IRTC anim topic: Impostor!

General discussion about JPatch

IRTC results!

Postby sascha » Mon May 23, 2005 6:26 pm

First place, yippee!!! :D

Altough it scored well in all three categories, I think it was the plot that was decisive (it's got the second highest "Concept, Originality, Interpretation of Theme" ranking of all time! - still on top is the unbeaten Pool Shark from the first IRTC animation year 1998).

David, thanks a lot (again)! I wish I had such funny plot ideas on my own!
I'm looking forward to see your "Explosion" animation, I think the plot is (once more) great, so it has a good chance to do well.

I'd also like to thank all of you here on the forum for your comments and feedback!
sascha
Site Admin
 
Posts: 2792
Joined: Thu May 20, 2004 9:16 am
Location: Austria

Postby dcuny » Mon May 23, 2005 9:55 pm

Thanks, but it was your hard work that brought it to life! I was amazed to see how well the final animation worked out. 8)

If you want to have a go at animating the Explosion script, you're welcome to it. At this point, I'm just struggling with JPatch, trying to get it to do things that it's not quite ready to do... :?
dcuny
 
Posts: 2902
Joined: Fri May 21, 2004 6:07 am

Postby pndragon » Tue May 24, 2005 8:20 pm

Congratulations, Sascha!!!


--Jim
pndragon
 
Posts: 591
Joined: Sun Dec 05, 2004 1:27 am
Location: North Carolina

Postby sascha » Wed May 25, 2005 7:22 am

Thanks!!!

If you want to have a go at animating the Explosion script, you're welcome to it. At this point, I'm just struggling with JPatch, trying to get it to do things that it's not quite ready to do...

It's your turn :D
Seriously, I'd better work on JPatch - I think "The Impostor" showed what is possible with JPatch right now, so I'd better focus on implementing new features before making a new animation.
This does not mean that one could not do better, and your script for "Explosion" has got a lot of potential.

As for the modification you sugested: I think implementig the scene graph (or object hierarchy) is not difficult, I just have to take care to make it "forward compatible" to allow dynamic object bindings in the future.

Still, using that to fake bones is overkill - I'd rather limit the motions and go with morphs (I'd say you can rotate up to 10 degrees with a morph - thats what I've used e.g. to turn the clerks head, etc. And, you can use multiple morphs building up upon eachother, like I used for the hand pressing the button - but this still is quite limited) or wait until JPatch supports bones.

Having multiple viewports and parallel projection should be no problem, but right now I have no clear picture of how to integrate the modeler and the animator - If I rewrite the "viewport" code, I'd like to be able to use it in both, the modeler and the animator (e.g. to have a perspective preview available in the modeler).

I'll be on vacation next week and I have a lot of other things to do, so I really can't make any promises.
sascha
Site Admin
 
Posts: 2792
Joined: Thu May 20, 2004 9:16 am
Location: Austria

Postby dcuny » Wed May 25, 2005 12:05 pm

Still, using that to fake bones is overkill - I'd rather limit the motions and go with morphs (I'd say you can rotate up to 10 degrees with a morph - thats what I've used e.g. to turn the clerks head, etc.
I put together a simple model of an arm, and added a slew of morphs so that it slowly bent at the elbow. Each morph was about 12 degrees. I then tried to animate it with something like:
Code: Select all
frame 1: morph 1, 100%
frame 2, morph 1. 0%, morph 2 100%
frame 3, morph 2. 0%, morph 3 100%
frame 4, morph 3. 0%, morph 4 100%
frame 5, morph 4. 0%, morph 5 100%
...
You can guess what happened: all the morphs began at frame 0, instead of the prior frame. What I should have done was::
Code: Select all
frame 1: morph 1, 100%
frame 2, morph 1. 0%, morph 2 100%, morph 3, 0%
frame 3, morph 2. 0%, morph 3 100%, morph 4, 0%
frame 4, morph 3. 0%, morph 4 100%, morph 5, 0%
frame 5, morph 4. 0%, morph 5 100%, morph 6, 0%
...
Argh. :roll:

Or are you suggesting that I limit the total rotation to 10 degrees in one direction?

