge and widespread I actually just saw a Mavic 3 being used by the Ukrainian military as a spotter drone on Youtube!.

My Mavic Mini has never missed a beat including many 360 panos using Litchi so I do have very positive experiences with DJI.

The thing the main camera on the M3 can never do is give you a high quality x2 x3 or x4 shot so that's what I'm wanting the Tele stitch for as so often my main subject is further in the frame. As for higher dynamic range I can combat that by taking two sets at different exposures and hand blend or auto blend in PTGUI or any HDR software.

Thanks for your comments currently have local supplier on notice to let me know when they have at least February manufactured M3's as may as well try to get an Oval tele lens even if it's not conclusive yet that have improved image quality.

The main camera is actually quite good at up to 2x in Normal mode, and maintains the perspective of the telephoto lens, when both are shot from the same location. It almost replicates the M2Z in a much larger sensor. The other disadvantage of a telephoto lens is compressing any haze and heat in the air. Flying closer (with a realistic 5 mile range, if needed) would avoid that. Even exposure blending won't overcome the limits of the 1/2 inch sensor in low light, like at night. The max ISO is quickly reached at f/4.4, and you currently can't manually control the shutter speed to lower it, as everything is currently all Auto and JPEG.
Might as well get the newest M3 version, since it is certainly at least as good as the original version, and has a small possibility of being better, if you happen to get a really good lens on it.

DJI still makes the best stuff, and continues to innovate, rather than just copying the competition, or sitting on their laurels! As far as support, that's what we are all here for! Depending upon the manufacturers for customer support can be disappointing! That's not DJI's forte. Innovation is!

 

[Bussty]

Never been a fan of digital zooms, can just take the 1x image and blow up in Gigapixel but never the same. Especially if doing large prints though does depend on subject. I think I would get in trouble real quick if I flew 5 kms away

I'm looking forward to experimenting with the M3 you are slowly talking me into it you know

 

[GadgetGuy]

Never been a fan of digital zooms, can just take the 1x image and blow up in Gigapixel but never the same. Especially if doing large prints though does depend on subject. I think I would get in trouble real quick if I flew 5 kms away

I'm looking forward to experimenting with the M3 you are slowly talking me into it you know

Indeed. The up to 2x digital zoom in Normal mode is more practical for video composition on the fly. Just because DJI advertises it, doesn’t mean you should, but we are an international forum, and not all countries are under the jurisdiction of the FAA and its VLOS requirements. Some people also like to claim their eyesight is way better than others, while those with normal eyesight can still use spotters, under certain circumstances.

IMHO, the M3 is truly the best drone DJI has made, when paired with the RC-Pro. The 32-34 minute flight time alone is worth the upgrade from the venerable M2P and P4P, neither of which have a practical way to add external batteries to extend their flight time.

 

[Bussty]

Indeed. The up to 2x digital zoom in Normal mode is more practical for video composition on the fly. Just because DJI advertises it, doesn’t mean you should, but we are an international forum, and not all countries are under the jurisdiction of the FAA and its VLOS requirements. Some people also like to claim their eyesight is way better than others, while those with normal eyesight can still use spotters, under certain circumstances.

IMHO, the M3 is truly the best drone DJI has made, when paired with the RC-Pro. The 32-34 minute flight time alone is worth the upgrade from the venerable M2P and P4P, neither of which have a practical way to add external batteries to extend their flight time.

I thank you for your valuable feedback and very good points.