Motion Capture Equipment (Positives / Negatives / Reflection regarding use)

So for about 2 or so months, I have working around and using the Perception Neuron equipment.  After my initial post regarding input of the software and how it works, my initial assessment was correct in regards to what was being inputted and how the equipment prcoesses the data.

The Suit

The suit is sometimes a pain to work with, the initial set up on your own is difficult and although it is designed to be put around the outside of clothing, the error can be seen that the sensors can easily move due to the straps moving as the are attached to a Velcro strap.  As you move, or the more you move it.  It is apparent why most of my feedback from people say that it is essentially a 1 minute record sessions with re calibrations in the middle as it  does move.

The location of the hub on the software design wise is confusing.  I am intrigued to the location that they are making you put it into as quite frankly being on your side makes no sense.  There is a good solid location on the back for a slot or compartment to hold in the hub which makes it out of the way and more effective in the overall design layout.

The Positives

Price
For the price, it's a definate bang for your buck, for indie devs with not a lot of cash, your main choices are going to be either one of a couple of things.

Perception Neuron - https://neuronmocap.com/
RoKoKo - https://www.rokoko.com/en/
XSens - https://www.xsens.com/
or Shadow - https://www.motionshadow.com/
go barebones and work with what you can do like Ninja Theory at the beginning of the Dev experience for hellblade.

So for around £2000 / £2500, you can get a suit or possibly two that'll allow you (to a point) record your and capture motion capture information.  This has it flaws which will be looked at in the negatives but overall the suit price to the results is justifiable.

Software

The software that was provided by the perception neuron is a positive and negative.  The positives are the plugins, the plugin that directly feeds into Autodesk Motionbuilder is amazing.  A simple, pull in of a neuron bot, make it live, make it real time recording and attach a mesh to the skeleton and click record and you're on your way.  Although confusing at first, it does make sense how to work it and you can directly record within Motionbuilder and do the clean up straight after with ease.

The Negatives

The Negatives... Oh boy, there are a couple.

Drifting and Constant Calibration

When I mean drifting; the explanation is based around the motion capture suit over compensation of the inertia the suit thinks that you are doing when doing a movement.  I did bring it up as an overall problem initially when observing the software but when looking more closely overall, I didn't know to the extent that it was going to be.

It really overcompensates on movements in the leg area specially if you move you legs in the Y axis in the cartesian coordinate system (it really doesn't like it) ,or when you bend down with you legs crunched up as it conflicts with each other.

As well, with the constant calibrations (needed to be done every 20 minutes due to desync), The sensors will need to be re-calibrated which is problematic which means you need to practice the capture before hand, do calibrations and then do a capture.  This is time eating and the other issue is that you are limited to your recordings which are 1 - 2 minutes meaning you are restricted to what you can capture, which if you have a big list of stuff to do... is 60% of your time wasted.

Software

The software provided is partially open and accessible which for someone who has paid vasts amount of money on the suit is annoying.  The pro sections can be helpful especially the calibration of each individual sensor but as you can't access the pro software, it's essentially half a good program that will only provided.

What's the point of half a program with the rest locked behind a Pro doorway that you can't seem to access or pay to unlock.

Suit

The sensors can be difficult to work with as you have to be very delicate with them.  This being said the sensors can disconnect for mostly one of two reasons, the cable connections do not work properly or the nodes themselves pop out of the sensor and do not work properly.

The sensors as well can pick up static or need personal calibrations separately.  The problem with this is, you can't de-static them normally.  This could ruin the sensor which are in short supply. The other way to manually calibrate each sensor manually which is horrible but the software is something I am restricted on and it's difficult to work with but grin and bear with it.

Price / quality of recording?

The diminishing returns from the quality of work in the raw data files against the time to set up / calibrate doesn't work fully with what the product should be.  Although cheaper than other kits, the feedback from industry is these kits are okay but for long term, they aren't the best inertial capture kits that are out on market.  This can be seen in the data capture with most recordings lasting 30 / 60 seconds with clear errors usually within the first 30 seconds. With the cost to have the equipment, and a bit more money you can get a good set of optical capture cameras which industry experienced animators and engineers state are better and give a better result.

Reflection

Overall, I understand the statements that have been said regarding bang for your buck but looking more closely and trying many times to get recordings and captures for data for implementation, I am not impressed with the suit.

The software provided has parts that are unnecessarily blocked, the face that cables between the parts can pop out and that the legs and arms can move so easily due to only being on a velcro strap makes that bang for your buck seem more of a calibration game as they constantly need to be re-calibrated.  Not worth it in the long run.

If given the chance, and with input from industry.  The two kits that are promoted constantly are either Vicon or Xsens.