Forum : Technical Preview

Get all your questions answered about our latest Photosynth Technical Preview.

Topic: A photosynth 1 dismantled to photosynths 2

Report Abuse Report Abuse
Being thrilled with the promising possibilities of Photosynth 1, I had my own try-out project on 27 of April 2011. An antique town square.

Cuba, Sancti Spíritus, Trinidad, Plaza Mayor: 864 photos, 34 panos, 1 spin. With some earlier material resulting in
I think the Photosynth Team gave as one of the reasons to embark on the Photosynth 2 project, that navigating a scene in Photosynth 1 is not that natural as one would prefer. That is exactly how I felt about the above mentioned photosynth of mine. Now that there is Photosynth 2, I set myself to decomposing the original material in photo sets per shooting location plus the spin of the statue. It breaks down in 4 major parts. Photo sets at the street side of the Plaza Mayor park, at the border of the park, in the cross in the park and the statue in the middle. Following are the Photosynth 2 results.

I am very curious what functionality the Photosynth Team is going to provide to link these up into one experience and when.
Some corresponding Google Spheres can be found at

I intend to post a link to a cross reference database in another topic.

Note that Google has a policy, that does not allow for non-360 panos and that it requires sufficient vertical coverage to accept a Photo Sphere as a Google Maps View on the map.

I very much appreciate that Photosynth is not imposing these limitations. I believe there is a lot of material in people's archives which would be a great contribution to the Maps experience. Having this sort of restrictions would reduce reuse close to zero %.
In other topics on this forum there have been posts about migrating existing photosynths to Photosynth 2. Having done the migration of all my material manually, I do not have the impression that a fully automated migration is possible. Maybe it would be possible to have tools to support a manual migration.
PhotosynthTeam (6 months ago)
Joscelin --
This is a great data set. Thanks for doing all the work to lay it out. Do you have "walks" that connect your panos in the dense areas? 

I ask because our work on connected synths is progressing, and using walks as connectors between panos, spins, and walls seems like a really natural way for people to capture a place.
You are most welcome. I wish I had these "walks" you ask for. With the knowledge I have today, I would have shot differently. I am staying on Curaçao now. But, I am in love with the antiquity of Trinidad the Cuba and I might visit there again in September.

I am interested to engage in such an experiment. But, I suggest you to post an open request for material with instructions. Surely many of us would like to help out.
The link to the above mentioned cross reference database of my material is in this topic
PhotosynthTeam (6 months ago)
Thanks Joscelin -- I'll do just that when we've got some instructions on best practices for shooting synths that are intended to be connected. I don't want people to spend hours shooting and end up with something that doesn't work well because our model was slightly different than what you were expecting. Hold tight for a little while!  -- David.
DontPanic (6 months ago)
Do you have an eta, I have some projects planned too I was trying to figure out a solution. I'm sure I'm not alone.
eta = estimate time of arrival ;-)
@PhotosynthTeam. Sure, walks are the natural way to connect. Having experience with connecting Google Spheres into a Constellation, I currently cannot imagine that you intend to do this fully automated and have all components and the constellation thereof oriented properly on the map. I am curious to see what you come up with.
NateLawrence (6 months ago)
@Joscelin, very cool! 
Sadly, I don't have an installation of Access handy to open your database.

I think that if we had a tool that provided us a view of the camera frusta in a Photosynth 1 synth (like Greg Pascale's iSynth app or Henri Astre's WebGL photosynth 1 viewer did) that allowed us to select a group of camera frusta and send those corresponding photos to make a Photosynth 2 synth with, that would be an ideal tool for a user to select photos with.

I would also point out that Photosynth 1 already knew where 'objects' and 'panoramas' were within synths + changed navigation options when navigated to (the 'torus' or 'donut' at spins and click + drag within panoramas to navigate to neighboring photos in the pan) which could aid in automating conversion.

@Photosynth team, I'm sure that you have plenty of tools to take a look at the camera positions in a synth, but here's a snapshot of Joscelin's plaza synth in Henri's WebGL viewer:
To have this Cuba, Sancti Spíritus, Trinidad, Plaza Mayor 2011 Photosynth experiment migrated to a Google Maps Views Sphere Constellation turned out to be a challenging adventure. Here is the (limited) result.
Migrating the Photosynth to a Google Maps Views Photo Sphere Constellation did become sort of an adventure. Shooting was back in April 2011, with a different aim and guidelines. It was a test project for something rather different. Therefore none of the 'Spheres' is a sphere, all are cylinders. Camera was Panasonic Lumix DMC-TZ10 with 16:9 aspect ratio sensor and GPS sensor. GPS co-ordinates were fine tuned with #GeoSetter which uses EXIF Tools and Google Maps. Stitching was done with Microsoft Image Composite Editor (#ICE) with default settings. Adding the Google Sphere required EXIF properties was done with I ran into trouble there and found out that this site has a 60 Mega pixel limitation. I used ICE to downscale the stitch as needed before saving. Next I came across some little inconveniences when loading to Google+ photo albums, which propagated to loading to Views. Not a showstopper however and to be addressed elsewhere.
In this case the biggest possible showstopper of all turned out to be the acceptance criteria for Google Maps. Acceptance seemed to be arbitrary, not really depending on vertical coverage. In the end I called on +Adam Lasnik . He commented on vertical coverage and had a side note on top and bottom borders being jagged. The latter is the clue in this case and is not detailed in the policies. In this respect it should be noted that the scenery is on a slope, which made me change vertical direction of the camera while shooting. I set myself to restitching and cropping top and bottom. This resulted in more panoramas to be accepted, but not all. Another restitch, fine tuning the balance between cropping for jagged borders, retaining or losing parts in the scene and vertical coverage did the trick. I got all material I could possibly reuse because of the 360 requirement accepted today.
Completing the Constellation, aligning all Spheres and orient them all properly was a precision task that took time. It would have been easier with better satellite imagery at a higher zoom level, being projected plain and not tilted. So, this is not a Constellation of beautiful Spheres, but Plaza Mayor in #Trinidad de Cuba is a beautiful historic UNESCO #worldheritage   place. I enjoy it. Enjoy it too.
Reminded about this topic of mine by a recent reference, I have to note the following. In a Photosynth Preview update, all geo co-ordinates were removed from the uploaded images. Currently, I am in the process of adding Preview update 3 features to my synths. I will resynth all my synths that lack the geo coordinates. As a result, soon most of above referred synths will no longer be available at the mentioned url.
Also, I am removing from my library Photosynths v1 which are in some way redundant to my Photosynths v2, as I see no cause in carrying forward the legacy.
Find the new versions on the map here...,Panos&lat=21.8050524595493&lon=-79.98439777932589&zoom=20&user=Joscelin.Trouwborst
You need to be Signed In to add a comment. (Are you new? Sign Up for a free account.)