*** Paul's Playground ***

[Paul Forwood]

Nope. Not Jesus.

C_I_C04.mov

I haven't played with dynamic hair for a while. I'm not sure if there is a damping option for hair.

[TheSpleen]

nice!!!!! Good styling!

[largento]

Nice model as always, Paul! Not sure about the hair, but I haven't really messed with hair much.

 

I hate to hear about your hard drive. I hope you didn't lose too much stuff.

[robcat2075]

Increasing the "stiffness" may be what you're looking for.

[mtpeak2]

You have to set the targeting to spring to get the damping option.

[robcat2075]

You have to set the targeting to spring to get the damping option.

 

a-ha!

 

What is the intended difference between "constraint" and "spring" for hair?

[Paul Forwood]

Thanks for the comments, guys!

 

I think that the spring constraint allows the hair to stretch and spring back to it's groomed state. Not really what I was looking for but I will give it a try.

 

Here are a few frames of my last attempt:

C_I_C07.mov

[HomeSlice]

You have to set the targeting to spring to get the damping option.

 

a-ha!

 

What is the intended difference between "constraint" and "spring" for hair?

 

From A:M Properties - a quick reference

http://www.hash.com/forums/index.php?s=&am...st&p=327831

 

Constraint : Constraints will be used to correct the bone towards the underlying modeled or animated position. The value represents is the amount it will move towards the target in one frame. It can range from 0% to 100%. Where 0% is the same as no targeting, and 100% is completely stiff.

 

Spring : This method uses the a traditional damped spring using Hook's Spring Law to encourage the chain towards the modeled or animated goal position. Frequency and damping are used to control the behavior of the spring.

 

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

Still not sure what that means in real world use though... <_>

[robcat2075]

Hook's Spring Law? More big government!

 

 

I suppose a side by side test with one hair might reveal the difference in behavior.

 

My first guess is that Hook's Spring Law is more computationally intensive.

[NancyGormezano]

Hook's Spring Law? More big government!

 

 

I suppose a side by side test with one hair might reveal the difference in behavior.

 

My first guess is that Hook's Spring Law is more computationally intensive.

 

Spring hair is ...uh...springier. Bounces more wildly. I like it - but is harder to control. I usually set drag to 20%, Turn respond to gravity = OFF (so that hair returns to original styling and doesn't fall flat), but still respond to forces. Constraint type seems stiffer, less noticeable

 

It doesn't (or didn't useta) seem to take any more time to compute.

[Paul Forwood]

This one uses the spring constraint with the damping left at the default value of 5:

C_I_C09.mov

I think the object collision detection is a problem with a low patch model.

Springs are coming back to me now.

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

[Paul Forwood]

Hmmm. Where did I put that bow? It's around here somewhere.

 

Am I going crazy? One minute the hair had a "bow" option and when I next looked I couldn't find it. Where is it hiding?

[NancyGormezano]

Am I going crazy? One minute the hair had a "bow" option and when I next looked I couldn't find it. Where is it hiding?

 

There are now 2 crazy people in the room - I have seen that happen as well. I can't remember exactly what I did to bring it back...but more than likely it was to start the hair material from scratch. Blah.

 

Before doing that, maybe instead close A:M and restart?

 

I have noticed that funny stuff might happen when switching between Muhair and NOT muhair? Things go missing, or remain set, don't change. Also maybe from switching between dynamic type options (spring, constraint).

 

Did you bake the hair?

 

Was this a hair material created in a previous version? (what version are you using?)

 

Can you describe in more detail what you had been trying?

[mtpeak2]

If you are using the hair shader, it disappears. The hair shader takes care of it.

 

What version are you using?

[Paul Forwood]

Thanks for the help.

 

The examples that I have shown were entirely created in A:M14c. I did run "calculate spring systems" but no baking or hair shaders. This is just an image on hair.

I have just upgraded so am trying it in A:M15 and A:M16 (Beta5).

[Paul Forwood]

Back in A:M14c for the moment and still not quite there with the hair.

