itch.io

Is it really downloading files or just checking them?

Can you show the output from the terminal?

Do you have any error messages?

Yeah, I saw this bug and haven't figured out how to fix it yet. For now you have to restart an app when you change an avatar.

I have updated an app, so now it works fine with this model.

I have updated an app so it doesn't crash on this model, but it still have some problems with physics bones, which I will try to fix later.

Is it a free model? Is it available for download? So I can test too.

You have too many spring chains. I have limited a number of spring chains to prevent crashes, so extra spring chains will be ignored. You can try the newer version.

My bad, this bug has appeared during the previous version building. I have fixed it in the current version, so you can download the newer one. Thanks for telling me.

I have updated the app. I hope it fixes those bugs.

Ok, I will write the documentation since this software is already become pretty stable and I use it for my streams.

It's not open-source, but it's a freeware. You can use it for any purpose including commercial stuff, like streaming. No need to pay any fees. I need to mention it somewhere too.

Looks like it crashes. You can try to run it in terminal to see the errors.

Sorry, this is not an open source project.

I know that feel. I have tried VMagicMirror and VDRAW. I was disappointed that they haven't used PS4 gyro/accelerometer and VSeeFace don't support gamepad at all, but have a better tracking. So I made my own app to include every feature that I need. Unfortunately I'm using Linux and I don't have much free time to work on a separate Windows version.

I have updated the app, should be fixed in current version.

This is just an IR image. You will need 2 of them (left+right) for the stereo. And then you need to calculate and build the depth image. This part is done in camera's own SDK (for example RealSense, Leap Motion).

Found another possible alternative: MYNT EYE. Also some people are modded PS4 camera to make it work with OpenCV as depth camera.

Intel RealSense definitely has alternatives. Depth cameras are heavily used in robotics, so not only vtubers are needed them xD
Possible alternatives: Orbbec, Asus Xtion, Kinect (not sure if it's still alive).

 As an example, i’ve got a generic usb camera that’s branded as “windows hello compatible”.

I believe those are RGB-only with some extra sensors. I checked specs for the few of them. I need at least 1 RGB + 2 IR sensors in one single camera.

I see. Thanks for the info. Now I know where those models (with missing inverse bind matrices) are coming from.

You can try to adjust hands X offset in gamepad plugin settings. Does the gamepad rotation and movement works?

Added ID. You can test the gamepad.

Fixed rotations.

Yeah. It was another one. I have opened an issue for this problem to keep tracking it https://github.com/Moguri/panda3d-gltf/issues/111

Missing rotations is a bug. I'll fix it in the next update.

Yeah, that's all. Camera based hand tracking is incomplete at this moment. Full body tracking is planned.

Run an app from a console. You should see gamepad name with it's ID in a console when you connect it.

Do you have gamepad ID (it looks like 054c:#### ) ? So I could add it in the next update to check if it working.

I think I fixed it in the new version.

Looks like some troubles with importing model's armature or mesh. Does it have both armature and mesh? How did you exported the model?

I forgot a new dependency for a new capture plugin. Fixed.

Thank for the report. I have fixed this bug.

It looks like the app can't find the file "python39.zip". Make sure that you have it. You can also try to extract "encodings" folder from the file "python39.zip".

Can you provide the output of "env" ?

You should run "vtuber_webcam.sh".

Looks like "inverseBindMatrices" property become optional in newer version of file format. I need to update the importer.

Спасибо, поправлю)

Wow. This is great!