I'm planning to buy a good mocap system.
Just think about Ipisoft with 16 Go Pro Cameras because it has 4K resolution.

Would this avoid feet jumping?
Is this the top configuration to work with ipisoft? Is there another best setting?

I wanna get the best result to prevent working over the mocap.

Silvio Toledo
Stairsfilms

Hi

Higher resolution does not provide any benefit for our system, but may increase processing time. What's good about action cameras is they provide higher FPS which allows to capture fast motions more accurately. Also they obviously do not require any PC for recording. However, getting their records ready for processing is not as straightforward as with web cameras connected to PCs. Because individual records must be converted and synchronized into a single multi-camera video.

Our tracking algorithm does a good job preventing feet slide and similar effects with any cameras. Just enable foot tracking and ground collision options when tracking.

FYI using 16 cameras makes sense only for tracking of multiple actors. If you only plan to capture single actor, 6-8 cameras should be sufficient. Less cameras = less processing time.

Check out comparison table for various camera configurations which can be used with our system. Maybe it'll help you to make a decision.
http://docs.ipisoft.com/Multiple_Depth_Sensors_vs_Web_Cameras_vs_Action_Cameras_Comparison

Hi Silvio!

Good to see you here.

I don't have experience with Go Pro or other action cams yet because I only have a single Go Pro camera, but, what vmaslov says is correct: frame rate is much more important that pixel resolution. In general, you will want a higher frame rate but keep the resolution relatively low if the cameras are going to be near the subject. How near depends on the capture space you have and the quality of the lighting conditions (you basically want even lighting with minimal cast shadows.)

The only time higher resolution is useful when you're shooting in a large space and the subject is pretty far from the cameras. This is because the higher resolution provides more pixel data when the human figures are very small on screen.

As mentioned by vmaslov, the downside with higher res and frame rate is that the data size will be significantly larger. This can slow data transfer and tracking and you'll a beefy computer system(s). If you're in a hurry, you will need to decide if all that extra data is really worth the extra time it takes to process it. If the motions are simpler and slower, probably not. Fortunately, you can choose the settings that you feel are appropriate for your needs.

I think most user like the PS3 Eye cameras because it's a good balance: the frame rate is decent (60 fps) for many fast motions, and the cameras are reasonably easy to set up in a large (20 x 20 feet) space. The cameras are also inexpensive and you can connect multiple PS3 Eye cameras to a single computer. The potential downside is that it can be difficult to set up several PS3 Move cameras in a small room space because of the lens and lighting requirements.

In my situation, I'm still settled on multi-Kinect 2, mostly for convenience and because of my small shooting space (I usually record in our small-ish living room.) Because Kinect 2 captures 3D data, you can get away with using fewer devices in a small space. Also, lighting issues are almost non-existent with Kinect 2 since the video data is not even used by iPi Mocap Studio for tracking, it's only optionally there for visual reference. Multiple Kinect 2 is very easy to set up, it's basically plug and play and ready to go in less than five minutes. For most mocap data I've needed, this setup has been fine, but my needs may be simpler than some users. I've used Mocap Studio for mainly for small personal projects but have occasionally used it to create vfx for TV/film productions. The downside to Kinect 2 is that the frame rate is half that of PS3 Eye's frame rate, so it's less accurate for fast motions; also, you need a separate computer for each Kinect 2 sensor.

(I've been using dual-Kinect 2 devices for a while now, but I recently added a third capable computer. I'll post some triple-Kinect test results over the upcoming holiday vacation.)

I know some people are concerned that PS3 Eye and Kinect 2 is EOL by the manufacturer. This is true, but PS3 Eyes are still readily available and a great value, and Kinect 2 it's still the best option for depth-sensor recording.

The action cameras offer higher specs, but they might be more 'fiddly' to setup than PS3 Eyes.

For depth sensors, there are a few newer alternatives out there but so far the specs are only a little better than what the original Kinect produced. I'm going to stick with my Kinect 2 sensors for as long as they keep running, and hope that something better will come along before they die.

If better sensors don't arrive before then, I'll probably switch back to PS3 Eye or move over to one of the action cameras. Probably not Go Pro though, because of the cost, but maybe one of the cheaper alternatives so long as they support higher frame rates, and (and this is important) they have consistent video quality between cameras . I would be careful about going too 'cheap' with an action camera if the video quality doesn't match between each camera. (In the very early days, inconsistent video and color quality of cheap web cams was a real headache to track properly.)

Hope this helps.

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

Watch a one minute excerpt on Vimeo now!

Thanks a Lot Vmaslov and Greenlaw.

Kinect 2 will be dificult to buy once it's not on sales anymore. Once high resolution is not necessary, It seems PSEye is a good alternative.

So, with action cameras I can record video without realtime process and process late? Sounds good too.

I'm waiting investment to start a film next year.

Is hands capture also available in ipisoft?

