*** Paul's Playground ***

[Paul Forwood]

As I seem to jump from project to project, and test to test, some things seem to progress very slowly. I have decided to use just one thread to post most of my WIPs and tests that I find interesting or require some criticism.

 

So here are a couple of recent tests:

 

First a clip produced while testing AO in A:M 14:

GroupTest_A00d.mov

mp4 version:

 

 

 

This next one was produced from an action being used to tweak the weighting on the caveman:

BusyTest__FunA03b__A00d.mov

mp4 version:  

 

 

Edited September 5 by robcat2075
mp4 added

[steve392]

Both dam good ,I would like some more explanation on the weighting on the second one ,seems interesting ,I like the lighting

[Paul Forwood]

What would you like to know, Steve?

The caveman model is a very low-patch model, (1744 patches to be precise), which is great in most ways but makes it difficult to control body volume in extreme poses. Every control point has to be weighted to come under the influence of various bones. It takes time to get the balance right but is a must for organic models.

 

After rigging a model I create a series of actions, which put the model through all the extremes that it is ever likely to be asked to perform, and then I tweak the CP weighting from those actions. The action in the movie above was just really for testing the shoulders and for a bit of fun I just threw a few of them into a choreography.

 

The weighting is far from balanced yet.

 

The model was built in A:M 14 and is now being modified in A:M 15.

[Paul Forwood]

I added a crowd to the test, just to test if crowds are still working since the last time that I tried them:

 

BusyTest__FunA03b__A01c.mov Edit: (Modified movie... again!)

mp4 version:

 

 

[Rodney]

As I seem to jump from project to project, and test to test, some things seem to progress very slowly.

 

Slow and steady wins the race*.

 

Its always a treat to see what you are working on Paul.

Its great you are collecting everything here in one place.

 

You have such a wide variety of skills its inspirational to see them all come together like this.

 

 

Edit:

*No doubt this a quote subconsciously influenced by your Tortoise and the Hare project.

 

I just saw your addition of the crowd. Impressive test!

I'm sure you've seen Robert Holmen's tests on random action and are aware of Steffen's plugin... but if not that might work some nice variety into the cavemen's movements.

 

Very nice!

[Rodney]

Did I read that right...

 

You used Crowd Flocking on that?

If you did... that's one of the most effective crowd flocks I've seen to date.

[Paul Forwood]

Yes, Rodney. The foreground circle is all separate instances of the model. The background figures, (all facing the same direction), is simply a Crowd Flock. There does appear to be a problem with the action cycling on the Crowd.

 

Oh, there is also a refresh issue with flocking where you need to do a quick render in order to get the flock/crowd to show up in the choreography. Once you have done the render it remains visible in the realtime display for a while.

 

(I have updated the movie in my last post, above.)

[Dagooos]

I really like that caveman model. Its the first time I have seen him.

Low patch models are the best.

Have any plans for him or did you just want to model a caveman?

Any sketches of his design?

[Paul Forwood]

Thanks, David. This is a project that I started many months ago but didn't get too far with. I've just been fixing some of the rigging and weighting issues. When/If I get him sorted out I will then make a small tribe and dig out my storyboards.

 

You may remember this earlier experiment that I tried with another version of this model: Hunter Gatherer.

CycleOnPath_A07.mov

mp4 version:

 

 

This was just a single leap action, cycled along a path. No second pass has been made on it but that's what it needs.

[Rodney]

Yes, Rodney. The foreground circle is all separate instances of the model. The background figures, (all facing the same direction), is simply a Crowd Flock. There does appear to be a problem with the action cycling on the Crowd.

 

Regardless, it still turned out quite nice.

 

 

Oh, there is also a refresh issue with flocking where you need to do a quick render in order to get the flock/crowd to show up in the choreography. Once you have done the render it remains visible in the realtime display for a while.

 

Thanks Paul.

It'll probably be awhile before I do any flocking but I'll keep that in mind.

That workaround might come in handy for some during TaoA:M too.

[johnl3d]

Nice work now if I could concentate as well as you who knows what I might be able to do...

