When extracting ‘active’ from the Face Tracker node it doesn’t change its value when NVIDIA face tracking is selected, only when CPU is selected. Is this me just not having a clue what I’m doing?
Any help is very much appreciated.
Hi James
You need an Nvidia RTX enabled graphics card for the Nvidia face tracker to work…This might be why you can only get it to work with the CPU?
If you need further assistance please send in a support ticket to Notch - SUPPORT - notch.one
Regards
Will
Hi Will, thanks for your reply.
It all works fine, and I have an RTX card, my query was around extracting the ‘active’ parameter only works when CPU tracking is enabled. When RTX is enabled the active parameter always exports 1 regardless of face presence.
Good find - I’ve passed this onto the dev team to look into further. In the mean time, I’d suggest you try extracting a couple of the other positions, and check to see if they are stationary. If you are using live data, the chances of 2/3 properties staying at the exact same value for any period of time is exceedingly low, so you can use this to check if a face has been found or not. if its recorded data, the same probably applies but just be careful with changing the video playback speed.
– Ryan
Thanks for getting back to me, and sorry for my delay in reply. That sounds like a good workaround, would you do that with condition modifiers or in a javascript file? Thanks in advance.
Personally I’d use modifiers, it’s much quicker than going into Javascript. I’ve attached a quick tip of how to do that.
Ah thank you! I’ll leave you alone now 
Ok I lied, I do need to bug you again.
We have the modifiers working fine, the trouble is there is no way to force the face tracking node to ‘let go’ of the face regardless of our own boundaries set by the modifiers. So what is happening is, when a user walks away and stand a fair distance away from the installation no one else can then be captured as its still holding on to the original user.
Is there a way to force the face tracking node to reset or a call to lose current face? We are in a real bind with this project so any help is massively appreciated.
Thanks,
James
Hey there,
This is just how the NVIDIA face tracker is designed, there’s not much we can do about how it decides which face to track. you could try quickly setting active on and off to see if restarting the node forces it to pick a new face, but otherwise its sort of a black box to us.
Thanks,
Ryan
Hi Ryan,
Yeah I thought that might be the case. Its a shame the reset on face lost isn’t a controllable property but I totally understand thats under NVIDIA’s hood. Was thinking maybe if I used a camera with a shallow depth of field that will help physically lose the face in the feed, or transform the feed in some way. Worth a shot.
Thanks for your time.
James