Here's a RAM Disk from a company I'm familiar with. They have a 30 day trial.

See if it speeds you up?

http://www.farstone.com/software/virtual-hard-drive.htm

My experiences can only come in short bits at the moment. We have a toddler, so I'm only able to set up the cameras in our living room late in the evening, and then I have to dismantle everything before going to bed. Maybe this will motivate me to finally clean out our garage. :)

Anyway, once I have the workflow figured out, I hope things will go more efficiently for working in any reasonably sized space.



No, what you have it much beefier than my system. My workstation is a HP Pavillion, Intel Quad Core, 4GB RAM, and a NVidia 768MB Geforce 9600 GS (I think that's correct.) As mentioned, it's running on Vista Ultimiate (x64). I got this system last year after my old computer died.

My laptop is a three year old Dell e1705, which is an Intel Duo Core, 2GB RAM, and an 256MB Nvidia Geforce 7800, and running on XP Professional sp3 (x32). The video card is not compatible with the DMC software, but so far it seems fine for iPi Recorder.



Yes, this is my hope too. I'm a bit surprised that Recorder currently only writes to the primary drive, which, unless I'm not understanding something, seems like a bad idea.



Let me know how that goes. It might not be an option for me, but I'm very interested.

Thanks for posting the pic. I couldn't make it to Siggraph this year, but looking forward to next year when it's back in L.A. again. :)

Greenalw

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

Watch a one minute excerpt on Vimeo now!

Default installation of Microsoft's Windows Media Player should come with a functional MJPEG decoder. So you should be able to play MJPEG files on a computer without Morgan Multimedia MJPEG. Third-party MJPEG is important for recording, not for playback.

I think Logitech drivers do not allow that.

This really looks like not related to the speed of your computers.

If CPU was not fast enough, iPi Recorder would just stop recording after running out of its default 128 Megs of cache RAM.

If iPi Recorder displays low framerate during preview, this is a clear indication that your camera sends frames exactly at that rate.

You should play with camera parameters to increase framerate. You may need to turn off Anti-flicker, RightLight and basically all "auto" settings to increase frame rate.

Exposure is the most important camera configuration parameter directly related to frame rate. You should set Gain to max and Exposure to min possible value.

You may need to manually set Gain to max for all cameras. Logitech drivers seem to reset Gain to some random value after each reboot. This looks confusing sometimes. Probably we should change iPi Recorder to set Gain to max after Recorder restart as a workaround, as max Gain should be appropriate almost always.

You should aim to use lowest possible Exposure, of course provided that something should still be visible in resulting video. If you are able to clearly see actor in recorded video, our software will be able too, even if the picture looks a bit dark for your taste. Most people are used to high-quality digital video we all see on TV every day, but that level of quality is really not important for motion capture. Our software just needs some contrast of actor clothes and background to be able to tell actor from background; it works with color gradients, not absolute values.

Michael,

Thank you for explaining this.

After more testing, it does appear that can get the same frame rate with either the laptop or the workstation...there were slight variances in the settings, and the rates were comparable after carefully matching all of them. From what I can tell, either the workstation or the laptop are more than adequate for these cameras. (The real test will be when I get the fourth camera.)

I have to put aside iPi Studio for about a week because some personal things came up that I have to tend to, but I think I'll be able to focus on using the software afterwards. Hopefully, the Jimmy|Rig Pro beta will be ready for testing by then too.

Greenlaw

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

Watch a one minute excerpt on Vimeo now!

Hi,

Lots of questions:

I see in the Calibration Tutorial video, 'Auto detect initial camera positions' is disabled. When is it appropriate to turn this feature on?

I also noted that 'Auto adjust camera FOV...' is enabled, even though the FOV is manually set earlier in the tutorial. Is this correct? Does enabling this feature slow down the calibration process?

When would you use 'Recenter Coordinate System On Character'?

Regarding the option 'Enable foot tracking...' in the Tracking panel, why would you not enable this feature? Is this to be left off to speed up processing when the actor is standing in place?

When is it appropriate to use the 'Optional second tracking pass (High Resolution)'? Is this a more refined version of the standard tracking, or is it for higher res video? Same question for 'Refit Pose - HiRes'.

Thanks for any information.

Greenlaw

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

Watch a one minute excerpt on Vimeo now!

I just realized that some of my answers are hidden in the tool tips. I would still like to hear more details though. :)

Greenlaw

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

Watch a one minute excerpt on Vimeo now!

Auto detection works well if your calibration sequence has lots of good marker detections (more then 70% good points uniformly covering capture volume). If auto detection fails, in many cases you still can obtain good calibration by manually specifying initial camera positions.



Yes, it slows down calibration a bit. It is "on" by default because some users may forget to set correct FOV before clicking "Calibrate".


This is useful immediately before exporting animation. You may want to use it on first frame of your motion, so that motion starts in the beginning of coordinate system, instead of some random place in space.



Foot tracking is disabled by default because foot tracking requires accurately calibrated ground plane. Many novice users forget marking ground plane. Once you learn how to define ground plane, it makes sense to always use foot tracking.



Hi-res tracking pass uses sub-pixel resolution and results in much better accuracy even for low-resolution input video.

It is recommended that you use second tracking pass whenever possible. First tracking pass is lower resolution and is more a "preview" of capture. After first pass, you should visually check captured motion, and you may want to manually correct something and re-track some parts of it. Then you should apply second tracking pass. Second pass results in additional processing time, but it rarely requires manual intervention so you can run it completely unattended.

Michael,

Thank you very much for the detailed response! Your explanations for these features are very helpful and enlightening.

BTW, I did another session the other night, this time with better clothing, improved camera positions, and a fourth camera.

Here's some info that may be useful to other iPi users:

In Vegas Pro, I edited down my calibration video to just the frames I needed, which made it easier to manage in iPi Studio. When I exported from Vegas, the program informed me that 'No recompression needed', which was pretty cool.

I ran into a problem with my 'action' footage though. I think my mini-RAID failed and corrupted the data. The next evening, however, I thought to load the footage in the free Virutal Dub program, and found that it automatically repaired the data! I Stream Copy to export the video without recompression, and then trimmed the footage in Vegas and exported each 'action' as a separte video clip (making sure to include the clean plate at the head of each one.) This should make the data much easier to manage in iPi soft than trying to work with one very long clip.

BTW, in a future version, it would nice to be able to zoom into the timeline so I can accurately see the individual frames. Right now, with the full timeline always visible, it gets a bit tricky when trying to set the range and current frame to exact frames. As a reference, Fusion has a good timeline interface for this. Vegas Pro is pretty good too.

Calibration seemed to go well, though it took a very long time. (I can't wait to use the future 'all GPU' version with improved speed for this process.)

I really need to get my garage cleaned out so I can use the space for iPi shoots. It will be great to be able to position the cameras a little more permanently so I won't have to recalibrate every session. (My wife will be happy for a clean garage too; she really does care what my reasons for it are.) :)

I'll try to do some actual tracking before I go to bed tonight and let you know how it goes.

Greenlaw

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

Watch a one minute excerpt on Vimeo now!