jeelizFaceFilter
jeelizFaceFilter copied to clipboard
Refactor the dependency
Hello !
This is not to report a bug but to suggest you an improvement of your dependency. Are you interested by a big refactor of your dependency ? (I will include ThreeJS helper, within the core module in typescript. I never do some babylon project so I can't really do it :/)
I used twice our dependency, and I always copy/paste the projects files instead of use the package directly.
Hi @dupasj
Do you mean for the NPM package used in the https://github.com/jeeliz/jeelizFaceFilter/tree/master/reactThreeFiberDemo Or the static JavaScript demos?
Hi @dupasj
Do you mean for the NPM package used in the https://github.com/jeeliz/jeelizFaceFilter/tree/master/reactThreeFiberDemo Or the static JavaScript demos?
Hi !
I mean for the NPM package.
Sure I would be interested :) But I would prefer to keep Three.js helper either outside the package, or to put it in a separate package. Indeed, even if FaceFilter is mostly used with Three.js, this lib is framework agnostic.
Yes sure ! It's what I want to do !
Little outscoped question ^^' Will you provide more AR really experience ? Like foot tracking, body tracking, surface tracking... ?!
-
Jeeliz brand will be focused on glasses virtual try-on, especially through the Jeeliz VTO widget
-
Jeeliz FaceFilter and Jeeliz Weboji will still be maintained and improved,
-
Enhanced experiences are provided by WebAR.rocks brand ( https://github.com/webAR-rocks ), especially:
- WebAR.rocks.face has a similar interface to Jeeliz FaceFilter, except it outputs the position of face landmarks and the pose is computed differently. It gives access to new use-cases: high precision virtual try-on for hats, headphones, helmets, glasses, lipstick and makeup virtual try-on, deformable masks
- WebAR.rocks.hand is focused on hand detection and tracking, I am actively working on it. It can be used for hand based navigation, bracelet/wristwatch and ring VTO, hand object manipulation
- WebAR.rocks.object and WebAR.rocks.image are for objects or images detection and tracking, but their use is not fully standalone (it requires often custom neural network training), except for some use-cases like keyboard based SLAM or cup detection for example.
- surface tracking does not suit very well to neural network-based detection, and furthermore it will come through WebXR (as soon as Apple will decide to implement it correctly), so I have no plans for that
- body tracking is also not in my focus,
- I am working on foot tracking but the results are not good enough for commercial use yet.
Thus projects looks really cool ! Could I have the unminified source of the project ? I will try to merge all thus projects in one ;)
Hi,
The source code of the NPM package is here: https://github.com/jeeliz/jeelizFaceFilterNPMPackage But I don't provide the unmodified source code of the core of these projects.
All right, I can't merge thus projects in one if I can't view the unminfied code, I will do my best. If you don't want to provide thus lines of code for reasons of fairness, we can sign confidentiality agreements if you want.
I send you an email with mores details. Cheers