Sphere Pano - # of Shots
- Thread starter ccoltmanm
- Start date
M2P = 26 shots here. Don't know why the difference other than maybe a different field of view for the Mavic 1?
passedpawn
Well-Known Member
Mavic 1 lens is FOV 78.8° 26 mm (35 mm format equivalent) which is actually a wider angle of view than the M2P. Any more than 26 images is a poorly designed shooting algorithm, with way too much overlap. DJI must have realized that, but not backfilled the MP with the better algorithm , to shoot it faster with the fewest number of needed images. At it's core, the sphere only needs three rows (30° up, horizontal, and 30° down) of 8 images shot at increments of 45° around the clock, plus two straight down shots at 90° to each other. Hence 26 images.
It certainly does, which is why there are 26 photos instead of only 18! However, 30° above the horizontal still can't shoot the sky ceiling, so you either need clone one in from the sky that is captured, or add it from stock photos of skies, or cut off the pano at the top edge of the top row. The default created jpeg output file on the card uses the first method in a "good enough" automated method, which can look a little strange at the top seam, when you pan up in the image.
ITs significantly less after it done its thing than mavic 1. Last time i looked at gimbal angle it showed 15 or 20 degrees not 30.
The in-built sky fill is hideous for an image you want anywhere that isnt facebook and exposures often need to be balanced.
Hence photoshop needed to do it properly - and the extra missing sky is more work.
The in-built sky fill is hideous for an image you want anywhere that isnt facebook and exposures often need to be balanced.
Hence photoshop needed to do it properly - and the extra missing sky is more work.
In hindsight, I think you may be correct about that. Good to know, because that would explain some issues I had, trying to include the top of a building in the top edge of a spherical panorama, and I made sure it was visible in the frame when I elevated manually to 120°, before shooting the pano, but when I got home, it was cut off in the actual pano, and cloned in! Unfortunately, I haven't paid close enough attention to the exact degree of the top row angle. I will on my next Spherical Panorama, though! I know it does a top row above horizontal. My guess is that a top row that is a full 120° might not contain any horizon, and might ruin the whole stitch, with no common areas to stitch from.
I totally agree that the cloned in area can look very unnatural, depending upon the detail in the sky in the original part. The more even the original sky is, the better the cloned result.
Personally, I only use the final microSD card image as a proof of concept, because I save all the originals as jpg's, and use those to stitch my final pano output using PanoramaStudio 3 Pro, where it simply cuts off the top of the panorama where the top row ends, so no cloning is necessary, but the result is 360x122.8 rather than 360x180, so it is missing the top 57.2° from the zenith down. No one really needs to see the top of the sky anyway. Usually nothing of interest there. PanoramaStudio 3 Pro does a much better job of stitching, while also balancing the exposures, and allows a higher level of zoom, and can plot the pano on a Google Map with the jpg GPS metadata. It's now only $80 and well worth it.
PanoramaStudio 4 Pro | PanoramaStudio Panorama Software
Looking at the numbers above of 360x122.8, I think I have the solution to making sure I am at a high enough elevation when shooting the pano to just barely include the top of the building in the resulting pano. 122.8° is slightly over 120°, so if I elevate the gimbal to 120°, and put the top of the building in the very center of the frame, rather than at the top edge of the frame, the building top will make it into the original pano photos generated by M2P, which PanoramaStudio Pro 3 will use. I like it! You have solved a major conundrum for me! Thank you!
Last edited:
I'll check it next time (i must admit i havent checked since last firmware upgrade, maybe its changed) but from memory i used to see +15 on gimbal only, not +30 (M1 did +30).
My workflow is basically convert the DNGs to TIFFs (with exposure corrections here in ACR). Then import the TIFFs into Microsoft ICE which does a good job of stitching (and free).
Then into Photoshop to correct the sky, canvas and aspect and the 3D workspace in Photoshop to make the seamless joins and create the final panorama.
And i agree a 360 looks far better if your main point of interest breaks the horizon line and as you said, that involves an upward angle.
My workflow is basically convert the DNGs to TIFFs (with exposure corrections here in ACR). Then import the TIFFs into Microsoft ICE which does a good job of stitching (and free).
Then into Photoshop to correct the sky, canvas and aspect and the 3D workspace in Photoshop to make the seamless joins and create the final panorama.
And i agree a 360 looks far better if your main point of interest breaks the horizon line and as you said, that involves an upward angle.
I'll definitely check, too! I'm still on FW 1.00.00. Pity that GO 4 won't let us save both jpg and DNG on the panoramas as originals. It's one or the other, but not both, unlike single images where you can save both, if we shot them completely manually. Unfortunately, PanoramaStudio 3Pro can't read the DNG metadata for GPS, for the mapping feature, so I use jpg's instead of DNG, which are plenty good enough because of the 1” sensor and 20MB still image size, and the 100MB size of the resulting 100% jpg stitch.
Will now concentrate on keeping the top of the point of interest below the center of the frame at 120° gimbal tilt for inclusion!
ICE is a good alternative, too, but I also use PTGui, whose stand alone viewer is great for viewing the camera generated stitch for quick examination, to make sure it worked, without investing the time to do a full workup, just to see it wasn't worthwhile. Fun stuff!
You can use Bridge or DNG creator (both free) to convert the DNG to JPG with embedded GPS data if needed.
I dislike JPGs and always want to work with the highest quality/least compressed output until the final image. Hence DNG > TIFF (Because ICE cant read DNG). The only JPG is the final finished 360.
If its a 360 only going on Facebook i'll reduce the image size in Photoshop to 10000x5000 before going into the 3d workspace editor and cloning.
I dislike JPGs and always want to work with the highest quality/least compressed output until the final image. Hence DNG > TIFF (Because ICE cant read DNG). The only JPG is the final finished 360.
If its a 360 only going on Facebook i'll reduce the image size in Photoshop to 10000x5000 before going into the 3d workspace editor and cloning.
PanoramaStudio Pro 3 does such a good job on the stitching and exposure matching and white balancing that the 26 JPG files have been more than good enough on the M2P, since they are are 20MB each from a 1” sensor. Trying to keep the workflow as simple as possible. I was initially just thrilled with the 12MP stitched JPG the drone made, but realized that a 70MB JPG from the original JPG's was much sharper and had more zoom capability, along with the ability to fix stitching errors. I've shot up to 8 panoramas in a single flight, while recording 13 minutes of 4K video in between the 8 pano locations!