Kinects are still available in many stores. Looks for this list on Amazon for example
https://www.amazon.com/gp/offer-listing/B00INAX3Q2/ref=olp_f_new?ie=UTF8&f_new=true


Actually, processing with any cameras is not realtime with our system. We've recently added realtime capability for a single depth sensor only. But it's aimed mainly for preview purpose. For getting quality mocap data you still need to post-process that realtime data.


You can capture wrist rotations with the help of PS Move controllers. Fingers can be key-framed inside iPi Mocap Studio.

Just to add to vmaslov's comments, you can build a library of hand poses inside Mocap Studio. There s a timeline where you can insert the poses and Mocap Studio will interpolate between the poses. This works surprisingly well for most actions. But if you need very specific hand poses and animations, you're probably better off doing that in your animation program. You can, of course, combine the two methods.

As mentioned, you can use PS Move controllers for wrist rotations, which, IMO, really makes a difference in realism. A third controller can optionally be used to capture head rotations. Mocap Studio can actually track the head without a controller but for me, the data with the controller usually looks better.

Going back to feet slipping, Mocap Studio is actually pretty good about keeping feet locked to the ground. I think the feet can look more solid with PS3 Eye but it's really not bad with Kinect either. When I do see slipping (usually in a turn or low squat,) I can lock the feet down inside my mocap editing or animation program. So long as you can add IK to your rig and keyframe constraints, it's not a big deal.

Just curious but are you still using LightWave? I'm still mostly using LW 2015 but slowly transitioning to LW 2018. If you're using LightWave, you can use the native tools to edit the mocap, but I prefer using another program as an intermediary program and then importing just the motion data to my final rig in LightWave.

I used Motion Builder for a long time but nowadays, I'm experimenting with Webanimate and iClone. Another iPi user here, Snapz, can tell you more about these two, but in a nutshell, Webanimate and iClone 7 with 3DXchange are good alternatives to Motion Builder. Webanimate is based on Ikinema (same as Modo and Maya) and is pretty cheap. iClone costs a bit more but is more capable and it's still a lot cheaper in the long run than Autodesk. You'll need to get it with the 3DXchange Pipeline bundle though, to get custom mocap and mesh data in and out of iClone.

One neat thing about iClone is that it's also a realtime 3D animation and rendering program. I haven't used it this way yet, because I'm more comfortable with LightWave's workflow and prefer its render quality, but after I get my current projects finished, I think I'll try iClone's renderer for a small animation project. (Tip: iClone w/3DX seems to go on sale a few times a year, so keep an eye on their website and social media.)

Hope this helps.

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

Watch a one minute excerpt on Vimeo now!

Thanks again vmaslov!

Yes Greenlaw! I'm still a Lightwave user. I'm on LW 2015. Need to update to 2018. The render quality is totally diferent, more phororreal, but render times are bigger. The surface method and rendering changed totally, so I'll need to learn everything again.
I've looked at Blender and Maya, but Lightwave is still the best tool for my needs.

I'm preparing for doing a low ( super low ) budget feature film ( 4 people team :) ) and the plan is to reduce the necessity of animators using a lot of mocap.
So I need something stable in order I will not need retouch.

I'm looking at iclone as a good alternative for retarget and layer animate over mocap before sending data to Lightwave.

I'm also looking at UNITY as renderer for characters and, maybe, Octane through Lightwave for backgrounds.
UNITY does 4K images is realtime with a good fake SSS and nice fake GI. Unity licence is cheap per month.

I will need the max time of 4 minutes per frame in 4K resolution. Lightwave native skin shade will need too much more time. I will need a fake skin shade then.

I wait the new animation sculpt from LW 2019 ( comming ) will help us a lot to sold mesh deform problems.

6 years ago I had tried ipisoft trial, but it was not sufficient stable for something like that. I Think today it's a bit more precise than other systems. So I'm thinking using ipisoft. Just need time to get some cash and do the first tests.

Brazilian government will borrow me a little money, but some months are required until the funding process will be done.


The video samples at ipisoft site are from 2015. Is it the same technology that remains today? It would be nice to see samples from 2018 actually.

That's how I feel. For my personal use, LW continues to be very affordable and since I've been using it since the late 90's, I'm very comfortable with it. At my workplace, we use LightWave and Maya. Maya does the bulk of the studio's work, but I usually use LightWave because my 3D tasks tend to be 'smaller' and I can get such tasks done more quickly in LW. I'm also looking at Blender but mostly as a support program for things LightWave doesn't have. LW 2018 is, indeed, a different animal. The shading and rendering system has been completely re-written and many old rules and tricks are now out the window. But once you wrap your head around the new workflow, it's actually easier and the looks are more realistic. As for 'slower', I think that's mostly relative...if you do things the same way in 2018 that you did in 2015, yes, it seems less efficient. But if you optimize your workflow for 2018, it's really not that much slower and you get much higher quality for less effort. Once you upgrade, be sure to get Rebel Hill's training course. The course is making my transition from the old system to new a lot less bumpy than it would have been.


