stealthtech's dashboard

Along those lines the ability to have a private none public synth. Also allow more control over sharing and who you can share with. The ability to password protect and share that way as well. This way one does not need to go through the steps to be a member just to view. I can have a pasword protectd synth and share that way as well. Two steps view the site enter password. Not the twenty steps to become a member and fill out a profile (if one does not have a live account) and now you can see the synth. I don't want to ham string people into becoming a member just to view a private synth. I can hear the naggers now "Why do I have to become a member to see a private synth."
Some of this can't be done with out a friends list and so forth. But maybe at least a private synth and maybe a password protected as well.

IMHO the only reason not to have lengthy uploads is in case the output is not what you want. This is one reason for a preview before upload.


IMHO it is import to have as many pixels as you can give it. The time to upload should not matter if the quality of the outcome is better. As David said the preformance of viewing is not impacted by the size. A one time sacrifice for a long upload with better quaility outweighs a short upload time with bad quality. To the viewer they don't know if it took you 1 hour or 20 minutes, but they will know the quaility of one is better over the other. Viewing the two different ones will have no performance impact just quality. I think this is also why we have 20GB of storage.


But again the issue is not being able to preview which will then turn that half hour into wasted time because of a bad synth. I'm doing one right now that is about 80 images at 100MBs. I hope it turns out ok.


Think of it like raytracing (Toy Story) I can have the computer spend one day to produce a high quality movie or one hour to produce a lower quality. But playing either movie will still play back in real time.

Ok I did some sort order changing.
Take a look using the Grid View [~]. And use the space bar and shift space bar compared to the . (period) , (comma). Then the same in the 3D view.


The . (period) and , (comma) adjust to the changes. The space bar and shift space bar do the same which produces a different output?.?. I'm still checking.

With out re-order


With re-order

I got ya, I think the "Rule of 3" might be along those lines. If I understand now, it is when you go from lets say looking/walking forward but the point of intrest (poi) is on the right of you. How do you go from forward to the right. I think the texture matching is based on a texture appearing in at least three or more different images. I start to shoot pictures to combine objects together as I rotate until reaching the POI. Making sure object A (somthing in front) and B (half way to poi) are in the same shot then B and C (POI). Then I can focus in on C. This might require like 8 to 10 pictures. I Just start taking pictures to make sure that I have an object in at least 3 photos each time I rotate. Like you say "take three" if not more. I know it kills the upload time but the more the better. Am I closer?

Just got finished. I don't know the results yet hot our of the oven.

Here is the one with out sorting.
No sorting right from the camera (FireMantel)


Using the sorting application and grouping.
Sorting using Synth Renamer (FireMantel001)


I also noticed the image used on my home page used two different images.
My Home Page. If you look at FireMantel001 w/sorting compared to FireMantel you will see the thumbnail image is different.

I'm working on the same idea to see if there is a difference. If you look under the What Features Should We Add to Photosynth


The challange is renaming 60+ images I wrote a little application to help. I'm working on a zone or group idea representing each distance from the subject. As you pointed out you have pictures at 5 paces and some at 10 paces. These would be Group A and Group B. The challange is that if you are at 10 paces and decide to go back to 5 then back to 10 then the pictures are out of order.
This is all in theory. I'm working on some test. I will point out and others can confirm or deny but I think your subject might be in question because of the textures. Look at the PDF guide. Page 7 shows a single tree (good) instead of a bunch of trees (bad).
The Photosynth Photography Guide (1Mbyte pdf)


Another thing is to study the master or grid view [~].

I just watched (scanned thru) a video on Photosynth from Channel 10 posted on the blog. It is about 19 to 25 minutes into the video. I think it is you Blaise and Scott (good video) explaining that the synthing is done on the PC and then uploaded. Funny the upload takes longer than synthing.


Meet some of the Photosynth team


Which is why another post states there is no "queue" or "pipeline" per say. Nice insight into the engineering challanges.

The 3d free-flighT stuff can be done using the w-a-s-d keys. One draw back to the view application is that a lot can be done with the keyboard but not know unless you read it somewhere. I read about it in the PDF guide. The viewing application needs more "help" and I mean navigation help a quick tutorial. Reveal the key board short cuts and make them more up front.


