Attached Link: http://www.ssontech.com/

I use Syntheyes for 3d tracking for use with Lightwave, and it does a very good job. I was just at their site and see that they have a new forum, and that there is now a Poser 5 export script available. (Forum/annnouncements). Thought that you videocamera types might like to know...

Message edited on: 03/28/2005 16:14

fantastic. I have been watching and shopping match move software and have been planning on purchasing Pixel Farm's low end (under $100) product to play around with Poser and Video. That's why I've been asking about lighting Poser animated characters in Poser lit with IBL and composting in video. But it does not have direct to Poser export. I am headed to Syntheyes now. Thanks krimpr. ::::: Opera :::::

Attached Link: http://www.thepixelfarm.co.uk/

Attached file for The Pixel Farm, they have three or four levels of their system between $100-$4500. For Poser people who don't know what this software does... You have video. Perhaps an outdoor scene. It has to have movement...not just a static shot. You run the video into the motion tracking (aka match move) software. You have to tell the program, with mouse clicks, the location of prominent objects in the video over frames. The software figures out how to track that object. Then, you have to calibrate or give measurements.. You also have to indicate the ground and 'what is level' You export a "point cloud" file, which your 3D software can read and translate into 3D. Then, you can move your character around in the animating point cloud. You render the animation, and composit it with the original footage. Voila, integration of your animated character and the video. You can see why one would want to light the animation of the character with IBL from the original footage. krimpr, is that an accurate description? What file format does Syntheyes export that Poser can read, and have you tried it? ::::: Opera :::::

Attached Link: http://www.ssontech.com/mmove.htm

Here's a great page on the Syntheyes site that explains it better than I did. ::::: Opera :::::

That's basically it Opera, but Syntheyes actually simplifies it even further. You shoot your video, and then import it into Syntheyes. Often, all that you have to do is ask it to "autotrack". I'm working on a shot now that took my P4 2.4 about 20 minutes to analyze. It reports back with a report of possible "targets" which it has assigned to various points on your video. These are indicated by yellow "X" marks on your clip. You choose one of them to be an origin, and then a couple of others so it can calculate relative distances between them. (You choose a tracker by clicking on it and fill in approximate X,Y,and Z co-ordinates relative to your origin.) Once you have input this information, you rerun the track, and it then computes what movements you made with your video camera, which is exported into a Lightwave, Cinema, Houdini, Truespace, Particle illusion.......Poser scene. In this case, Poser's main animating camera would track as you had shot your live video. Now when you load your character into Poser with your video in the background when the camera moves, it is including all of the proper perspecive in relationship to your character. With the new Shadow catcher, you should be able to render directly right from inside Poser. Now Vicky can be rendered in a real world temple, complete with fluid, live action footage. It has been used for alot of big screen productions, including most recently "The Aviator". I just received an email saying that the new script is for P6 export. Great... now I need P6.

That is fantastic, thanks for the detailed rundown. The cool thing about P6 with regard to this concept is: IBL, or Imaged Based Lighting. That means this: You can plug an AVI of the original video into certain nodes in the material room. Then when you move Jessi (please forgive the departure from Miss Vickie) to stay in frame with your transitioning camera, she will be lighted automatically in harmony with the corresponding light conditions of the frame playing in the AVI! How cool is that for integrating the animated model into the footage at composite time? Can SynthEyes do any "reconstruction"? That is, can it generate textures based on the video footage and somehow incorporate them into the exported file and apply them to props spawned from the point cloud, so you actually "see" a 3D construction of the 2D video? The software is about $359, right? ::::: Opera :::::

During composite...or as you say, right in Poser6 with vid running on plane in background...I just 'groked' that.

The reason I ask about reconstruction is this: Instead of running the vid in the background...if you actually use the tool to GENERATE a 3D world, then your original footage is not frozen; it is merely the base from which you can fully animate, including, obviously, changing your mind about the camera movement.

::::: Opera :::::

Message edited on: 03/29/2005 00:55

Message edited on: 03/29/2005 00:56

Opera: Syntheyes' sole function is to recreate camera motion; like a pantagraph function for motion in time. In Lightwave (my only experience with integration thus far) it generates keyframes which are visible in both the timeline and in the graph editor. It also exports "nulls" (which are trackers which are chosen during the tracking process) of your choosing during the "export scene to"...Lw, after effects, etc., which can (again, in Lw at least) be replaced with objects of your choosing. Eg; if I exported a scene in which I had inadvertantly tracked a fire hydrant as part of the capture and had exported the associated null with it, I could replace the null in LW with a tree or an iguana, and it's perspective in relationship within the scene would be maintained. The camera motion can be edited via exported keyframes, but you would lose your perspective "link" with the running footage. It would be akin to the "slipping feet syndrome" with walking characters. Syntheyes effectively locks their location in space and time, (at the risk of sounding Trekky.) I'm not sure if that answers your question....
Yes, it's $349 and includes free upgrades (I've had two so far) and an extensive feature set which at this point is far beyond my rather feeble mental grasp and, for that matter, my needs at this time. 3D World (or maybe it was Digit) did a large article about it a year or so about it in which it was favorably compared to $9,000 software, so I thought it was interesting enough to take the plunge. I haven't regretted it. It also has about 15 exporter formats, and as new ones are developed they are put up in the downloads section and are free. Not often you see this type of customer loyalty. If you are looking for this type of software, I highly recommend you giving it a hard look.

Attached Link: http://www.realviz.com/purchase/index.php

I'd bet the $9000 sys it was compared to is Real Viz MotionTracker Studio. Link above. for $349, the features described... SynthEyes is WAY worth it. Besides solving for the camera in the automated or manipulated modes you described, to have the nulls to at least place objects...that's pretty darn good. And now the direct-to-poser script is great. Since there is a trial package of SynthEyes, you can bet you life I will be trying it after I get Poser 6 stablilized in a week or two. I would not change the camera keyframes, as you pointed out, if I were running back the camera animation against the original footage. The 'reconstruction' idea is something else..another app further along beyond match move. See next post. ::::: Opera :::::

What if you could generate a full blown 3D scene from photograhs, including textures dirived from the originals, in an automated way.

That's an exciting idea.

The example below starts with a photograh of a famous painting, that of Vincent Van Gogh's room in Arles. This is the photo of the painting:

Using a piece of software called 'Canona', which I now believe is either dead or has been purchased and incorporated into someone's reconstruction software, you could achieve this dream. It would be fully importable into your 3D app, hopefully including Poser.

Click here for Quicktime movie, 1MB, opens in new window.

Here is another example, made from photographs taken at a chateau/castle in Paris. Click here for Quicktime movie, 2.4MB, opens in new window.

By the way, Real Viz has a program called Image Modeler ($1380) which supposedly does this. Link here (opens in new window).

::::: Opera :::::

Those two Canoma samples are pretty cool. I've seen it (Canoma) from time to time on Ebay at ridiculously low prices; next time maybe I'll snag it. I've also read up on Image Modeler; it was offered free a month or so ago with any upgrade version of Lightwave, but it looked like something for down the road for me right now. Good luck with your matchmoving quest, I'm sure you'll have success whichever path you travel :)

You could do Canoma similiar method in Poser using UV Mapper.