thanks for the updates, i enjoy all your comprehensive breakdowns of all the available and new software and pros n cons of the various systems, i tend to "jerry rig" my characters to get them working, and do a minimum cleanup pass. im still using "perception neuron" , its the original "kick starter" rig.kind of busted up, but still working ok. Maya 2020 on? has some good clean up tools, but they aren't backward compatible, the scenes and some skinning techniques, so so you have to watch out. Cheers!!
I have always been interested in the technical aspects of these things. As a dirt poor person, I would rather try to implement it, rather than pay. But it's good so many free options are popping up. Too bad there is technical limit to what an off the shelf mocap solution can do. Video mocap have limited spatial resolution due to pixel frustum being too big and without any depth, and bigger res takes longer to process, and that's ignoring framerate resolution, noises and various blur issues. Ai can probably help but you would need hi quality reference input data, which mean synthetic data like 3d animation when you are poor, which might not have some hidden cues, like proper inertia, to infer the data. Raw Inertial sensor can be bought for kinda cheap (except availability where I am), and could provide the extra data within the margin of error of the pixel frustrum, with the pixel correcting drift within its margin of error, but that's not a compatible solution if you use synthetic data. And and we are back to square one, to build a good video mocap solution, you need the expensive equipment first, and I don't forsee these company kill their business plan by making the video solutions too efficient.
great video, really enjoyed it.
thanks for the updates, i enjoy all your comprehensive breakdowns of all the available and new software and pros n cons of the various systems, i tend to "jerry rig" my characters to get them working, and do a minimum cleanup pass. im still using "perception neuron" , its the original "kick starter" rig.kind of busted up, but still working ok. Maya 2020 on? has some good clean up tools, but they aren't backward compatible, the scenes and some skinning techniques, so so you have to watch out. Cheers!!
oh hell yes. all of my models are vrm based, so mocapi would work beautifully for me! that's some exciting tech
I have always been interested in the technical aspects of these things. As a dirt poor person, I would rather try to implement it, rather than pay. But it's good so many free options are popping up.
Too bad there is technical limit to what an off the shelf mocap solution can do. Video mocap have limited spatial resolution due to pixel frustum being too big and without any depth, and bigger res takes longer to process, and that's ignoring framerate resolution, noises and various blur issues.
Ai can probably help but you would need hi quality reference input data, which mean synthetic data like 3d animation when you are poor, which might not have some hidden cues, like proper inertia, to infer the data.
Raw Inertial sensor can be bought for kinda cheap (except availability where I am), and could provide the extra data within the margin of error of the pixel frustrum, with the pixel correcting drift within its margin of error, but that's not a compatible solution if you use synthetic data.
And and we are back to square one, to build a good video mocap solution, you need the expensive equipment first, and I don't forsee these company kill their business plan by making the video solutions too efficient.
very informative video but the avatar uncanniness is really disturbing me
Good! You deserve it.
Slime is the best and cheapest