PDA

View Full Version : V10.5 Alpha 4 *blended normals feature*


zandoria
03-08-2003, 07:45 AM
New alpha up on the ftp

ftp://ftp.hash.com/pub/updates/windows/am2003/alpha/
ftp://ftp.hash.com/pub/updates/powermac/am2003/alpha/

Release info
http://www.hash.com/am2003/alpha


One of the new features is "blended normals" which are described in the release notes. This seems to be the solution to creasing that a lot of people have asked for!:beer:

They have asked for feedback, so I hope everyone will test it out!

ragtag
03-08-2003, 11:22 AM
YES.....we might actually get rid of creases. :bounce:

Only down side is that now I have to upgrade from good old 8.5. :)

Ragnar

Wegg
03-08-2003, 06:50 PM
Can you make some examples for us?

JB
03-09-2003, 06:26 AM
Yes examples please, I won't believe it till I see it. :)

Drakkheim
03-09-2003, 07:04 AM
First, control points can now use another method of shaping splines: there are the original perpendicular normals, and now the new biased normals. Biased normals make models subtly smoother, and eliminate the Poissant “crease”, (named after Yves Poissant, who documented it so well).

before you know it Yves will have his own talkshow. :)

well it's doing some really funky stuff for me but it's still early in alpha, anyone else have more success?

-drakk

My Fault
03-09-2003, 09:14 PM
Originally posted by Drakkheim
before you know it Yves will have his own talkshow. :)

well it's doing some really funky stuff for me but it's still early in alpha, anyone else have more success?

-drakk

Same here, but still much testing to do. My problem is, is this really fixing the creasing issue or what Yves "thinks" is the creasing issue?

And why is it that Models taken out of AM and rendered in other packages not show these creasing problems? I've been pulling models out of AM and rendering in Lightwave and they look significantly better there.

Anybody else, who is technically minded, have opinions why AM has this issue?

Glad to see Hash is working on it though! :thumbsup:

My Fault
03-09-2003, 10:26 PM
The new options to the porcelain.mat let you dial in how much smoothing you want. 15-20% seems to fix the run of the mill creasing issues without losing detail. Very nice! :thumbsup:

Wegg
03-09-2003, 10:29 PM
Show us! Do a creasy model from the AM cd and do a rendering with 5%, 10% and 15%. . .

My Fault
03-09-2003, 10:35 PM
Originally posted by Wegg
Show us! Do a creasy model from the AM cd and do a rendering with 5%, 10% and 15%. . .

Will do, I'll have it up ASAP!

Just need to find a good creay model...

My Fault
03-09-2003, 11:32 PM
Here's one test using the Goofy model off of the CD. I'll try to locate an another model that may show off normal creasing better than this, but this gives you an idea of how it works.
http://www.2dornot2d.com/images/CreaseTest1.gif

My Fault
03-09-2003, 11:57 PM
Hmmm, was trying a few things out and am noticing problems with 5 point patches and hooks. I'll try to throw up some more examples later tonight.

Roger Eberhart
03-10-2003, 01:02 AM
Where'd they put the new porcelain.mat? I downloaded alpha 4 and I can't find it.

Drakkheim
03-10-2003, 01:12 AM
its the same one as is on the AM cd.
the new ver just does stuff to it :)

zandoria
03-10-2003, 01:38 AM
Here is one with biased normals (except for the teeth). When the teeth were biased, it rounded them off a little--but you can set the normals to be either biased or perpendicular point by point.

I used the porcelain at 50%, which was the best look. You can drag and drop the porcelain on different groups and set the normal strength independently on each group.

I'm still getting a PWS corruption when I try to drag a material from the PWS, but there is no problem if you use the library window. So I just made a shortcut to the material on the CD and dragged it onto the groups from the library.

The combination of these two techniques should help people get around creases, but the blended normals alone will not fix bad modelling (dead ended splines on the "blonde " and "brunette" models from the 2000 CD still caused a crease) I haven't had a chance to try porcelain on them to see if it will smooth that out.

http://www.zandoria.com/images/Bugs.jpg

My Fault
03-10-2003, 01:47 AM
Originally posted by zandoria
Here is one with biased normals (except for the teeth). When the teeth were biased, it rounded them off a little--but you can set the normals to be either biased or perpendicular point by point.

