[wings]

This is new for me, but I need to create a panoramic photo for a project. This is a good time for me to inquire about how to go about this.

Here are the requirements:

• End file needs to be 2048 wide X 1024 high
• 32-bit RGPA
• PNG format
• Vertical sweep = 180° altitude
  • From Zenith (+090°) top of image
  • To Horizon (0°) middle of image
  • Bottom (-90°)
• Horizontal sweep = 360° azimuth
  • Start on left edge North (0°)
  • One quarter across East (90°)
  • Half way across South (180°)
  • Three quarters across West (270°)
  • Right edge to wrap around to North (360°)
• Image must contain an alpha or transparency layer that defines the parts that are opaque, everything except the sky and defining the parts that are transparent, the sky.

Here is what I have:

• I am waiting on my K-1 from being upgraded
• I do have a Sodial 360° panoramic head panning base
• The only vertical graduated device that I have is an old Bogen tripod
• I do not have Photoshop, but do have Cyberlink Photo Director which does have stitching ability, but not sure about this project.
• I have various lenses, but perhaps some that might be used are
  • K-mount Rokinon 14mm, SMC Pentax-A 50mm F 1.4
  • Older SMC M42 24mm F 3.5, SMC 55mm F 2, SMC 50mm F 4, Soligor C/D wide-auto 28mm F 2.8
  • I have larger lenses including zooms, but not sure it that would suite this project

Since this is new to me I need to find out what is needed.

• Do I need additional equipment?
• How do I take these photos?
  • Format 4:3, 16:9 etc
  • Lens
  • Other equipment
  • Camera settings
  • How do I image the sky?
  • How do I image the ground and eliminate the tripod?
  • Anything else?

• Post processing
  • Software?
  • How to include the transparent layer on the sky?
  • How to eliminate tripod?
  • Anything else?

So there you have it. I would appreciate any advice from those who might have experience with this.

[tuco]

It sounds like you need a 360° camera. 2048x1024 is only a 2:1 panoramic aspect ratio for a basically a hemispherical picture requirements.

[Bob 256]

Your only solution with the K-1 is to take a large series of individual shots (covering a projected sphere) and stitch them with software. Creating an alpha channel would also be done in software. It would best be done with a longer lens such that the field of each shot is limited so less edge distortion results and stitching is simplified but will be a tradeoff between focal length and how many shots you need to take. As for eliminating a "tripod", the support for the camera would have to be articulated and moved between shots so as not to show in the stitched result. Your format doesn't make sense since 2048x1024 implies a rectangular frame - not a spherical one. In any case, the resolution would be determined after stitching and by the software used to develop the spherical format.

As mentioned, it might be easier to use a spherical camera and combine two or more frames into a single 360 degree spherical shot.

Not an easy task and not suited for scenes with motion involved. Special cameras have been developed to capture a single shot 360 degree spherical image, but they are special, and usually involve two fisheye type cameras with more than 180 degree coverage, pointing in opposite directions, and even then, the support will show in the image and will have to be removed via software.

[wings]

Your only solution with the K-1 is to take a large series of individual shots (covering a projected sphere) and stitch them with software. Creating an alpha channel would also be done in software. It would best be done with a longer lens such that the field of each shot is limited so less edge distortion results and stitching is simplified but will be a tradeoff between focal length and how many shots you need to take. As for eliminating a "tripod", the support for the camera would have to be articulated and moved between shots so as not to show in the stitched result. Your format doesn't make sense since 2048x1024 implies a rectangular frame - not a spherical one. In any case, the resolution would be determined after stitching and by the software used to develop the spherical format.

As mentioned, it might be easier to use a spherical camera and combine two or more frames into a single 360 degree spherical shot.

Not an easy task and not suited for scenes with motion involved. Special cameras have been developed to capture a single shot 360 degree spherical image, but they are special, and usually involve two fisheye type cameras with more than 180 degree coverage, pointing in opposite directions, and even then, the support will show in the image and will have to be removed via software.

