3D, VR, 360° and Street View Photographers  |  Real Estate Agents  |  3rd Party Service Providers
Should I buy a Camera? Quiz How Much to Start 3D Tour Business? Calculate How Much Can I Make as a Pro? Calculate Win More Listings? Calculate
We Get Around Network Forum
WGAN Training Academy (Learn more)
WGAN-TV Channel
Panoskin to Google Street View Demo and Discuss
for Newbies with Wojciech and Tom with Panoskin

Jagged and crooked doorframe4121

onepsg private msg quote post Address this user
This has happened to about 90% of my models. The rest of the scene is prefect but the door frame is jagged or crooked. I wonder if it had happened to any of you before and how to prevent it?

Since its happened in about 90% of my models so I am pretty sure its not the problem of imbalance tripod, weather, lighting, or location.




Post 1 • IP   flag post
Basic
Member
JohnLoser private msg quote post Address this user
@onepsg

Not being able to see your model, I can guess it's a "bad stitch" caused when you get something very close to the camera in multiple pictures of the same pano. The stitching can cause the jaggies.

The only possible solution is to try to be more than 18" away from door frames. This is not always possible -- it can cause failure to align of opposite sides of the door. So you take the best choice for a middle point.... That comes from trial and error.
Post 2 • IP   flag post
onepsg private msg quote post Address this user
@JohnLoser

My bad. I edited and upload some pictures.
Post 3 • IP   flag post
3rd Party
Service
srennick private msg quote post Address this user
@onepsg I am not as technical or experienced as most on here, but if you need to scan in the doorway to get alignment that is fine, just HIDE that particular scan in Workshop...you don't need it.
Post 4 • IP   flag post
3rd Party
Service
Metroplex360 private msg quote post Address this user
This is caused by parallax error due to the design of the Matterport Camera. As it rotates, the lenses do not stay fixed on a 'nodal point'.

http://www.johnhpanos.com/epcalib.htm

Look at the image below - this is how DSLR/Fisheye based photography works. The lens is DIRECTLY in line with the center of the tripod. As it rotates, the lens is in the EXACT same spot.




Now, think about your Matterport Camera as it spins -- do the lenses stay centered to a nodal point? No. Therefore, Matterport has to compensate with its stitching. Problems are WORSE if things are CLOSER to the camera -- thus door jambs and stairways have stitch errors.

Now, with a Ricoh Theta S or the upcoming IRIS 360 - you don't have to worry about this as there's no rotation and the lenses are calibrated for a one-shot no-parallax error experience.

I wouldn't look at this as a design flaw -- it's a compromise. In order for Matteport to work with an array of 3 3D-sensor/camera combos, it's designed the best way that it can be designed. This is just one area where there's no solution.
Post 5 • IP   flag post
iGuide Alex_iGuide private msg quote post Address this user
To reconstruct proper spherical image without parallax errors, all light rays from all the directions must pass through the same point in space. This is achievable only with a single lens rotated around the "nodal point" (correct term is "entrance pupil" of the lens). And even then, it just an approximation and is not 100% perfect. Thus, even Ricoh Theta will have a parallax error for close up objects. It will be better than MP, since the distance between the entrance pupils of the two lenses in Ricoh Theta is somewhat smaller than the inter-lens distance in MP or Iris cameras.
Post 6 • IP   flag post
42178 6 6
destitute