I shoot indoor spherical panorama with Gigapan and floor shot by hand. (picture with my legs ) Puctures are shooted with 3 steps of bracketing and after bracketing is "Enfused" with EnfuseGUI. Enfused pictures and settings of Autopano Giga are here:
After a lot days of work, I am not able to make spherical panorama without errors. Specially I have big trouble with ceiling. Autopano is not able to make controlpoints on celing using Similarity or Homohraphy... Can anybody told me, where i am making mistake ?
first - it can be done - thats the good answer. but ...
if I take all 25 images of your zip file - it fails. + some reasons maybe: mix of focals - the nadir handheld shot does have 18mm the others 15. - i did not use the handheld nadir image with your feet (handheld + non 90" to bottom angle will in most cases not work). + the 15mm images are recognized as non fisheye. this maybe correct or may not. one rule: if having troubles with 15mm or smaller that are nonfisheye recognized than try it with changing in image properties to fisheye. - I did change it to fisheye. + you shot with a regular image matrix - this leads to an rather big overlap in zenith row - I finally did use 5 images less... so in sum I used only 21 image to still cover the sphere. + you shot in landscape mode. - this may work or may not - but general practice is using portrait mode images. Advantage on indoor panos is you need less rows... and you've less ghosts from moving people (if there are ones...) + in screenshot you can see that autopano detected CPs between some images and between some no CPs could be detected. so I did delete all links between images not directly being neighbor images. and I manually added links between images to link the images. this dramatically reduced the detected RMS and the ceiling errors could be corrected as shown in preview mode... + you've dust on your lens or on your sensor. if dust is on the images (always on same x,y position) than make sure not CP's are detected by accident (didnt care about it and it worked...)
I did use apg 3.0 final. I used Detection Quality Standard CPs: 50 with Ransac model: Similarity Blending settings: Multiband Level 0 with Fusion and Cutting enabled Priority to Ghost.
3. mage: using nonfisheye properties and with default recognized CP connections 4. image: using with fisheye properties and after adding CPs to connect images 2. image: is the preview (preview feature with apg 3.0) 1. image: finally the resulting pano using 30% original size and 75%jpg quality compression to fit in the forum
and what with the nadir: I would use krpano tools to create cube files and postprocess the regular pattern. the corrected cube files can than remerged to a rectangular image with the kroanotools...
+ you shot with a regular image matrix - this leads to an rather big overlap in zenith row - I finally did use 5 images less... so in sum I used only 21 image to still cover the sphere.
+ you shot in landscape mode. - this may work or may not - but general practice is using portrait mode images.
That's because he used a Gigapan robotic head - Gigapan robotic heads can only shoot a regular matrix of images, and can only mount the camera in landscape orientation. Just two of the reasons why they are not really suited to shooting sperhical panos.
Last edited by mediavets (2012-12-07 00:17:30)
Andrew Stephens Nikon D40, Nikkor 10.5mm fisheye, Sigma 8mm f3.5 fisheye, Nikkor 18-55/50/35mm lenses, Nodal Ninja 5 Lite, Nodal Ninja 4 with R-D16, Agno's MrotatorTCS short. Nikon P5100, CP5000, CP995, FC-E8, WC-E63,WC-E68, TC-E2, Kaidan Kiwi 995, Bophoto pano bracket, Agno's MrotatorA. Merlin/Orion robotic pano head + Papywizard on Nokia 770/N800/N810 and Windows 8/XP/2K.