ChI_A02b.mov

[Paul Forwood]

A little more progress:

ChI_A06.mov

[Paul Forwood]

Another character in progress:

AM16_Barin_C01i.mov

 

I've just started a test render with SlowAO and one klieg light. This is the first frame:

A few frames with AO:

AM16_Barin_C01i_AO.mov

They were taking over 6 minutes per frame with 3 passes which is too long for an impatient guy like me so I will see what FakeAO can do. I haven't tried it yet.

[Paul Forwood]

I think that I had better read the instructions for FakeAO.

[robcat2075]

I think that I had better read the instructions for FakeAO.

 

Make sure the first prong of the FakeAO under the camera is set to "Depth" not "color"

[Paul Forwood]

Thanks, Robert. First buffer is set to depth and second is set to color but I'm just getting similar results to that last image.

 

I really like the effect that real AO produces but would need to find a way of preserving the highlights and reflections on shiney surfaces, preferably without compositing. I'll try messing about with FastAO again tomorrow.

[NancyGormezano]

I really like the effect that real AO produces but would need to find a way of preserving the highlights and reflections on shiney surfaces, preferably without compositing. I'll try messing about with FastAO again tomorrow.

 

I found that setting AI=100%, AO=0% and having 1 Klieg (% intensity tweaked to taste) with diffuse=ON, spec=ON, NO shadows works well with FastFakeAO.

 

You will have to tweak settings for FastAO according to your models, chor camera focal length

 

here is with no reflections (not sure what Fakeao camera focal length was - probably 70)

 

here is with reflections, but FakeAo focal length changed to 250 (probably too much)

[robcat2075]

I had a case today where FakeAO was making that incorrect result.

 

I deleted it from the camera and the post effects, saved, reloaded, then made it over again and it worked right.

 

You do have Depth Buffer on in the render settings

[Paul Forwood]

Thanks Nancy and Robert. I have tried your suggestions and dozens of other configurations but I just get the same results. I am thinking that my models might be a little too low on patches to get the required results out of FakeAO.

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

Edit: I found the problem! I was trying to render to QT rather than an image sequence. I will test a bit more now and post my results.

[Paul Forwood]

FakeAO_Test_A00_qt.mov

 

Much better!

One klieg light at 50%

AI = 50%

AO = 0%

FakeAO = all default values except Focal Length, which I set to match the camera's focal length.

Rendered 3 passes to png format.

 

I'm still not sure that I will be using FakeAO for animation, even though the average render time for this sequence was only about 4 seconds per frame. I look forward to seeing what others produce with the full version and in the meantime I will keep experimenting.

[steve392]

FakeAO_Test_A00_qt.mov

 

Much better!

One klieg light at 50%

AI = 50%

AO = 0%

FakeAO = all default values except Focal Length, which I set to match the camera's focal length.

Rendered 3 passes to png format.

 

I'm still not sure that I will be using FakeAO for animation, even though the average render time for this sequence was only about 4 seconds per frame. I look forward to seeing what others produce with the full version and in the meantime I will keep experimenting.

That looks good Paul ,nice colouring ,nice charector aswell

[NancyGormezano]

Edit: I found the problem! I was trying to render to QT rather than an image sequence. I will test a bit more now and post my results.

 

Weird. Because I was able to render to QT with beta 5 (PC 32bit). Surprised me that I could because it was reported that you should only render to tga, png, exr to use depth buffer. Perhaps it was a fluke?

 

Character looks good. Doing Shakespere, are we?

[Paul Forwood]

Weird. Because I was able to render to QT with beta 5 (PC 32bit).

That is weird! What compression did you use in your QT movie, Nancy? I was using mpeg because I had been experiencing problems with H263 recently, thoughthat was probably not with A:M16. I have just started a render with h263 compression and it is actually rendering correctly, as far as I can tell, to the QT movie format. I note that it is not possible to access the depth buffer settings in QT mode as you can when rendering to a still image format. Maybe it is neccessary to set a render up for stills first and then change to QT? Anyway it is working now so I will press on.