I used the porcelain at 50%, which was the best look. You can drag and drop the porcelain on different groups and set the normal strength independently on each group.

I'm still getting a PWS corruption when I try to drag a material from the PWS, but there is no problem if you use the library window. So I just made a shortcut to the material on the CD and dragged it onto the groups from the library.

The combination of these two techniques should help people get around creases, but the blended normals alone will not fix bad modelling (dead ended splines on the "blonde " and "brunette" models from the 2000 CD still caused a crease) I haven't had a chance to try porcelain on them to see if it will smooth that out.

http://www.zandoria.com/images/Bugs.jpg

What did the render look like without the biased normals?

Also, I'm not having any issues with dragging the materials from the PWS. Is anyone else having the same problem William is with that? What video card do you use?

zandoria
03-10-2003, 01:59 AM
It is a Radeon 7200 all-in wonder.

I didn't render the model without the normals and porcelain, I noticed that you could drag the percentages in the properties panel, and the shaded mode would change and show the mesh changing in real-time so I just fiddled with it untill I liked the look of the mesh. Then I dragged "bugs" into the choreography, posed and rendered.

My Fault
03-10-2003, 02:03 AM
Originally posted by zandoria
It is a Radeon 7200 all-in wonder.

Jut a quick note, most of my PWS issues went away when I dumped my ATI card and moved back to Nvidia. Not saying it is definitely the problem, but AM, Lightwave and Messiah all had less problems on my machine when I made the change away from ATI.

zandoria
03-10-2003, 02:10 AM
Well it works for me fine. I like being able to watch TV on my computer too :)

Kalimol
03-10-2003, 05:18 AM
Too...cool! Finally! Toon renders will rock so hard with that!

What was that number you guys were calling for that free upgrade from 9.0? :drool:

hoochoochoochoo
03-10-2003, 09:28 AM
What was that number you guys were calling for that free upgrade from 9.0?

Don't think it works that way.:shrug:

This feature looks really cool, maybe I can now withdraw the knife from my back that was stuck there not long ago -

http://www.cgtalk.com/showthread.php?threadid=43936&perpage=15&pagenumber=2

All I can say is maybe this forum should have started YEARS ago, think of what else AM would have solved from user pressure under cold public gaze....

John Keates
03-12-2003, 09:16 AM
Playing around with the blended normals feature, it seems that it does the job nearly right but it is not perfect. It goes 80% of the way but I still have to get bias twiddling to realy smooth things out.

Any thoughts?

John Keates
03-12-2003, 09:40 AM
Also, has anyone else noticed the renders being realy slow in 10.5. I am trying to make a big push for a faster renderer and would like to see the oposite happen.

zandoria
03-12-2003, 01:44 PM
If you are using multipass, that may be why your renders are longer... The "read me" warns of an increase of (I think) 6 to 20 times longer than with the normal antialias.

My Fault
03-12-2003, 04:37 PM
Yeah, multipass certainly fixes many issues but at an extreme speed cost. The renderer is definitely a few generations behind C4D, Lightwave, etc. but it's still an improvement over what was there before. It's a shame Martin wasn't more up to speed on current render technology. The programmers of those apps use a scalpel to improve their renderer, Martin is still using his damn hammer :p

zandoria
03-12-2003, 05:01 PM
Why knock it? At least now you will have the option based on the level of quality you want.

It seems to me that Hash is trying to give people what they want, but that doesn't seem to be enough for you 'My Fault'.

What project are you working on that is being held up by the rendering speed? We can all talk about the program all day long, but maybe we should just get back to work...

Haven't you noticed the other forums seem to be full of people showing off some amazing(often) work that they are doing, while this forum for A:M seems to have lost a lot of steam after it's initial cathartic venting? Now that Hash is addressing the users complaints, let's see some productive work being done!

My Fault
03-12-2003, 05:15 PM
Remember though, that this is still an alpha, so I'm sure speed will increase.

The main reason I'm pointing out the speed issue is that...
1. I don't think Martin will keep working so dilligently on the renderer if we don't keep the pressure on him.
2. It is much, much slower compared to the renderers in other competitive 3d packages.

I refuse to settle for second class tools when it is obvious Hash can do so much more. They are a pretty damned talented group and I think they are more then up to the challenge. Like I said before, this is still an alpha and I'm sure they will bump up the speed.

