Digital Art Zone

fyredrygyon - 02 November 2012 06:53 PM

Oh, I did have a question.  In the Dragon Rage images, there are these cute little Asian dragons in some of the pictures.  Does anyone have any idea where they came from?  They don’t look like the Eastern Dragon but I could be wrong.

Rezca is MIA atmo, I’m clueless but hopeful of enlightenment later on

wendy♥catz - 02 November 2012 02:39 PM

no, I cannot make my own Mimic files, fortunately have most figures, just some like the Lorez Arab for example I don’t have dmc’s for and cannot figure out how to do them inspite of reading all the tutes, I keep missing Something, may click one day!

OK Mimic is one package I used a lot over the years, not sure if I’ll be able to pull off a tute like the big boys but I’ll give it a go because making mimic phonemes has become routine *sigh* Don’t know what that says about me.

What i’ll do is draft the tute here, trying to follow the best traditions of the complaint thread, and when it comes together I’ll post the whole lot in Wendy’s thread. There will be a lot of pictures so I’m going to reserve a couple of posts hopefully in sequence.

Wendy`s thread is here: http://www.daz3d.com/forums/viewthread/9199/

HOW TO MAKE PHONEMES AND EXPORT A CONFIGURATION (.DMC) FILE USING MIMIC PRO

I’m using a Windows machine with Vista installed for this, don’t have a later windows version to use sorry. My definition file might differ from yours, I seem to have a couple of dozen definition files called Default, they all seem slightly different, apologies if ymv.

I’m using Mimic Pro 3.1.0.7 here.

First and obviously some kind of audio is needed to assemble phonemes around. I use the 40 or so phoneme international scheme, I’m also comfortable with other schemes like the Disney 13 and SAPI 21. One thing I have found useful is getting actors to record a standard line so whatever work has been done on making phonemes for one voice can easily be transferred a tweaked for another voice. The phrase I use is a line of dialogue which contains all the phonemes in the international scheme:

A note about the bored worker who bashed his fat chum’s head in and got the boot.

Here is Mimic loaded up and ready to go with the line of dialogue imported via the session manager and M4 ready to go with all the windows you need open to create a new phoneme set. Is important to remember that some windows only open from other windows. For example you need to define a phoneme in terms of the morph channels an object offers for lip movement, you can only open the channels window by clicking on the containing object in the OBJECT TREEE WINDOW, in this case “head”. So, reiterating, open the object tree window and double-click on the “Head” object to see the morph channels.  You can close the object tree window now if you need the screen real estate.

If you look along the timeline you will see that mimic has parsed the phrase and displays a set of phonemes for each spoken word based upon the default mimic config file. It is this default file we are looking to change.

There is a timeline and three windows to keep an eye on when you create a phoneme. The phoneme you are looking to change or tweak is on the timeline in a horizontal strip labelled PHONEME TRACK on the left of screen. The first word of the recorded phrase is “A” so is unsuprising that the first phoneme offered by Mimic is AH. All phonemes have unique labels by convention and it is good to learn them by heart if you can.

Mimic’s interface can be a little puzzling at first because if you do this for the first time all you see in the Phoneme Palette that resembles AH is +AA(Odd). But if you click on the little “+” symbol to the left of the +AA(Odd) item it will expand to show you all the “A’s” on offer.

Double-click on the AH entry in the Phoneme palette, the Definition window pops up.  The Definition window is useful in a few ways. First and obviously it is telling us that the phoneme AH is a copy of the phoneme AA, hope that is clear. Click on the “preview” button, and drag on the upper red bar to change the strength of the AH phoneme, you get a real-time preview of the phoneme in the Display window.  Change the preview of the morph if you want to by dragging on the lower red bar next to “Preview Strength”. Phew hope that’s clear. 

Because the phoneme AH is a child, or copy,  of AA you can edit the parent directly and the children will inherit the changes, one of the real strengths and timesavers Mimic offers. But if you decide to edit the AH phoneme directly you hit the “uncopy” button, which will unparent the AH phoneme from the AA one. You can then edit the AH phoneme directly following the next couple of paras.

What I’m going to do is edit the AA phoneme so that the AH phoneme inherits.

An important window in the next step is the Channels window. To recap, you invoked the channels window by double clicking on the “Head” entry in the Object Tree window. Depending upon what morphs the Head object contains you will see a large list of channels. We are interested in the ones related to speech and lip movement. Those are built in to the M4 base model, scroll down until you see entries starting with the Phoneme “IY” and below, those are phoneme labels that follow standard naming conventions.

Double Click on the AA phoneme in the Mimic phoneme palette invokes its definition, rather than its child as double clicking on the AH channel did. Delete the entry in the Definition window by highlighting it and pressing the <DEL> key on the Keyboard.  This is where you can assign a phoneme to the Definition Window from the channels list by double clicking on the morph channel you want. Double Click on the morph channel in the Channels window and the channel appears automagically in the definitions Window. You can add as many morphs to a phoneme’s definition window as you like by double clicking on them. If you make a mistake use the <DEL> key on the KB to delete the entry from the definitions list.