Awesome! Best of luck with your film project.

Regarding the need to retouch, I think you will always run into some of that even with high-end systems. With iPi Mocap Studio, I don't think it's that bad, considering the cost of the system, and most corrections can be done directly in the program quickly. For me, it's usually just a matter of tracking backwards through the error, or keying a re-pose and tracking from the pose. (Actually, the process is not that different from fixing errors that can occur with 3D camera tracking.)

Tip: record your takes in smaller segments. This makes it much faster and easier to run through the tracking process and make corrections. If you record complex takes that run several minutes long, you're probably asking for a lot of grief.

Naturally, you want to minimize errors as much as you can. I think the PS3 Eye camera is less prone to error under the right environmental conditions. As mentioned earlier, I'm using Kinect 2 mainly because I don't have the ideal environment for PS3 Eye, and there are a few benefits of convenience. But if you want the best quality and you have the space, and you dont mind fussing a bit more with preparation, go with PS3 Eye.

What also helps is understanding how the system works and adjusting your performance for the system. In other words, avoid doing stuff that you know is difficult for the tracker. Sometimes this simply means making a slight adjustment to a stance or movement, so you can still get a good performance. iPi Mocap Studio is actually quite capable of recording complex moves, and sometimes it's actually 'easy' stuff it that trips it up. After some practice, you'll get a feel for what works best without compromising what you need.

I'll write more later...need to get ready for work now.

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

Watch a one minute excerpt on Vimeo now!

...

Some quickly tracked samples of motions recorded and processed in 2018 with iPi version 4 using a 6-PS Eye set up and appropriate other needed requirements for recording.

There has been updates since these were tracked to correct some aspects and it is even better tracking now.

These are single pass tracking takes using only the integral auto clean up tools, no manual editing was done, videos exported right out of the iPi Studio view port using the Export Video feature tool inside iPi Studio.

https://www.youtube.com/watch?v=ES64gElBlN8

https://www.youtube.com/watch?v=PtgBZ8P4O4c

https://www.youtube.com/watch?v=YaJG5stOeKA


I have a 22 ft (7 m) x 12 ft (4 m) recording area and can use a 15+ ft x 11 ft recording volume.

Excellent results compared to other solutions available in my opinion, clean up is minimal if recorded under specific requirements needed for better tracking.

Exports work in any major 3D editing program, you just have to know the specifics on how some programs import files.

...

Let me know how that goes for you. It's been hard for me to work consistently on my personal projects this year, so I'm still figuring out a lot of my 'new' tools. (I've actually had iClone for a few years...that's how far behind I am. :P)

I've had a few successful tests getting stuff from iPi to iClone and then to LightWave, but I'm still figuring out how to streamline my workflow with the newer tools. I had something in place a while back with my old tools but, silly me, I keep expanding the parameters of my projects. I really need to just get stuff done so I can move on to new projects.


To me, 4k seems like a daunting challenge. At home, I was primarily compositing at 720p for my personal short films and other projects, and when I started doing 1080p for freelance work a few years ago, that was killing my system. But, TBH, that workstation was pretty old and it only had about 9GB of RAM at the time, and it finally did die last spring. I recently built a new workstation and I think should handle 1080p easily but I don't know about 4k. The new computer might be okay for simpler 4k rendering and compositing but probably not enough power for heavy compositing and many scenes like in a film production.

So, if you can pull it off, my hat is off to you. (If I had a hat.) :)


Oh, yes, Metamorphic is pretty neat...I managed to get in on that before it was acquired by LW3DG. I've only had a chance to play with it a little but it has great potential, and it will be interesting to see how Metamorphic gets integrated in a future LightWave. It's obviously much easier and far more capable than using one of the FX tools in Layout to edit the mesh or a morph.

In the meantime, you may also want to look into the 3rd Powers suite of Lightwave plugins. The entire suite is fantastic and, for character rigging purposes, I think their Paint Weights tool for Layout is an essential plugin. Paint Weights has pretty much eliminated my need to use Maya to prep characters for mocap. (BTW, Metamorphic can paint weights in Layout too but I don't have experience with that feature yet. What I do know is that 3rd Powers Paint Weights works great.)


With the recent release of version 4, a lot of improvements have been made for better tracking and final mocap quality. This version adds support for newer capture devices, like Go-Pro type action cameras, depth sensors like Intel's RealSense and Orbec's Astra, and Logitech's recent web cams.

I haven't posted video examples in a long time but I'll post some recent tests in the upcoming holiday break. In the coming year, I'm hoping to focus on creating tutorials on a variety of production topics including motion capture. Will let you know more as I get closer to releasing stuff.

Again, good luck with your project and keep me posted on your progress! :)

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

Watch a one minute excerpt on Vimeo now!