Do you have an idea for an awesome feature we should add… or hate the way we’re currently doing something? Share your ideas and suggestions here.
Is it possible to add new photos to a synth that's already made?
Not what you're probably thinking of, no.
It is, however, easy enough to make a synth with the old photos and new photos all in one. The calculations will take as long as normal, but the uploading should be quicker than normal for that total number of images as it won't re-upload anything that you've got on your account already.
WHOA! NOWAY! So if I've got multiple sets of images from a single location, different days but all vaguely at the same time of day, I could potentially generate synths from combinations of images without having to re-upload the images themselves? HMMM...
Does this depend on the pathnames to the files? I tend to make my synths with folders sitting on my desktop, then shuffle them off into my photography archive tree once the initial work is done with the photos.
Thanks for pointing this out, Nathaniel!
Tom, it's dependent on the hash signature (see http://en.wikipedia.org/wiki/Hash_function ) of the file being the same as the last time you uploaded the photos.
Put simply, if you've edited the file at all (even just added one tag in the metadata - changing none of the actual pixels) since the last time you uploaded them, then they'll re-upload again, since it is assumed that if you have updated your file then you want the newest version of the file to be used in your new synth.
I suspect that the reason is that when architecting Photosynth's backend, they thought of people using Creative Commons photos found on the web in their synths. Really if multiple people are just downloading the same photo and uploading the DZI to the Photosynth servers, they only want to host one copy of it.
You will find that it doesn't even need to be you who uploaded the file before. If anyone used it, and you put it in your synth, it will skip the upload.
For really interesting reading, check out Bitplane's ideas here:
And if I didn't make it clear above, no file path shouldn't play a part, otherwise they would get redundant uploads, since everyone would store their copy of flickr photo x in a different folder.
I keep thinking of little things that relate here.
The ONLY problem that I foresee with throwing in a heap of images from multiple takes of one site (especially in your case, Tom, where you literally have hundreds or more than a thousand shots from each take multiplied by the fact that you almost always shoot extremely image-feature rich subjects) is that you may have so many features that the synther will crash.
The benefits would be that the combined pointclouds should be far more robust in density, as well as color accuracy. If you compare the pointclouds of two synths of the same subject - one under direct sunlight where it's difficult to avoid overexposure and one where you have receding rainclouds and all the colours are rich and vibrant due to recently being rinsed, as well as being lit with scattered light from the clouds, you'll see a massive colour difference, not only in the photos, but in the pointclouds.
Combining shots from throughout the day or different days should give the pointcloud a more accurate rendition of what the objects' colours actually are, rather than what they are under one specific lighting condition.
I had one last thought here, which is that everyone should use a photo tagging program (like Adobe Bridge, Microsoft Pro Photo Tools, or Windows Live Photo Gallery) to add, at the very least, their name to the 'Author' or 'Photographer' field before making their synth.
Even if you don't want your real name associated with your photos on your Photosynth account, you could at least add a nickname or just your Photosynth username so that it appears next to your license symbol in the lower lefthand corner of your synth.
Again, it's better to do this before you ever use the photos in a synth because if you use them before adding your name and then later decide that you want your name next to your Creative Commons or Copyright symbol (after all how will people attribute the photos to you if they don't see your name next to the (cc) icon?) you'll have to re-synth and re-upload every photo that you've added your name to.
Hahahahaha! NOW you tell me! ;)
I got one a while back with just that in mind, but like a doof I haven't been using it on images I'm using for synths. >sigh< I need to re-vamp my workflow...
I may play with this with all the photography I've done at Kiholo Bay, just to see how far I can push things before it gives. I've got 4GB of memory on this machine, and a pretty nice processor, so it may let me stretch things before it crashes. We'll see!
Here's an example of what a synth looks like when the shots are taken on different days with different lighting conditions:
http://photosynth.net/view.aspx?cid=f3ae070d-096b-462a-b93f-20f5dd7b275e (Glacier Point, Yosemite National Park)
It would be good if it was possible to 'pull' images from another synth and add your own to create a larger synth.
That idea was with http://photosynth.net/view.aspx?cid=f3ae070d-096b-462a-b93f-20f5dd7b275e in mind!
an option to hide the white outlines...so it looks more real. Also the option to take photos to allow the user to almost walk down a street or through a house.
Here are two conversations more along the lines of what you're thinking about:
As to the discussion from above, I don't believe that editing the filename is enough of an edit for the synther to re-upload a photo. It has to be a change in the metadata or the pixels, if I'm remembering correctly.
Like you can go fowards with arrow keys aswell as drag with the mouse to see the panorama in google, i think you should be able to do this in photosynth so that its not just roads covered, people could actually go out take photos of a stretch of canal or a place where they walk their dog.At the moment photosynth doesn't really allow you to move about.it allows you to turn around and zoom in.You could incorporate the way google street view navigates, into photosynth, but allow the user to move in any direction and not just foward, back, left or right.
Sometimes photosynth can get 2 images that are a fair distance from each other but because they look similar they can be joined by mistake when they are two completelly different areas.I think photosynth as smart as it is, it could get alot smarter, and the possibilities are amazing.when do Microsoft live labs plan on taking suggestions into account to start working on an update to make a much more flexible versi
Dean, are you talking about the panoramas that are hosted on Photosynth of late or actual Photosynths?
It is absolutely possible to make a synth where you move around or down a road or canal, rather than stay in one place and turn around. It's been possible ever since Photosynth first came out.
Here's an example of moving down a canal (just keep pushing the spacebar key to see all the photos):
Here's someone walking along a road. They only walked one way and not particularly quickly, but try out 'Overhead View' from the drop down list and click on any bit of the road to swoop down to that view.
If you're just wishing to have multiple panoramas strung together, then I believe that you're definitely on the trail of what the Bing Maps team is up to (that's where Photosynth moved after it launched out of Live Labs). Plans to link nearby synths together have been around for a while now and I have to think that when it happens, it will include panoramas as well.
For more information about the Photosynth team's intention to allow multiple synths to be linked together, try reading this topic:
Hopefully we'll get to the point where we can jump from one synth to another and from there to a panorama all along the same street in Bing Maps, even out to the official Streetside panoramas, if they've covered our area yet.