 

Character looks good. Doing Shakespere, are we?

 

Thanks, Nancy. Not Shakespeare, though I can see that these recent models would fit the period quite well...at the moment.

I did start a model of Shakespeare a few years ago but, alas, (poor Yorik), he remains nothing more than a head. This will be more 30's sci-fi. Think "Saviour of the universe"! (No, Still not Jesus).

[NancyGormezano]

That is weird! What compression did you use in your QT movie, Nancy? I was using mpeg because I had been experiencing problems with H263 recently, thoughthat was probably not with A:M16. I have just started a render with h263 compression and it is actually rendering correctly, as far as I can tell, to the QT movie format. I note that it is not possible to access the depth buffer settings in QT mode as you can when rendering to a still image format. Maybe it is neccessary to set a render up for stills first and then change to QT?

 

I only use "animation" compression with rendering to QT with A:M. I believe that is NO compression. I will then recompress using QT pro for posting to forum in order to play with quality versus file size issues. I use h264 usually with QT pro.

 

And yeah, my guess is that one has to render a still first to access the depth buffer, and then from then on, can render to QT. Have not verified.

 

I have found that it is not a good idea to do a Render (Q) to the screen when using FakeAO, nor to experiment with turning the post effect on/off/on in the render settings. It is then that the weird fakeao results show up, and will not go away, even when doing a "render to file". I find I have to delete the post effect, and delete any orphan channels created, and re-add it, in order to have it correct once again.

[TheSpleen]

That is a very very cool character! I can tell he would have tons of expression!

[Paul Forwood]

Oops! 264 not 263! That's what I meant.

I have found that it is not a good idea to do a Render (Q) to the screen when using FakeAO, nor to experiment with turning the post effect on/off/on in the render settings. It is then that the weird fakeao results show up, and will not go away, even when doing a "render to file". I find I have to delete the post effect, and delete any orphan channels created, and re-add it, in order to have it correct once again.

I ended up starting a new project before I got FakeAO working but I suspect that had I used the correct compression settings in the first place it might have worked as it was. Don't know though.

 

That is a very very cool character!

Thanks, Gene! I wanted to add a touch of Disney to him so he has slightly exaggerated proportions.

[Paul Forwood]

I wanted to try Subsurface scattering with FakeAO. This is what I got:

FakeAO_Test_A00_Render2QT_sss_01.mov

I only had one klieg light, set to 50%, in this sequence. Turning on the other two lights brings the skin tones back to where they started but washes out the AO. It's a balancing trick.

Rendering examples at the moment.

[Paul Forwood]

FakeAO with SSS and 3 point lighting:

FakeAO_Test_A00_Render2QT_sssAndAllLights_01.mov

Average render time: about 4 seconds (I forgot to check after the render finished but I noted that it was around 4 seconds at about 75% rendered).

Not bad. A few little tweaks to the model would sort out the strangeness, like his left lower eyelid.

Better than expected!

Well done, Jenpy!

[NancyGormezano]

The sss looks terrific with the fakeao!

 

Care to post your settings? I never had much luck with sss. Is it related to the size of model as well as lighting? Can you have decals with sss? I seem to remember the sss group(s) must have certain surface characteristics, as well as be the last ? first? group. Not belong to other groups? It was a long time ago that I played with it.

[Paul Forwood]

Here is a comparison of RealAO, FakeAO with SSS and No AO or SSS:

I forgot to add the average render times which were approximately:

4 seconds per frame for FakeAO with SSS

6 minutes per frame for RealAO without SSS

 

This is the RealAO example:

AM16_RealAO_Test_A00.mov

 

The two versions of AO both have qualities that I like and qualities that I dislike. The standard render still wins but I'll see what I think after the texture maps are added to the mix.

I'm impressed by the speed of FakeAO and SSS! Has Steffen been tweaking SSS recently or did I just get lucky with that render?

[robcat2075]

Is it true SSS and RealAO don't work together?

[Paul Forwood]

Care to post your settings?

