....

Most anywhere other than Cable Matters they were upwards of $60 and more for a 10m cable.

I posted the link above for it through Amazon.

...

Ah, I see. Yes, Cable Matters is pretty good. I use one of those for the Bluetooth dongle since it doesn't need to reach so far into the room.

Here's the link for the longer cables:

Cable Matters® SuperSpeed USB 3.0 Type A Male to Female Active Extension Cable 10 Meters/32.8 Feet

Its twice the cost but also twice the length and with the option to connect a separate power source.

As mentioned earlier, daisy chaining multiple cables works well but with the setups getting more complicated, I'm looking to simplify and be able to store as few pieces of gear in our kit box as possible.

Tip: I took the time to label the new cable connections and the devices this morning. We've always found labeling very helpful when making connections and identifying which real world item corresponds to its virtual counterpart. For example, our sensors and cables are now labeled R01 to R04 to match each of the four render box they will be connected to. Previously, we labeled all our PS Move devices R, L and H to correspond to the body parts. Labeling the Move devices is even more useful because it leaves out the guess work when you start tracking.

G.

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

Watch a one minute excerpt on Vimeo now!

...

Just ordered a second Kv2 and adapter, should be here by end of the week if not sooner.

Actually I think I have enough reach for my area just with the cameras cables, since using a laptop for 2nd camera, I suppose the laptop can run through a WiFi connection on the network, I won't know until I get it all hooked up, but it should without direct Cat5 cable to router.

It may not be as completely accurate for very fast motions and feet placement as the Eyes are, but the trade off will be much faster processing times for most actions and much smaller file size for videos when using only depth data on recordings.

So far with one the results are very good, not sure how difficult the calibration will be for 2, but no worse than 2 - v1's I am sure and that wasn't too bad.

I will post about it after I get it up and running, but you may know with your setup sooner and post more on it also.

...

Cool! Good luck with your system.

The way we're set up at home, we keep our computers in an adjoining 'computer' room, and we run the cables from there to the living room. To view the master computer, I use a laptop or tablet and the free version of Team Viewer to operate it nearer to the sensors. This is how I control the motorized tilt in the v1 sensors during setup.

Actual operation, however, is usually done remotely using the buttons on the PS Move controllers. Since I perform the motions myself late at night, it's an ideal system for me. When Alisa or Sienna are involved in a session, one of us might operate the master computer directly.

Once a recording is stopped, sometimes I'll go into the computer room to rename any files if necessary. (For example, I like to add a 'Cal' designation for calibration.) If I have a complicated session, I'll record info about each captured motion in a spreadsheet--info like character, scene, action, take, etc. This has been super helpful for the music video because I need to record motions for three different characters, sometimes interacting with each other, for many takes, and for many scenes. ('B2' currently has about 60 scenes and most of it has mocap.)

This data entry part is based on the workflow I've observed at big mocap facilities like Giant and House of Moves on various video game jobs I've worked on. Often times we'd have only two performers doing many takes of motions for dozens of different characters, so it's very important to have detailed script notes for everything that's been recorded.

G.

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

Watch a one minute excerpt on Vimeo now!

...

Scene actions labeling can get a bit hectic I am sure, it is best to have some kind of rhyme and reason for organization.

Using the Moves to start and stop on your own can get a bit confusing, but reviewing and naming after isn't that hard.

I have to do it anyway when I edit the video files for length, I do this even on separate takes in the same scene recording, it wont accept the same name of course, so have to add and (ed) to naming, but I usually trash the original right after to free up space, since it's basically worthless after that.

First time trying this distributed recording thing, so I hope it falls into place well.

I am sure will take a good minute to get it working, I just hope the video sequencing works well now, they did have issues with it earlier.

...

...

I have been playing with the floor height and with 1 cam what I see is worse results when the floor is adjusted to actual grid height at actual performer height, I am just trying different things now.

I feel it works better when depth was left as cam sees it, or adjusting the floor down, (actually it moves the depth cloud when off camera view, moving depth cloud up), just slightly and increasing the feet size. This would be pure preference to do, as well as anything anyone would do actually :)

I do feel I get pretty good results overall with just setting actual actor height/build and leaving floor as imported.

The actor and performer cloud don't line up precisely at performers actual height, but that really doesn't affect the tracking and just using a much different actor height than actual performer doesn't work well, what affects the tracking worse is the floor height, especially on foot tracking, squatting or bending moves.

I use the same center point of the room for alignment and the actor appears very close on file load into the Studio, which is good.

I guess each user will have their own interpretation of what works best for them, I just thought I would mention what I am experiencing, but 1 camera is limited in capturing, so can't wait to see what 2 cams will do.