Here’s the advantage to working on a parent/child system. For each child you can edit the phonemes strength, that is the intensity to which it acts upon the model’s mouth. So is possible to make one phoneme and copy it to many children at differing strengths, a handy feature that gives you a great deal of flexibility animating speech. Of course every phoneme you define can be unique, just sayin there’s a lot to be said for keeping things simple and flexible with parent/child phoneme relationships .

That’s really all there is to it. You follow the timeline left to right modding the phonemes Mimic selected for you when the audio file was parsed. Each time a particular phoneme gets modified all phonemes with that label, and any child/copied ones belonging to them,  are updated on the timeline, and the phoneme definitions are automagically updated internally as well. When you are happy with the results - or are by now dead tired from modding phonemes for hours - you go Menu->File->Save Configuration File (DMC) and a text file will be written out with the suffix .DMC. That’s your new definition file. If you save the project your work will be saved along with it, but the definitions file is what you really want, that will load in to all the plugin versions of mimic, for example in Studio, that don’t offer the phoneme editing functionality. Of course the object you apply the .DMC file to must contain the named morph channels you assigned to each phoneme for the config file to work.

A few words about “L”

“L” doesn’t fit comfortably into phoneme systems. Is it a mouth shape? Or a tongue shape? There is no agreement.  I guess the wonderful thing about standards is there are so many of them. To deal with “L” you need to decide what its allophones are in context. For example the sound of “Lie” can be parsed as “IY”,  ignoring the L as a mouth shape. The linguists who study this stuff can’t agree on what to do about sounds that aren’t mouth shapes. This is worth looking at and thinking about because a MOUTH_L morph is often supplied as part of a phoneme scheme but it’s up to a user to decide how to deploy it.

If anyone is so moved this tute needs testing out, lemme know if I missed anything or if anything is unclear and I’ll try to fix it up :)

WOW, thx, will have to fire up the pooter and follow this, does it matter I only use wav files not text?

wendy♥catz - 03 November 2012 12:02 AM

WOW, thx, will have to fire up the pooter and follow this, does it matter I only use wav files not text?

The audio may not (mostly will not) parse properly until you enter the text of any dialogue you recorded. You can do that directly by typing into the Session manager if you don’t have a file containing the text. There are advantages to putting text of the dialog in a plain ol text file and saving it alongside the audio file. There are distinct advantages in keeping the audio in any given session as short as practicable.

ps1borg - 03 November 2012 12:05 AM

wendy♥catz - 03 November 2012 12:02 AM

WOW, thx, will have to fire up the pooter and follow this, does it matter I only use wav files not text?

The audio may not (mostly will not) parse properly until you enter the text of any dialogue you recorded. You can do that directly by typing into the Session manager if you don’t have a file containing the text. There are advantages to putting text of the dialog in a plain ol text file and saving it alongside the audio file. There are distinct advantages in keeping the audio in any given session as short as practicable.

I should then, actually have a text file of Phonemes too if I use Speakonia.
thanka for this, will link it to my Mimic thread asking Daz to put back the on site dmc files so it is easy for others to find with search too
(and me if I misplace the copy I make)

Partly a complaint, but kinda an observation…

Poor Genesis lacks many Phonemes :( I’ve used what is available heaps… But there are many that do not exist :(

Spyro - 03 November 2012 12:15 AM

Partly a complaint, but kinda an observation…

Poor Genesis lacks many Phonemes :( I’ve used what is available heaps… But there are many that do not exist :(

So this gives you a way of creating more from what is there, is always the more we are most interested in *wonders how to import genesis to mimic without a CR2?*

eta I guess you could make a DMC for Genesis using M4 and Genesis wouldn’t know the difference, would take some tweaking I imagine but it would be doable.

yes, creating a cr2 does not work well.
but it does at least have a dmc

wendy♥catz - 03 November 2012 12:30 AM

yes, creating a cr2 does not work well.
but it does at least have a dmc

Is this making sense so far Wendy?

So mimic effectively creates phonemes? (I have never used mimic so u know LOL) I always though it was built the same was as morphs. That you’d have to hand carve and shape the face in a full head morph to make one.

Spyro - 03 November 2012 12:36 AM

So mimic effectively creates phonemes? (I have never used mimic so u know LOL) I always though it was built the same was as morphs. That you’d have to hand carve and shape the face in a full head morph to make one.

You make the phoneme from whatever morphs are available and tweak in mimic. Mimic creates and exports an animated pose file you apply to the model in studio or poser or carrara or whatever, the pose file animates the model to simulate lip-sync’d speech. Usually what is available in an object can be combined in Mimic to produce something reasonable that is then applied to the object as an animated pose.

ps1borg - 03 November 2012 12:33 AM

wendy♥catz - 03 November 2012 12:30 AM

yes, creating a cr2 does not work well.
but it does at least have a dmc

Is this making sense so far Wendy?

trying to follow, on my Android, when my video finishes uploading I can use the Android as a wifi for my laptop and save it all
no broadband left on desktop for a week.