[steve392]

Thank's Paul tha\t answerd my question .I allways have problems with waighting and that sounds a good idea.Yours looks very good to me

[Paul Forwood]

I played with Flock Crowds a little more and discovered that they do not cast shadows so they are probably not very useful for real crowd situations. Better to use individual models anyway so that each one can be placed precisely where it is needed.

 

Other things that I learned are that flocks have a probem with fog on frame 0, you can put a flock crowd on a path and give it an action but you can't use it's Stride Length, you cannot randomise the direction that the objects in a crowd face.

Has anyone else made good use of flocking yet?

 

Total chaos: CrowdTest__FunA03b__A00d.mov

mp4 version:

 

 

 

Does anyone know how to set a QT movie to loop without QT Pro?

[mtpeak2]

Open the QT, go to view and you can set it to loop. When you close the QT it will ask you to save it.

 

 

[EDIT] Hmm, I had an older version of QT (7.4) I just updated and it no longer gives that option.

[Rodney]

You can use a little HTML markup to run the loop.

You should be able to do that here in the forum too but I'd have to test.

 

The Loop attribute:

LOOP=value (optional...requires QuickTime 3 or later)

 

LOOP is either TRUE, FALSE or PALINDROME. By default, LOOP is set to FALSE. When set, the LOOP attribute makes the movie play continuously. Setting LOOP to PALINDROME causes the movie to play alternately forwards and backwards.

 

So the code in the HTML would be:

 

 

Didn't know about that Palindrome setting... I'll have to experiment with that! Testing... Testing...

[Paul Forwood]

if I could concentate as well as you who knows what I might be able to do...

Hundreds of glutenous special effects dancing together in space, accompanied by project file. Sounds good, John! Now concentrate!

 

When you close the QT it will ask you to save it.

Thanks, Mark, but that only works in QT Pro now. Apple dropped that feature from the free viewer around the time that QT7 appeared.

 

You should be able to do that here in the forum too but I'd have to test.

That would be cool, Rodney, but maybe I should just get QT Pro, or post AVI movies instead.

[Kelley]

Everyone seems to be ignoring that delightful cluster of buildings in the first post. I'd love to see something happen with that.

[Paul Forwood]

Thank you, Kelley.

Most of those buildings were thrown together very quickly, just to test out Ambient Occlusion, which I hadn't really tried for some time but want to use for a CGTalk image challenge. The actual image will use new models but these were suitable for the AO test. It's funny how, when I saw the first AO render of this cluster of buildings, I thought of your wonderful brick and stone buildings that you have shown in the past. Did you use AO for those renders? It certainly adds a lovely quality to the images, giving a more natural shading and bringing out the details. The two thatched cottages are for another project.

The clusters of leaves have not been arranged into a natural formation, they are just to get a quick idea of the effect and the amount of depth that AO creates.

--------------------------------

I was just looking through some folders, searching for files which turned out to be on my other PC, when I came upon some test renders from another CGTalk challenge that I attempted over a year ago. I never completed the challenge, which was to build and texture one of a selection of characters suitable for Halloween. They provided several designs that had been contributed for the challenge and I chose to do a model of Bobby Chiu's Kangamolebunny. http://digital-bobert.cgsociety.org/gallery/

I only got as far as the basic model, rough rig and no textures. I was just playing with ideas for animating Kangamole's drool/saliva when I ran out of time. It hasn't been touched since but here is a clip of where I was going with the tests:

KMB_reduced_Test_B04f.mov

[Paul Forwood]

Here is the head of a Sweeny Todd model that I made some months back, before adding textures:

Depp_360__A01e.mov

[Paul Forwood]

And this is what happens to Todd when he gets M.A.D.:

[steve392]

Good looking charector ,did you animate him at all .Would like to see it

edit How come I can't see these qt movies unless I dl them to my hd

[frosteternal]

I played with Flock Crowds a little more and discovered that they do not cast shadows ...

Flocked crowds do cast shadows...at least they used to!

[Paul Forwood]

How come I can't see these qt movies unless I dl them to my hd

I don't know what the problem is there, Steve.