And "zandoria", it's hard to get back to work when I have to spend 5 hours for a render. It's great that you have all the time in the world, but I don't right now. But your right, at least the render looks really nice after those 5 hours. Maybe if I pour caffeine on my processor that will speed up the render times :p

My Fault
03-12-2003, 05:18 PM
Originally posted by zandoria
Haven't you noticed the other forums seem to be full of people showing off some amazing(often) work that they are doing, while this forum for A:M seems to have lost a lot of steam after it's initial cathartic venting? Now that Hash is addressing the users complaints, let's see some productive work being done!

Totally agree with you WIlliam. I think a lot of us are still exploring some of the new changes and that's why your not seeing a ton of great work yet, but it will come!

Sorry to have sounded so negative in my other posts but let's face it, renderers can never be fast enough. More power...MORE POWER!!! :eek:

John Keates
03-12-2003, 09:38 PM
Originally posted by My Fault
renderers can never be fast enough. More power...MORE POWER!!! :eek:

Couldn't agree more. Do you think that the people at Maxon are thingking "Well, our renderer is one of the fastest around and it looks stunning, lets just sit on our arses and relax". No- they are probably working on speeding it up and adding more power. If other renderers are 10 times faster than AMs' then that is like buying a render farm. Now that is value. I would love not to have to think too much about render speed so that my creativity is unhindered by such concerns but at the moment much of my stuff is almost dictated by the slowness of the renderer.

You will indeed see some quality work appearing if the renderer is improved. (oh, and I was using the normal renderer when I noticed slow output, I guess that it is an alpha thing).

jonmalexander
03-13-2003, 01:47 AM
I know this thread has moved to render issues, but I have a question back to the porcelain material in 10.5 alpha4. I can't seem to find where I can change the percentage of the porcelain material on my model. I look under properties and drop down all the triangles I can on the material, but I don't see any percentage options. Can somebody please point me in the right direction. Thanks!

My Fault
03-13-2003, 02:12 AM
You should have a property called "Normal Weight", that is where you change the percentage. If you do not, your porcelain.mat file may be corrupt. When I tried saving the porcelain.mat file in 10.5Alpha v4 it removed the "AverageNormals=TRUE" in the .mat file, thus making the porcelain feature not work. Either open up your .mat file in a text editor and re add it back (under the [ATTRIBUTE] section BTW) or re grab the porcelain.mat file off of your AM CD.

If you want to set your percentage, make sure you do it on the material link in the model and not to the original material file, that way it doesn't screw up the original .mat file and it will still be there after you close your project..

amsmf
03-13-2003, 08:53 PM
Just in case anyone here didn't see the other thread, HASH NEEDS FEEDBACK ON THIS:

From: support@hash.com
To: animaster@hash.com
Subject: Bias Normals vs Perpendicular Normals in v10.5a4
Date: Thu, 13 Mar 2003 09:59:40 -0800

We had hoped to get some concrete feedback supporting this (radical)
change in technology-
We got some nice "oh this is cool" comments, but no concrete evidence
that it was better, or that users preferred it with supporting data- If
we don't get any input by Friday, we'll be going back to the
traditional perpendicular normals rather than offering the bias normals
option-

Steve Sappington
Hash, Inc.
400 West Evergreen Blvd.
Vancouver, WA 98660
360-750-0042
-3D Modeling and Character Animation Tools!-

My Fault
03-13-2003, 09:14 PM
Ack, they need to give us a little more time to test it then just a week!

amsmf
03-14-2003, 02:37 AM
Originally posted by My Fault
Ack, they need to give us a little more time to test it then just a week!

I agree completely. I suggest everyone who is even *REMOTELY* interested in this new feature write steve at support@hash.com and say something along the lines of "please let us have some more time to test". I already did so and I would like to think that if enough people show interest that they may be willing to give a bit more time.

Kalimol
03-14-2003, 06:50 AM
D---. You'd think "this feature rocks!" would be enough! I don't have AM v10 yet, so I can't really respond to this...would mentioning that Blended Normals is the reason I'm planning on upgrading help?

Someone send Steve the bunny pic (he's a very convincing rabbit), and maybe a death threat or two. Geez!

amsmf
03-14-2003, 03:07 PM
Originally posted by Kalimol
...would mentioning that Blended Normals is the reason I'm planning on upgrading help?