Thanks for the replies. A spherical camera would not work for this application. The format is determined by the application which uses the complete stitched image. It becomes sort of a map projection instead of a spherical image. It is essentially a panorama image including the sky.

[interested_observer]

Look at a Theata V...

• Product | RICOH THETA V

[wings]

Look at a Theata V...

• Product | RICOH THETA V

Those are nice, but spherical cameras would not work. Thanks anyway.

[wstruth]

Those are nice, but spherical cameras would not work. Thanks anyway.

Actually the Theta images are made in a 2 to 1 ratio and at well over that size, I have shots on Flickr that were uploaded as 5376 x 2688. My Theta SC will only shoot jpegs though.

As far as using a DSLR. The other posters have described what you need to do.

[wings]

Actually the Theta images are made in a 2 to 1 ratio and at well over that size, I have shots on Flickr that were uploaded as 5376 x 2688. My Theta SC will only shoot jpegs though.

As far as using a DSLR. The other posters have described what you need to do.

Thanks for that reply. Although those specialized cameras might work, it is not really necessary in this case. Panoramas can be shot using a regular camera and software such as Photoshop and others are used to stitch them together to make a panorama shot, which is how it was done prior to the 360° cameras which have recently become popular.


Best wishes.

[Hank]

A few interesting points:
Only 2048x1024 in resolution for the final product- so the loss of detail isn't a concern.
A defined horizon at the 50% mark of the frame.

Question:
Concerns about spherical projection or other lens distortions?

My solution:
What I'd do is fairly simple, use the 14mm lens on the trippod, with the camera held at the 90 orientation, shoot overlapping photos at every 15 degrees, left to right (0, 15, 45, 60, 75, 90, 105, 120, 135, 150, 165, 180). and then use autostitch (AutoStitch) to make your panorama photo, and then rescale to size.

[wings]

A few interesting points:
Only 2048x1024 in resolution for the final product- so the loss of detail isn't a concern.
A defined horizon at the 50% mark of the frame.

Question:
Concerns about spherical projection or other lens distortions?

My solution:
What I'd do is fairly simple, use the 14mm lens on the trippod, with the camera held at the 90 orientation, shoot overlapping photos at every 15 degrees, left to right (0, 15, 45, 60, 75, 90, 105, 120, 135, 150, 165, 180). and then use autostitch (AutoStitch) to make your panorama photo, and then rescale to size.

That sounds like something on the right track. I am not familiar with autostitch, but will look at it. It would need to allow one to select the portion of the image that is the sky from rest such as trees, structures and ground and hopefully would have a solution to remove the tripod shadows. The complete sky & ground would have to be included for this project.

[DeadJohn]

I wonder whether Microsoft ICE can do much of the work for you. A full 360 may or may not be beyond ICE; I've never tried going that far. Image Composite Editor - Microsoft Research

My use case for ICE is stitching stills but it can handle video input. Your modest output resolution 2048x1024 makes me wonder whether you even need a tripod. Put a wide angle lens on your camera, video mode, rotate the camera through 360 degrees including up and down. Try to rotate the camera around a single point in space, imagining where the sensor is located, moving your body to stay out of frame.

You'll definitely get some parallax error handheld, but ICE processing plus the relatively low resolution should be forgiving.

Image must contain an alpha or transparency layer that defines the parts that are opaque, everything except the sky and defining the parts that are transparent, the sky.

ICE cannot handle that part. The only way I know how to do layering is with Photoshop. GIMP, free software, can reportedly do it, also.

[Kozlok]

I’ve tried 360 degrees in photoshop, and it didn’t go well. Never did figure it out. There’s no way I could figure out how to tell it where to split the image.

[wings]

I’ve tried 360 degrees in photoshop, and it didn’t go well. Never did figure it out. There’s no way I could figure out how to tell it where to split the image.

That was my very first thought about them, it might be possible, but I would not know how to make it fit the criteria. I appreciate your efforts. I wonder if you could take screen shots and reassemble them?