Please refer to the image above, Nancy.

 

I never had much luck with sss. Is it related to the size of model as well as lighting?

Lighting definitely has an impact but I haven't experimented with size. I would think that the more screen space that the SSS occupies the longer the render time, but I don't know if a 10 cm sphere renders any faster than a 10 metre sphere. I think it is all down to screen space, pixels. You only calculate what will be seen... I think.

 

Can you have decals with sss?

Yes. Well last time I tried, (Clint Eastwood model), you could. SSS has a smoothing effect, not unlike the porcelein material, so bumpmaps tend to get a bit washed out. As I recall, displacement with SSS did work but can be a bit messy. Probably just needs to be used judiciously and with careful planning. I should try it again with A:M16.

 

I seem to remember the sss group(s) must have certain surface characteristics, as well as be the last ? first? group. Not belong to other groups? It was a long time ago that I played with it.

I simply create a group called "Skin", which contains a selection of all the geometry that represents skin and then set the SSS properties in that group's surface properties. Yes, heirarchy matters.

 

Is it true SSS and RealAO don't work together?

I don't know, Robert. I was going to set that one up to run overnight.

[Paul Forwood]

I tried to render with RealAO and SSS but the skin rendered black. I wasn't able to get them to work together.

 

FakeAO and SSS work so well together so far. Changing the lighting affects the render time.

This is one frame of the same sequence that is rendering at 1 min: 25 secs per frame which is not bad, considering that my PC is slow by today's standards, but it is much longer than the previous 4 seconds per frame that I was getting. (Maybe I misread those results).

Shadows are turned OFF in this example:

FakeAO_Test_A00__SSS3pntLights_09.mov

I think I am going to love this new rendering option.

[Rodney]

Very nice Paul!

 

Thanks to Nancy and you I see SSS is making a triumphant return.

[Paul Forwood]

Thanks, Rodney,

 

FakeAO, SSS, 3 point lighting, shadows ON:

With raytraced shadows: FakeAO_Test_A00__SSS3pntLights_Shadows.mov

With Z-buffered shadows: FakeAO_SSS_Test_A00_Shadows02.mov

[mtpeak2]

Those look great, Paul.

[itsjustme]

Impressive, Paul!

[Paul Forwood]

Thank you, Mark and David.

 

Next I'll add some textures.

[Paul Forwood]

Here is an example using FakeAO, SSS, Raytraced shadows, Colour map and Bump map. The bump map runs the full scope from back to white and is applied at 100%:

FakeAO_SSS_RTShadows_Col_Bmp_100_a.mov

The maps are just quick and rough for testing purposes.

What looks like freckles were actually meant as a test for pours but the darkening, caused by the SSS, makes them too dark for the purpose. I will experiment on some basic geometry to find a mix of SSS and bump map that suits my needs. The colour map is very understated.

 

Hmmm. I guess that I should up-res the maps.

[robcat2075]

What looks like freckles were actually meant as a test for pours but the darkening, caused by the SSS, makes them too dark for the purpose. I will experiment on some basic geometry to find a mix of SSS and bump map that suits my needs. The colour map is very understated.

 

 

Very interesting experiments.

 

Perhaps on the bump problem... you could do one regular pass with SSS but no bump maps and a white pass that only has bump maps turned on and then multiply them together in compositing to get the shading of bump maps without SSS distorting it.

[Paul Forwood]

Thanks, Robert. That is worth trying but I would like to avoid compositing if possible.

 

This was my last effort:

Mapped_A04i.mov

[NancyGormezano]

looking terrific!

[MJL]

Wow, That looks good!

[TheSpleen]

looks great!

[Paul Forwood]

Thanks, people!

 

I'm going to get on with modelling today and will be working out some of the details for his costume. This was a quick test of bump map strength in A:M 16, applied to the front of his waistcoat only. It looks too much like a smoking jacket, so I won't be using this, but I am happy to see that bumpmaps are working very well in A:M16 Beta5.

 

Mapped_A04k.mov