I do appreciate the Photosynth import plugin being built into Meshlab.
I would like to see the plugin updated to import the other two basic categories of uploads on Photosynth.net, in addition to the original synth type that the plugin currently imports.
The first of the two remaining categories is your basic stitched panorama in cube map format which debuted in early 2010.
The second category is generation 2 Photosynths which were publicly able to be uploaded via http://photosynth.net/preview since 2013 December 10. Technically, there are several subtypes within this category, but the download process should be virtually identical for each.
I know that time is precious for developers, so I would encourage you to review Christoph Hausner's code in SynthExport 2, which can be found at http://synthexport.codeplex.com/ since he has already implemented the ability to download both original resolution cube maps and point clouds and images from Photosynth 2.
There are a couple of things that I would like to see done in addition to what Christoph has implemented - namely:
1) downloading the original photos' metadata and applying it to downloaded images
2) downloading and exposing the depth maps which Photosynth 2 generates to project the photos onto. Note that these depth maps are a per-photo model, rather than a global model as the sparse point cloud is, but my instinct is that they could still be used collectively for visualization of scene depth. If that is not the case, at least they could be used to project the photos in a similar manner to the Photosynth 2 WebGL viewer.
3) downloading the smoothed camera path used for photosynth 2 navigation/playback.
When you download photos from original photosynths (what your current plugin already imports) I'd also like to see original metadata downloaded and applied to the photos. Henri Astre's PhotosynthToolkit http://www.visual-experiments.com/demos/photosynthtoolkit/ already proved that this is possible, some years ago, although one must download the thumb.jpg version of each photo in the synth in order to acquire the EXIF data to apply to the full resolution copies which are pieced together from the tiles at the base of each image's multi-resolution tile pyramid.
Most of the photos uploaded to Photosynth are liberally licensed by the uploader with a Creative Commons Attribution license and by preserving the original photos' metadata when downloading the images, we retain any 'Author'/'Photographer' information they may have applied to the images before uploading, which is more than reasonable in my eyes.
Camera model and settings are also useful if one is going to further use the downloaded images for further photogrammetry purposes.
Also, if your importing plugin could pre-set the lighting in Meshlab and per-vertex color settings so that the point cloud immediately is recognizable as the same data seen on Photosynth (in the same way that opening the same point cloud with SynthExport or PhotosynthToolkit has the right settings when opened), I would greatly appreciate it.
Thanks in advance,
Nate Lawrence
+1 vote