Community Model- AManda

[John Bigboote]

Meet AManda!

 

Jirard had posted a model, and I saw great potential in it and built upon it... and thought it would be a great idea to use as a community building/learning project- see where it goes! I learned a lot by looking at Jirard's wonderful minimal splineage, and my model- if studied, will teach some of my tricks and 'bad habits'...

 

Now, Jirard was going to take it and add teeth and... whatever from there. For now- it is just a female head. I added minimal bones and some poses. Feel free to DL this model or project and add to it... we'll figure-out a way to combine progress as we go. Show us YOUR tricks!

[John Bigboote]

Here are the files, MDL, PRJ and all the image maps thus far. V19... Less than 1mb!

Jirards AManda.zip

[robcat2075]

Great makeup!

[RS3D]

Thanks for posting this.

[jirard]

Nice work Matt

Someone else will have to make the teeth. I have been trying but they look like horse teeth.

[John Bigboote]

I know it is 'wrong'... but I kinda like the sphere in there... it does not draw attention- like teeth can... maybe I'll cut it in half for uppers and lowers and add a toungue/mouth behind it... sometimes you need to stylize.

[jirard]

Yeah the sphere fits the look. It was too hard to line up teeth. I made stylized teeth but theylook like horse dentures.

[John Bigboote]

I have teeth from other models as well that could be copy/pasted in...

[jirard]

I'd like to see them. I have problems creating teeth and ears.

teeth.mdl

[John Bigboote]

OK--- did a little work on hair today... still a lot to tweek. I will let a test animation render overnight. I dug-up this old thread where I learnt to use an image to control hair placement, giving AManda a nice widows-peak that does not rely on patches... https://www.hash.com/forums/index.php?showtopic=34198&hl=hair+decal&do=findComment&comment=290226

[jirard]

Amazing work Matt

[John Bigboote]

Thanks-we'll see if it all flies caber-wonky in animation.... the trick here is a FORCE on the hair... I finally nudged it's magnitude all the way up to 50,000... it is constrained to her head-bone to move with it... so in the 2nd picture you can better see the 'widow's peak'...

 

ALSO- you will see in the animation that I added your teeth model to her with some gums and a tongue.

[Rodney]

You should be able to use an image to control the length as well and that might be highly advantageous.

 

 

Looking good thus far!

[jirard]

Oh no not the horse teeth 😀

[John Bigboote]

You should be able to use an image to control the length as well and that might be highly advantageous.

OH YEAH! Really good idea there, Sarge!

[John Bigboote]

SO- when making a hair decal- I know that an alpha channel is not needed and that white means hair and black means, no hair... so with length I suppose it would follow suite where... black means very short hair and white is very long hair and the shades of grey in between are the intermediaries? I will investigate this tomorrow!

[John Bigboote]

AND- as predicted, the hair went caber-wonkey upon render with the simple head move. Not too bad...(movie file attached) it is mostly a cavalcade of hair-mass falling forwards as the head tilts and overpowering the force- I am not sure the force is the way to go, I have experimented with them in the past and wound-up with models that have 4-5 forces attahed to the head and I am forever adjusting magnitudes... so-

 

I will experiment with these image-driven hair properties... length and direction. Now- how would the direction option work? I imagine like a normal map where various colors shade the normal in differing directions?

 

FIRSTLY... I will need to make a pose where I flatten my hair group of patches so that I can properly UV these images I will be making in Photoshomp.

 

NEXT- our AManda will be needing a body... anyone?

AManda4.mp4

[John Bigboote]

This is the Reader's Digest version of UV mapping in A:M... rather simple. I selected the scalp group (I had named 'hair-here'...) hid everything else and made a new pose, named it 'HAIR FLATTENER' and did a distortion, and scale operations and then played with bias handles and moved CP's as I saw fit until I got a nice flat mass to apply decals to.

[John Bigboote]

Looking good. Next is to experiment with the 'direction' feature... LOTS of tweeking has been done, but I will provide the .prj which will let you peek behind the curtains if there are any questions...

[John Bigboote]

You can tweeek endlessly when it comes to hair! Thought I would post a WIP, this is a fun project to have going on in the background. In this image, you can see the force which is keeping the hair at bay from the forehead and face.

[John Bigboote]

I am getting really good feedback from the image-based hair controls for 'density' and 'length', and I am using a different method for 'color' (the hair's color property offers the ability to change color and other factors such as ambiance, specularity, variation amount... down the length of the hair strand- which is really very cool and you can see it in the image above where the hair's roots are darker then get blonder as they lenghten...) These 2 controls work off a simple greyscale image, for length black means short and white means long and for density black means no hair and white means full hair.

 

BUT I can not get a handle on the 'direction' option. I've tried greyscale and a 'normal map' type color image, and experimented with variants of both but can not seem to determine what makes a hair go off in a predetermined direction. Anyone have any input on how this might work?

 

ALSO- the key I am finding is to test-test-test, with full animation feedback. I am using my NetRender WITHOUT baking the hair at all, and it is working fantastically.

