Deviation Actions

AddestorionVayanis's avatar

VMDSpectrum Tutorial

Published:
27K Views
So i discovered that VMDSpectrum isnt really popular, so i'll try to re-introduce it to the english MMD community, since i already did it for MMDAgent addestorionvayanis.deviantart.… and ARToolkit_MMD addestorionvayanis.deviantart.…

Introduction
-------------------------------------------------------
VMDSpectrum is a software which creates 'visualization' facial motion data based on .wav files. This can be used with pmd models made specifically to display the vmd motion data as some sort of media player visualization for use in MMD

---------------------------------------------------------
Download
---------------------------------------------------------
The download via niconicodouga is here (comes with a 15-band VMDSprectum .pmd model): www.nicovideo.jp/watch/sm59036…

If that doesnt work, try searching bowlroll:  bowlroll.net/file/keyword/vmds… (link type updated on 15th january to correspond to the new way for bowlroll's search links)

---------------------------------------------------------
Start
----------------------------------------------------------
Models which do support these usually have facial bones which are usually named band1, band2, band3 and so on.

By my TRIAL AND ERROR, this is what i know of it:
Top left corner is the number of bands you want to create. Some models have less bands, so pick accordingly

Top right corner is the frame skip.
no frameskip means the motion updates every frame, giving the fastest response, but also reaches maximum motion commands sooner
Longer frameskip samples the motion at longer intervals, so you get a somewhat smoother transition of motion. I prefer skipping 2 frames.

Bottom left is the name of the band. Its generally (NAME) followed by (NUMBER) in case the model uses a different name, say... bar1,bar2,bar3....
you can just rename it at this menu so that it becomes bar1, bar2, bar3. Its default is 'band'


The warning box means you reached the maximum number of motions per model with that motion, and it will split it to multiple parts. Usually, picking large number of bands with very low frameskip will cause it to break up into more parts.

So, you'll need to use unique models for each of the motions.

Load the motion at the indicated starting frame.  For the second motion, you insert it at the starting point (frame zero), but it wont begin until the 3000th frame (as shown above).  Until the 3000th frame, you should hide that model, and show the model only at the 3000th frame, and hide the previous one so that it seems like a perfect transition.


Let the first model be on display until its motion runs out, then hide it, then display the next model so that you get the impression that it's a single seamless model.

Let it run, and be happy. lol

note:
the frequency band at the far end usually doesnt have much going on for it. If you still would like to make that end move, just select more bars than exists in the model (like... divide to 6 bars when the model only has about 4) so that all frequency bands get some attention.

----------------------------------------------------
To ModelMakers
----------------------------------------------------
It will also be possible for this to be used on things other than just bars or displays, so long as the model has an 'OTHER' facial expression which is named [something]1, [something]2, [something]3 and so on.
Which means... that if you somehow can program the visualizers on miku's skirt and arms to become facial expressions, it will be possible for the bars on miku's skirt to respond accordingly.
Or the visualizer on SeeU's costume
Or controlling parts of the model using facial expressions.
Or controlling strobe lights
Or controlling laser orientation and/or intensity and/or color

Possibilities are quite endless really.

----------------------------------------------------------
Usage
----------------------------------------------------------
the very first one i saw [ARToolkit_MMD Interstellar Flight] www.youtube.com/watch?v=W-gBRb… / www.nicovideo.jp/watch/sm10027…
the second one i saw www.nicovideo.jp/watch/sm59036…

videos made by me
ARToolkit_MMD secret garden - www.youtube.com/watch?v=yLaPU1…
SNSD Gee - www.youtube.com/watch?v=vFppK1…


UPDATE - 21/3/2013
Ula LAT's Rin Append's headphones actually have visualizers, which use 'sp' instead of 'band'. So if you save it as sp from 1-6, her headphones actually will give you visualizations.



----------------------------------------------------------
VMDSpectrum useful links
----------------------------------------------------------
RuchiiP
General sound bars tfmsetna.deviantart.com/art/So…
Electric Love Bars tfmsetna.deviantart.com/art/El…

Circular expanding - www.nicovideo.jp/watch/sm21167… [23nov2013]this one doesnt use vmdspectrum. it uses mouth facial motion data]
Image details
Image size
989x1628px 321.81 KB
Comments219
Join the community to add your comment. Already a deviant? Log In
alexxrat's avatar

Ik this is old but i have a problem, i downloaded everything alright but when i open the VMDSpectrum it says some kind of warning in japanese but opens once i press OK. Well then when i try to open a wav file in it, it says that i'm missing some files?? About DirectX stuff? Does that mean i need something else too to get this to work?