Yep, your right it auto updated as soon I booted up the program. Now I have to order 2 more cameras. Have you noticed any better tracking with the addition of the extra cameras? Thanks...

Yes, I think so.

But, the calibration is important. I think the virtual cameras should line up and the ground planes should look correct for each camera. Make sure there are no reflections. Make sure lights are not in the path of the calibration (you can turn them off after your t-pose which is needed to initially align the cameras).

And I may need to re-think the distance from the ground for each camera. I'm not positive about the best heights, and some have to be about 6 feet up so they can see the full scene (as I have limited space).

I'm not exactly sure what you mean here. The motion video actually needs to be much better lit than the calibration video (which should be pretty dark when you switch to Calibration Mode,) but I'm guessing this is not what you meant.


My setup is a semi circle so I don't know how this info relates to your full circle setup exactly, but my outside cameras are at least six feet high, while my two inside cameras are at maybe four or five feet. When start using the two new cameras, I'll probably vary the heights a little more. I think the main thing is to maximize the coverage but not to have the cameras so far apart that their is no overlap. (The overlap helps the software figure out the relationship between the cameras, improving calibration and tracking.)

Hmm. Maybe that last bit of info might help you out? Let me know.

G.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!

I meant that you can turn the lights off during calibration after recording a t-pose. This way the flashlight won't intersect with any lights.



Makes sense that overlap would be a good thing. I wish I could place them in a circle, but I just don't have enough room. It's a rectangular space with the cameras at each end, but none near the exact middle of the long sides of the rectangle.

BTW - did the last 'gold' LW10 fix any of the remaining FBX issues?

Oh, I see now. You don't need a T-Pose for Calibration though. T-Pose is just for when you start recording motion.


I'm still testing the latest FBX IO but one important fix that came recently is that the axis orientation is now correct for Lightwave (y up) which makes it totally worth using over LW 9.6 for my iPi Studio-Animeeple-Lightwave workflow. Newtek also added Endormorph/morph map support. I'm not especially concerned with morph maps in FBX right now but it does mean they started working on vmap support. That's probably as much as I'm allowed to say in public until Lightwave 10 is officially released.

G.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!

I'm still testing the latest FBX IO but one important fix that came recently is that the axis orientation is now correct for Lightwave (y up) which makes it totally worth using over LW 9.6 for my iPi Studio-Animeeple-Lightwave workflow. Newtek also added Endormorph/morph map support. I'm not especially concerned with morph maps in FBX right now but it does mean they started working on vmap support. That's probably as much as I'm allowed to say in public until Lightwave 10 is officially released.

G.[/quote]
Isn't the T-pose needed during calibration to initially position each camera so that it is close to the correct orientation?

Or, are you letting the calibration auto-detect the camera positions?

No, the T-Pose is only used to set up the rig for motion tracking, not scene calibration. You don't need to mess with the T-Pose until after you've created and saved your Scene calibration file.

After you've tracked a motion, you can optionally reset the orientation of the scene to match the 'front' of character at any frame, but this has nothing to do with the original calibration. (Sometimes I've had weird results when using this feature so it's a good idea to save before clicking the button.)

Hope this helps.

G.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!

I don't follow you. The iPi tutorial video shows using the t-pose to 'very roughly adjust initial camera positions'.

Maybe I'm misreading it, and instead of focusing on having the t-pose virtual character match the t-pose actor from the calibration (to get the camera 'positions' correct), maybe I should just focus on the camera positions themselves?

Possibly all the orbiting and positioning I'm doing before starting the calibration is messing with my ground planes (but the cameras have been lining up perfectly), as I do try to match them up exactly.

Exactly. The T-Pose has nothing to do with camera calibration. It's only for rig tracking after you have established good camera calibration.

Hope this helps.

One of these days (hopefully soon,) I'll post one of my own calibration sessions to show what I'm doing.

G.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!

Just to be clear, the only thing you need in the calibration is a few seconds of clean video at or near the beginning and followed by the maglite motion. Remember, you need to set the range for the clean video (if it's not at the very beginning.) This is important because it's used to help key the person out of the calibration video.

Before starting the calibration you should enable Autodetect Camera Positions, which should automatically position the cameras for you. If this fails, then you can move the virtual camera to roughly where they should go. (FYI, I believe I've only needed Autodetect with recent builds of iPi DMC. This probably depends on how clear the video data is DMC.) Don't bother with setting FOV or using Autodetect FOV if you're using PS3 Eye cameras; just use the default setting of 75.

Don't bother setting the camera scale/fine tune until after the calibration process is done. Now click the Calibrate button. When calibration is done, find the ground points by scrubbing through the video and then define them. (Use the < and > keys to step frame by frame.) Choose at least three ground points to define the ground plane. If you only choose three, they should be in a triangular shape and not a straight line.

Once that's done, set the scale of the scene by matching as closely as possible the height of any virtual camera to the height of its corresponding real world camera. (Normally, you would choose camera #1, but any camera should be fine if you have the correct height information.) Use the Fine Tuning slider for additional precision in setting the camera height.

Calibration is done. Click Save Scene. Using Save Scene is different from selecting Save from the file menu, which saves the project file. This Scene file is your calibration data so you may want to name it accordingly.

When you're ready to do some tracking, load your action footage, set your clean video or background range, and then load the Scene file using the Load Scene button to position the cameras to the calibrated data.

It's at this stage you will set up the T-Pose for your Actor. After setting the pose, remember to save the Actor for this scene. You can reload this T-Pose for additional scenes from this session, assuming your real world T-Poses were generally assumed at the same spot. (When shooting your action videos, it's helpful to mark the spot with a piece of tape on the floor for your T-Pose. This consistency will speed up the tracking setup.)

Bear in mind that I've only done this using four PS3 cameras, but I imagine the process is the same for six.

Hope this helps.

G.

_________________
Greenlaw
Artist/Partner - Little Green Dog | My Demo Reels (2013,) (2015,) (2017,) and (2019)

Watch a one minute excerpt on Vimeo now!