[Rodney]

For direction you'll want to check out Colin Freeman's appication as well as the basic insights into how color directs hair orientation.

 

http://www.colins-loft.net/hairbrush.html

 

The application is flash based and so will not run in many modern broswers but there is a downloadable program on that page. The painted results in the window can then be exported for further testing.

 

 

During one of the Live Answer Sessions we explored this and Robert even demo'd basic use via Photoshop.

[Rodney]

Here's an example of what that painting produces after blurring the image:

 

 

(I applied the same image as color.... but not sure why the color and the direction aren't the same in each instance as they seem to be inverted vertically)

[John Bigboote]

Awesome- Thanks Rodney- I'm OFF!

 

EDIT- I see it is a rather old demo... uses the 'old' style of hair... hope it works with today's hair. (did you use the new hair for your test?- I'm sure you probably did...)

 

EDIT-image- WHOA! Look at all the properties the old-style hair could have image-controlled!

[Rodney]

I think that perhaps I don't know what 'new hair' is.

I was under the impression this was all working with the newest hair available.

 

Edit: I do see where his site references old hair. Wow. I had forgotten that was with old hair.

As an aside keep in mind we can still use the old hair locks by holding down the Shift key when creating the new hair.

And technically speaking we may have all those old options... which I didn't think we still had.

 

 

Matt, you sly fox.

You've revealed a long lost feature (at least lost to me!)

[robcat2075]

I suspect the "direction" maps may be of most use in animal furs that have a very definitive direction out of the skin.

[John Bigboote]

Rodney- I think Martin rewrote the code(or had one of his guys do it) around version 12 or 13. The new hair was supposed to be better ar collision-detection... which was something I was very excited about. If you look at the image of the AManda with the hair resting and her shoulders and not passing through her shoulders That is the look I have strived for.

 

Rob- I think there is a myriad of uses for direction mapping. One thing that I did learn in my tests. Is that the color map affects the base root Of the hair and that as the hair gets longer and further away from the base the dynamics take over and anything can happen. But having that initial direction can be instrumental in getting the look you are after.

[John Bigboote]

Progress on hair.

AManda_Loop1.mp4

[RS3D]

The changing highlights on the hair as the head turns really adds to the realism. Nice job.

[John Bigboote]

More toying with IBL renders... the 1st image uses the image I posted as the lighting source, the 2nd has the image used as fog is lighting the scene, both have 1 additional light near the camera with shadows and specularity on. I do not know where the green fill is coming from on the 1st image...

[Rodney]

I do not know where the green fill is coming from on the 1st image...

 

 

Perhaps you have a negative light?

[John Bigboote]

Rodney, no... just 1 light in that scene... renders quick at just under 3min per frame @1280X720.... an inverse(negative) white light would render as eminating darkness... I suppose a red light would be green in inverse-land. Negative lights are a powerful feature I stumbled on by chance!

 

Here is an animation! Hit Cntrl_L in quicktime to make it loop.

AMandaY_LOOPER.mov

[largento]

In real life, a colored light casts a shadow that is its opposite on the color wheel. We did this in one of my freshman art classes. We took a white cue ball on a white poster board and put colored gels on a light in a photo darkroom. A red light would cast green shadows. The point of the exercise was to expose us to the idea that shadows weren't simply black.

[John Bigboote]

Okay- I am done toying with A:M hair- it is what it is... here is the zipped assets if anyone is interested in seeing settings or learning about long hair in A:M... or just wants a cool model! I borrowed some VO from Scarlett JoHanssen in the movie 'Her'. There is a longer clip in there- I may try some more lip-synch with.

 

I've found that it is not necessary but for Collision-Detection (CD) it helps to bake the hair... the resulting pai and par files I have omitted from the folder because they were 60mb+... so you would need to perform a 'bake' which for this 10 second clip takes about 30 minutes. ALSO- to help the CD, I have a thicker mesh that expands out over the face and shoulders that should be set to (PUSH MESH POSE 100%) and the force that keeps the hair off the face has a pose that can be activated/deactivated as well.

 

You can use any image in the chor settings to light the scene... I will attach the image I used if needed... have fun- hope someone gets something out of this someday... I will return to add a rigged body to AManda someday, maybe.

AManda_Loop7 (1).mp4

Jirards AManda (2).zip

[Mechadelphia]

Okay- I am done toying with A:M hair- it is what it is...(snip)

 

Hi Matt. Are you disappointed with your hair results?

 

It looks fine to me. Were the solutions that you came up to address your earlier problems practical ones or do they seem to be more trouble than they are worth?

 

Thanks for the demonstration model. I am gong to have a look at it and try to learn more about A:M's hair system.

[John Bigboote]

 

Okay- I am done toying with A:M hair- it is what it is...(snip)

 

Were the solutions that you came up to address your earlier problems practical ones or do they seem to be more trouble than they are worth?