I can see the composer rendering a preview. Might not be the end result be something close. Or download a preview from the cloud. Before uploading to the cloud. This comes back to the question of what benifit does the cloud offer or what is the clouds role? Is it is only for hosting and publishing or does it do the actual work too? If the work is not done in the cloud that would mean the work is being done on my PC. So why not preview before upload? Now if some of the stiching, matching and figuring out textures is done in the cloud then I can see why a client side only would be challange. Does this process require a lot of horse power that is obtained by uploading the images to the cloud? I understand "what" is going on with Photosynth I just don't know "where".


If it does require horse power then I could see a comercial entity buying the servers to produce and host the synths. Like selling Surface, but for consumers that might be a bit more challange.


I'm all for a private and controlling the sharing of synths. Just having the C on the lower left corner is not too much of a warm a fuzzy fealing. Also for privacy.


I compare this S+S to the likes of DeepZoom the difference is the I can *choose* to upload to PhotoZoom or not. Or I can host somewhere else or just keep it local for an Intranet.

Ok, I was going to run some tests and see the difference in the 2D picture view. See what the order of the pictures are and if it changed from before. The challange is renaming 60+ image, so I wrote an application. That took some time.
Now that I'm ready to test. It looks like the site is being worked on and/or it is midnight (pst) and it is time to stop.
I will try tomorrow.

Here is a follow up on the sorting idea.
What I'm going to try is to come up with a naming convention that will sort the way I intend it to be. The issue is that after stepping in for a close up if I need to step back then then those images are our of sequence. I'm using the focal lenght or the subject length as part of the file name. I use three zones, far back, medium, and close up, and maybe a fourth (very close up).

filename_slg_a_00001
filename_slg_a_00002
filename_slg_a_00003
filename_slg_b_00001
filename_slg_b_00002
filename_slg_b_00003
filename_slg_c_00003
filename_slg_c_00003
filename_slg_c_00003

Subject Lenght Group A thru C

A room synth would be a bit different. The image in the center would be group A. Then the corners and side wall will be group B. Then Group C and D would be zoom in of subjects in the room. Base on the PDF guide.

I'm hoping that renaming the files into groups will help. If I took an image in group c then figured I needed to take an image that belongs in group b. Based on existing sorting b would come after c. It should have been before. Renaming the files would then make sure that the images are sorted correctly. instead of being after 005 in the sequence it would before.

One other issue I found is that not all cameras work well with the file name. My cell phone camera made the images out of order. I think I got a call in the middle of synthing (damn them interupting my synthing).

Maybe the GUI should have a way to group the images in a subject length.

I go back to the DeepZoom composer. You could have the images imported. Then build a canvas or various subject lenght groups then put the images in those groups by dragging and droping from the import image list.

Blaise, good becasue I left to get something to eat came back and guess what. I'm looking at "Publishing synth." My last post was about an hour ago. I'm going to quit and try again.
The last one made it but some are still missing "Faild Publishig". Am I now not going to be able to upload and use a name used during a failed pulish. Last time I got a snyth already exist.
Off to do some synthing.

Same here, then when you try to upload another one with the same name you get an error of duplicate synth.
If something is going on in the cloud could we not see what is going on? Can we get a view of the pipeline/queue that we are waiting in?
That way we at least know that the synth is in the cloud but not ready?
A status at least (uploaded, waiting to be published)?
It is hit and miss. I have started a new synth, uploaded, and published quicker than waiting.

I was wondering that myself. If it has some importance should that be clearified with the community? I was figuring that the GUI did not do it there for it was not important. Could I suggest that may adding the abiltiy to re-arange by dragging and droping images. Or some GUI icons to represent chaning of the order. Again if it is not a big issue then no problem. But if it is an issue to have a zoomed in image first instead of a panned out shoot then it would be nice to re-order the images. Then in the composer would be nice to do it before uploading.

GPS, The ability to transition from one Synth to another. Synth 1 is the room Synth 2 is the lamp in the room. I would like to transition from Synh 1 into the Lamp.
Tagging
Mac/Linux or Just make the Client SilverLight plug-in. I know there is some DirectX 3D stuff for performance preventing cross platform.
Is the building of the Synth on the client or in the cloud? what is the purpose of the cloud? I would think it is done in the cloud but I read thing that contradict this. If not then how about a pure Client Side for hosting elsewhere? Or for an Intranet/private projects? If it is on the client then the preview before uploading would be nice otherwise you can't preview until the synth is created in the cloud. But it can still be done, after uploading have a preview with a commit. Or the work around would be to delete.

And when can I get Mac applications to run on Windows. The price we pay for picking the platform we choose. Is it MS responsibility and/or Apples to make thier software work on all OSs, or is it the user who made the choice.