I think that saying that would be an even better tactic than just asking for more time!!

amsmf
03-14-2003, 03:37 PM
>From: Steve Sappington <steve@hash.com>
>Reply-To: steve@hash.com
>To: animaster@animationmaster.com
>Subject: Bias (blurred) Normals
>Date: Thu, 13 Mar 2003 17:23:17 -0800
>
>
>Thanks to all of you who have provided useful feedback- for those of
>you working on providing actual project data to help us make our
>decision, we've extended the deadline to next Friday before we'll make
>our decision-
>
>Steve Sappington

Now we have a week to come up with examples.

My Fault
03-14-2003, 07:42 PM
I don't understand why they need to remove them if it is already in and working? Odd, very odd.

Roger Eberhart
03-14-2003, 07:48 PM
Biased normals are in and working....but they don't help. At least from my testing. I specifically recreated the "Poissant" crease and biased normals do nothing to alleviate the problem.

Wegg
03-14-2003, 08:19 PM
Just do a regular crease. One that comes from a bunch of 3s jammed together. . . Does it help with those?

Roger Eberhart
03-14-2003, 08:54 PM
Nope, doesn't help there either.

John Keates
03-15-2003, 01:23 AM
look: biased normals are a good step in the right direction. It means that I can *nearly* see what a model will look like once I have adjusted the bias handles (as I model). It doesn't go all the way, but I like it any way. Yes, it is wrong to expect people to use an alpha when we as paying customers should only be expected to use a release, but hey, we are the AM users.

Mind games... mind games.

Those folks at hash should work for the CIA.
I work hard for them. Corrupted files... "Please give easy steps to re-create problem". ....ooooooh

Life is hard enough.

Biased normals=good.

oochabooba

The new porcilain feature will get rid of 3p patch defects even on 0%. Yes... I don't know how that works but it does. Just use bias normals and porcilain on 0 and a whole load of crease issuse go. OK, a lot else happens instead, but there you go, we are using an alph (whatever that means).

I have been able to use 3 point patches and that makes me happy!!!

Ok, I have also just sold a painting to the township where Bosch comes from but I am happy for hash as well.
:applause:

Don't forget the new specular light. I love it.

Good night.

John.

koon69
03-15-2003, 02:18 AM
What are these new specular lights??

John Keates
03-15-2003, 09:09 AM
Sorry about the randomness in my last post, the wine got to me.

The new specular highlights can now be very big. Before, if you set the size to 1000 it made very little difference. Now, you can get a huge highlight that covers a whole side of a model (if you want). It is great for the soft sheen on skin and stops things from looking to plasticky. Of course it isn't all working yet (looks funy in a chor.)

Hope this helps.

zandoria
03-18-2003, 07:02 PM
The biased normal makes the surfaces seem "rounder", but it isn't going to smooth out a bad surface from poor modeling.. However the new control over the porcelain material does seem to be a panacea for smoothing selective areas, by assigning it to named groups.

I like the combination of both. I'm going to send in some more examples, but right now I'm unsure of the value of this feature... I think maybe the porcelain changes are the better solution.

On the "bugs bunny" picture, I used a combination of the two, not just blended normals --so it may have given the wrong impression of which technique was more responsible for the smoother appearance...

HellBorn
03-22-2003, 11:17 PM
Just got my AM v10.
I got really exited after working for several ours without a crash.
So I gave a human model a try and..the dreaded creases are still there. I had just decided that it was time to put AM back on the shelf untill they fixed it when I read about the new Blurred Normals. Suppose I give it a try before giving up. The old porcelain did never work for me as It always looked ugly at hooks.

I'm really proud to be one of the so called naysayer on this list. We really got Hash moving didn't we.
(The people on the AM list and Hash would of course deny it ;) )

My Fault
03-23-2003, 12:00 AM
Aye, it's great to see Hash being so commited to really working the kinks out. The way they are going this could be a great year for Hash and their users. It's tough being patient, but I think (hope, hope, hope) it will be worth it! :thumbsup:

HellBorn
03-23-2003, 08:58 PM
Hmm..
The new biased normals option...
It just look as it moves the roundness aliitle. Hard to explaind. But I can say that I get less creases.
The new variable porcelain settings seems to be of more use. At 50% it removes most creases while not smoothing the shape to much. It also seems that Ii wont create artifact at the hooks.

CGTalk Moderation
01-14-2006, 03:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.