 

did you animate him at all .Would like to see it

Nothing much, Steve, but I'll see what I can find if you like.

-----------

Try these test renders:

Hairy_A01.mov

 

BobAbout_B00.mov

 

He did also participate in this test that I did with Luuk Steitner's wonderful mocap software, ZignTrack:

http://video.google.com/videoplay?docid=-8...94721&hl=en

 

 

Flocked crowds do cast shadows...at least they used to!

Great! I couldn't get them to show up in A:M 15 but maybe there is something that I have overlooked. As you can see, in those crowd tests, shadows are rendering for all objects in the choreography except the crowd components.

[Kelley]

Most of those buildings were thrown together very quickly...

Thanks, Paul, for the second serving on the houses. Very nice. It looks like you did not use 'hair' on the thatched roofs. Might be a good place for it. I especially liked the vines growing up the walls. [i'll put that in long-term memory]. And...funny you should mention the 'brick and stone' houses I was playing with a while back. I was just in that folder. ['getting nostalgic for that project. One of my first in A:M]. Did not use AO on them.

[largento]

Great stuff, Paul! I got a big kick out of seeing Alfred E. Neuman. :-)

[MJL]

Very nice work, Paul. Would it be possible to show us a wire frame of the Sweeny Todd head?

Myron

[Paul Forwood]

Kelley, you should do something with those models. You had a good thing going there.

 

Mark, thanks! Glad you recognised him!

 

Myron, not normally but as it's New Year's eve...

(Image removed)

 

Right. And on that note I had better get out of here and join some revellers for a few hours.

 

Happy New Year everyone!

---------------------------------------

 

Edit: As it is well past New Years now here's the new wireframe:

[steve392]

I love the one in the chair ,and that file plays ok but the other bump about one didn't ,strange

You should make a short with that model it looks good and interesting

Happy new year to you ,ohoo my head

[Kelley]

Kelley, you should do something with those models.

Indeed. They've been sitting for too long. Through a friend of a friend, I found a kid who with an interest in 3D. The deal is, I'll get him up and running on animating, and we'll see if we can get some of my older projects finished. First on the list is "The Sentinel", second is "Moving Foxhole". He's animating the tank treads now.

[Paul Forwood]

This is just another test render that I did of Alfred E. Todd:

ATodd_Comp1a.mov

As you can see there is much work still needed before this model is ready for action.

[Paul Forwood]

These are just a couple of tests of A:M 15e Fluids:

DomeEmitter_A00d.mov

 

Notice how the surface starts to boil once the particles have built up to a certain depth. I would like to see the surface creating waves that radiate out from the impact zones of incoming particles but all I get is boiling at the moment. Does anyone know if that is possible yet?

TapFlow_A00d.mov

[Paul Forwood]

Playing with A:M15 hair and SSS here:

SSS_TestA00.mov

I'm still modelling so there are no textures yet.

There are so many shaders that I have never tried, This is Muhair's default, which I have tried before, but I have never learned to tweak all of the properties to get a mix that works from all angles and all lighting conditions. Something else to play with, I guess. I'm aiming for silvery-white hair that doesn't look black in the shade and doesn't look overly specular.

The blue patches at the back of the neck and head are just patches with inverted normals that I missed. Well, that's what these tests are for.

[itsjustme]

That is a fantastic looking character, Paul! He looks like a Clint Eastwood caricature.

 

The eyebrow hair could use either more density or more thickness at the base of the hair and the specularity etc still needs more tweaking, but other than that, it kicks ass.

[Paul Forwood]

Well spotted, David! I haven't even put any detail in there yet. That just shows the skill of, Fabricio Rodrigues Garcia, the guy who designed this caricature. I saw this great work on CG Talk early last year and felt compelled to model Clint but I have only just stumbled upon the painting again. His characters look like 3D models already so I just had to have a go.

 

For more examples of Fabricio's great exaggerations check out his blog: F R Garcia

 

What was the name of the program that would allow you to distort a face like this? Goo?