Hi Mack- well, no. I have been 'toying' with A:M hair for 16+(YIKES!) years now... trying for something big. I was a little disappointed in this test seeing the hair 'sink thru' the shoulders toward the end. It was way before its time once(the A:M hair feature) and is still timely. Even watching a Blender hair tutorial I see a guru lamenting after his failed demo that 'you just have to eventually 'live with' the hair penetration that will inevitably occur. https://youtu.be/2YCQLZMuvkE skip to 10min15secs to see the hair fail. Other programs have great looking hair, but don't really work whence animated.

 

Ultimately, I would like to see hair that reacts realisticly under animated conditions, yet will not sink thru geometry when collision-detection conditions occur.

 

My solutions were more scientific than anything... the 'mesh' approach is a reaction to the hair particle needing a dense geometry to react to (theory) so I hide a dense mesh within the character and have a pose that expands it out OVER the skin area while the hair simulation is happening... once simulated, the pose can be de-activated for render. I make the dense mesh by copying the CPs where I expect collision to occur, like the neck and shoulders(and cheekbones) and paste them into a MESH group. Then, I use the Plug-In/Wizards/Push CP's feature to expand the patches along their normal and make a new pose for the result... I then make that new pose a +100/-100 range and in the -100 I shrink the CPs to be out-of-sight gone.

 

There are other theories I chase as well. For instance- there are 2 places where the hair's density can be set in the material: One, in the 'Hair System' and the other, in the 'Hair Emitter'. One of those is for how dense the 'mimickers' or 'populators' should be, and the other for how dense the 'hair guides' should be. The hair-guides I believe are the ones that actually react to grooming and collision-detection while the 'populators' merely mimic the nearby hair guide and even if the hair guide is properly not passing thru geometry, it's populators will because they are just 'offsets' and the offset may place it thru the mesh.

 

Then- there is the hair's 'Control Points' value which can make a strand stiff like a piece of uncooked spaghetti or limp like a piece of cooked spaghetti (hey! this value should be called 'al-dente'!) I would have to consider those points along the hair's length are used by the collision-detection as well.

 

Then- there is the basic dynamic targeting modes, Constraint, Spring or none. Then, there is bounce and friction... I honestly can not say what these effect and have tried values from default...zero, 10,000 with no discernable differences.

 

And THEN- there mis the mysterious 'Cull'... seems to hold some big powers- but how, what...?

 

There is much to explore and many hours to be frittered, warning!

 

Thanks for your interest - let me know what you learn.

[John Bigboote]

I should note to anyone who DL's the folder that my latest .prj was AManda_MC_PRJ3... sorry- somewhere along the way I backed-up a version...

[John Bigboote]

More experimenting, shaded low-res render... no baked hair, so the hair calculates at render time(fast)- with the force pose ON and the mesh ON too- but shaded transparent so the hair recognizes it but the renderer does not.

 

EDIT- updated rendered gif. I guess I still recommend baking the hair- it really helps for CD.

[John Bigboote]

SO THEN I'm getting a last render ready to go over the weekend when BAM! All my(Jirard's) 5 point patch's go away! Crap! Which brings the added joy of having to re-apply all decals that were on those patches in any way...grrr!

 

And some of them just don't want to close no matter what you do... and I see Jirard has a habit of 'hooking' to 5-pointers... which further complicates things.

[jirard]

Ouch

If the 5 point patch won't close, there usually is a hook next to it.

I detach the hook and close the 5 point patch.

That sucks...

For stubborn 5 point patches I select the points, press H, select all and that usually works or I select them all in a different order.

[Mechadelphia]

Thanks Matt for your comprehensive post!

 

I tried to understand the hair system long ago but it just seemed so complicated to me that I gave up. For example, understanding what all of those settings that you mentioned actually do is daunting to me.

 

Every time I see your results though I reconsider trying to learn about it again.

 

I'll let you know what I learn for sure. It may not be much at first but I'll share what learn anyway.

[John Bigboote]

Jirard- yep, that is my method too... fixing them 2 by two.

Mack- yeah- LOTS of settings, many seem redundant... and I haven't even TOUCHED the grooming tools on this... which are pretty nice in A:M. (I have a force connected to the model found in the bones folder.) that is always 'pushing' the hair back, then gravity does the rest.

[John Bigboote]

I noticed... she had lost her 'widows peak' somewhere along the lines... turns-out- setting the 'culling' setting anything less than 50%(default) makes you decal-driven 'density' go away and hair will populate on the entire hair group.

[John Bigboote]

Reduced lip-size a little, added freckles and a wisp of hair in front... this is a combination of a full IBL lit shot and a standard default lighting setup halfmixed in After Effects... lens-flare by VideoCoPilot. I am finally mostly happy with the hair's collision-detection and interaction with the shoulders. I made a pose that I only turn-ON for the hair-bake-simulation, which brings the dense mesh out into position and tucks away things that could get in the way like the ears and earrings- then I turn the pose OFF for rendering.

AManda_talking (2).mp4