I am also trying using a black material on the floor, so the floor returns no data (yellow)(this doesn't change the floor depth the camera sees though) and will only show the performers data, concentrating on better foot tracking, prelim results are looking good and I can see the whole foot depth/size better, where before the toe was always not shown when feet were touching on floor. If it continues in tests to work out better, that's what I will use. You may think using remove background in tracking scene may do the same thing, it doesn't, not to me anyway.

This may all change with 2 cams, but if it does I will adjust to that as well.

...

Yes, in my experience, settings can vary from project to project depending on the motions I'm capturing. Sometimes I find I need to adjust the actor height during tracking to get the results I want. It all depends on how the internal collision objects need to interact with the capture 'shell'. Changing the actor or floor height can produce different solutions.

The things to watch for when using Kinect is reflective surfaces and fully matte black surfaces. Reflective surfaces will bounce the IR rays away from the sensors and fully matte black might not bounce them back. Also, very bright indoor light source can interfere with the data, and bright sunlight coming through a window can really mess things up.

We have a shiny wood floor so the carpet helps there. It has a little black in it but I think the fuzzyiness and glancing angle makes that okay.

For the windows, I pin a blanket over the brightest if I'm recording during the day. In our computer room we installed 100% blackout shades to protect our UV sensitive chemicals (an unrelated project) and we're thinking of installing these shades in the living room if we do much recording during the day. But usually, we just wait till evening to record.

For indoor lighting, we turn off any bright lights that are within body level. Good ambient lighting is not nearly as critical for Kinect as it is for PS3 Eye but it's still something to think about.

G.

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

Watch a one minute excerpt on Vimeo now!

...

These are all just personal tests seeing the reactions, since I haven't used Kinects much at all, but would like to more if I can get good results now with v2, and yes, every session/settings could be a bit different from the last.

If any of this help some users, great, as I am kind of new to Kinects also, if it doesn't, they will soon learn on their own anyway... I am a trial and error type I guess, to learn differences and when I need to adjust for differences in each session ahead of time :).

The actor scaling could be adjusted at different points in the tracking, that's an option when needed, I would just rather leave that alone if possible and have it track relative to the actual performer throughout knowing the capture movements limits, since I am only going to be using 2 cams... that is one good thing with processing in iPi, changes can be made right inside the program easier than other systems.

The black non-shiney material on the floor is working well, (jury is still out though), it doesn't return any data on the BG eval. or the tracking, which allows more of the full foot length/depth to be seen, which is seeming to be better for alignment throughout, especially when touching the floor.
Without it, it seems to me to want to push the feet off the cloud more with the one cam, and the toe bend data is still iffy if it returns correctly either way, but no big deal there.

Results are good, much better than V1, so that is a plus in its self, I just wanted to test different scenarios and methods since I have not worked as much as others with the Kinect v1 or v2, and post the outcomes I see, and your right, it all depends on the person and the capturing situations, and their expectations.

I can deal with any tracking issues, but I like less of course, and I am impressed with the v2 tracking and export quality so far, and not worrying with clothing color, this will make it easier to get different actors to perform and even an easier set up to make portable to go to them, I may need to get another laptop for travel though.

...

Yes, one of reasons I'm a big fan of iPi Mocap Studio is that it offers many options to suite many situations. We started out with web cams, which worked okay but then PS3 Eye camera support came along to simplify things greatly. This was when the system started to become practical for professional use. The problem we had with PS3 Eye was the space requirements. We were using our garage, which was just big enough, but of course we need out garage for other things so it wasn't always practical to set up in there. Especially since I also had to move computer gear back and forth.

When dual Kinect became available, that changed everything. Setup could now be done in our house without moving our computer, and it took less than five minutes for us to start capturing. Even though the system was assembled for our own personal film projects, I found it incredibly helpful on a few freelance jobs. And my wife is pleased that we can take it down in five minutes so we can have our home back. :)

That wasn't possible with the PS3 Eye system--not for us anyway. However, we keep the PS3 eye cameras handy just in case we need to record motions to cover a larger area. For that, I expect we would set it up in our backyard and record just before sunrise. So far, we haven't needed to go back to the PS3 Eye setup though. I'm hoping that the wider FOV of the v2 sensors will make it even less necessary for us to fall back on our old system.

G.

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

Watch a one minute excerpt on Vimeo now!

...

The v2 does seem to capture faster motions better, but it may and probably will, in my tests anyway, loose the tracking now and then, you just have to stop and re-adjust

the IK and it catches right back though, but after I reposition/refit, I run track backwards a frame or 2 or it will just sling the correction back off track again if you try to go

right to track forward, you are probably familiar with this though.

Yeah, I am lucky enough to have just enough space for the Eyes and to leave them permanently mounted, so I can switch to them at anytime in the same space.

...