[steve392]

That looks graet Paul very sleek looking hair ,hope you will do a tut or show some settings .That prog is Goo and winMorph and Wax allso

http://www.debugmode.com/

[robcat2075]

Good looking skin. It doesn't have the solid wax look .

[Paul Forwood]

Just making a start on the textures:

[steve392]

Looking good that reminds me of him out of house Peter Lorey is it

[Paul Forwood]

After playing with SSS for a while I have decided not to use it at the moment. It is difficult to make it work with bump mapping. If the bump value is low the bumps get washed out, a bit like porceline. If the bump value is high it doesn't play well with the SSS. (Of course this all depends on what you are trying to simulate.)

 

This is 3 passes without any SSS and with my bump map set at 30%: HeadC_Test_A00a.mov

 

This is 3 passes with SSS and bUmp set at 150%:

[mtpeak2]

Looks great Paul.

 

Did you try using the skin shader?

[robcat2075]

I have one suggestion on the hair... if you make the root of the hair do a quick transition from transparent to the intended hair color, it has less of the abrupt doll-hair look.

 

 

On the bumps I have no idea, but would any of the bump alternatives like normal maps or displacement possibly give a better result with SSS?

[Paul Forwood]

Did you try using the skin shader?

Thanks, Mark. No, I have never tried the skin shader. Are you suggesting using it with SSS or just generally?

 

if you make the root of the hair do a quick transition from transparent to the intended hair color, it has less of the abrupt doll-hair look.

Thanks, Robert. Do you mean setting the transparency along the length of the hair from the surface properties? I haven't tried that. Good thinking!

I haven't tried SSS with normal maps or displacement yet. Displacement test coming up...

[steve392]

The texture looks good the face lines are placed very

[Paul Forwood]

the face lines are placed very

Haphazardly?

They are only temporary, some very quick scratches to play with while I get a feel for A:M15.

[Paul Forwood]

Nope. Displacement gives similar results to bumps when used with SSS. Maybe by playing with the SSS properties and the bump/displacement value for long enough it would be possible to find a good balance that works.

 

Here is SSS with dispacement at 100%:

[steve392]

I meant to say (well) lol must have missed that

[robcat2075]

Thanks, Robert. Do you mean setting the transparency along the length of the hair from the surface properties?

 

yes. just at the start.

 

The rendered inset shows the transparency effect softening the roots.

 

[Paul Forwood]

Robert, did you do that in A:M 15?

I tried to get hair to start off transparent, fading to 0% transparent at 20% of it's length, holding at 0% until 90% of length and then fading to 50% transparent at 100% of the length, but for some reason it effects the whole length. Rendering produced images without hair though it was visible as semi-transparent in the realtime display, as you can see below. After several attempts I gave up and reset the transparency channel back to "not set" but the hair retained it's semi-transparent state, even after rebooting.

---------------

Edit:

I found that I had to remove the key frames from the timeline (for the transparency channel) before A:M would accept "not set".

[robcat2075]

Robert, did you do that in A:M 15?

I tried to get hair to start off transparent, fading to 0% transparent at 20% of it's length, holding at 0% until 90% of length and then fading to 50% transparent at 100% of the length, but for some reason it effects the whole length.

 

I did that in V13. I also tried using the transparency on the ends of the hair but it gave them all a dark tip rather than make them fade out, so instead I tapered the thickness to 0.01 (0 isn't possible)

 

 

Here's the PRJ of my test. frame 00:13 is a good frame. See what you get.

 

haritest02B.zip

[Paul Forwood]

Transparency, with keys along the length of the hair, doesn't seem to work in A:M15e. Whatever is keyed at 0% affects the whole length. Further keys have no affect. Transparency does offer another mean of softening the hair however but is not ideal and does seem to add considerably to the render time. The hairs already taper from 0.01 to .2 and then back to 0.01.

 

I have just opened this model in A:M 14 and it came through perfectly, even retaining the hair properties. That's cool!

-----------------------------------------

Edit: I just noticed that I lost all my constraints when I took my model back into A:M14 from A:M15. Oh well, at least the model came through alright.