For the same animation, I had made the arms independant objects, so that I could rotate them at the shoulders. SInce it was in wireframe, I couldn't really tell which direction the elbow bend was going, until I rendered the frames, and noticed that the arms were placed on the model backwards, so they bent in the wrong direction. Feh. :x

And, you can use multiple morphs building up upon eachother, like I used for the hand pressing the button - but this still is quite limited) or wait until JPatch supports bones.
Layering rotating morphs has the unhappy effect of flattening out the mesh. For example, I created a morph of the arm rotating at the shoulder, and one of it rotating at the elbow. Individually, the morphs were just fine. But when I combined them, they caused the arm and hand to flatten out - not the effect I was going for. :(

I'm sure working with child objects will come with it's own special set of challanges. :?
dcuny
 
Posts: 2902
Joined: Fri May 21, 2004 6:07 am

Postby MU » Wed May 25, 2005 11:44 pm

hey, congratulations :) :) :)

just visited irtc.org -what do I have to see - 1. place :)

I did not doubt that it would be at the first 3 places, excellent work!

Mark
MU
 
Posts: 16
Joined: Mon Nov 01, 2004 1:57 pm

Postby sascha » Thu May 26, 2005 9:46 am

Thanks Mark! :D

David,
You can guess what happened: all the morphs began at frame 0, instead of the prior frame.

That's easy to correct. I know you posted a method about how JPatch could automatically do that, but I think it woudln't work. It would work in simple cases, but imagine a character with 50 or more morph-targets. JPatch has no concept of key frames. It has individual (and completely independent) motion curves. I don't see a reason for treating the last key of any curve as the last keyframe - instead, JPatch treats the last key of the active motioncurve as the last keyframe. If you animate a characters hand, do you really care if there was an eye-blink 3 frames ago? - should that really make a difference on how the morph targests on the hand should be treated? I don't think so. What I could imagine for a future version are "morph groups": You could toggle between "standard" and "group" mode - where it treats a key in the active group as the last keyframe. This way, all phoneme morphs could be grouped, for example.
What I should have done was::
Code:
frame 1: morph 1, 100%
frame 2, morph 1. 0%, morph 2 100%, morph 3, 0%
frame 3, morph 2. 0%, morph 3 100%, morph 4, 0%
frame 4, morph 3. 0%, morph 4 100%, morph 5, 0%
frame 5, morph 4. 0%, morph 5 100%, morph 6, 0%

Actually no. I think it works best if the morphs "build up":
frame 1: morph 1 100%

frame 2: morph 1 100%; morph 2 100%
frame 3: morph 1 100%; morph 2 100%; morph 3 100%
frame 4: morph 1 100%; morph 2 100%; morph 3 100%; frame 4 100%

Of course you'd have to create the morphs that way. That is, first create morph 1, then leave it at its 100% position and create morph 2, leave it at 100% and create morph 3,...
And you'd have to use linear interpolation.

Or are you suggesting that I limit the total rotation to 10 degrees in one direction?

It depends. If you want to rotate something for 60 degrees, you could create 6 "layered" morphs, rotating 10 degrees each. But, if you e.g. want to rotate a head for 60 degrees and the head contains morphs to rotate the eyes, the eye morphs won't work anymore when the head is fully rotated (you would need true rotations and a hierarchy - aka bones).

Layering rotating morphs has the unhappy effect of flattening out the mesh. For example, I created a morph of the arm rotating at the shoulder, and one of it rotating at the elbow. Individually, the morphs were just fine. But when I combined them, they caused the arm and hand to flatten out - not the effect I was going for.


Yes, but normally you would not move your shoulder and elbow joints individually - if you reach out, you'll move them simultanously - so you could create a morph to "reach out" the arm using e.g. 4 morph targets (each morph building up on the previous one) - just like I did it in Impostor when he moves the mouse - I actually had only two morph targets - one for moving the hand to the right, one for moving it forward. By setting the limits to -100% to +100% I was able to move the mouse to any position. Both morphs affected the shoulder and the elbow.

I should note that I am absolutely aware about the limitations of this approach. It is a workaround that makes sense only in very simple cases. If you have just one shot that needs to show a character reaching out his arm, you can do it with morphs. If your animation heavily depends on gesticulation and complex arm and body motions, you will have to wait until bones are supported.

I'm sure working with child objects will come with it's own special set of challanges.

It's different to working with morphs, but I'm quite sure it's not any easier. The advantage of the "child objects" would be support for true hierarchical rotations, so you could simulate a bone system, but you have to do everything by hand - which will be very cumbersome. And it anly works with independent objects (i.e. the lower arm is a different object than the upper arm, the torso, the head, etc. - the bone system would work on individual controlpoints of a single mesh).
The advantage of the morphs approach is that you can preview and fine-tune it in the modeler, and that it works on entire characters (i.e. without splitting into several objects, one for each bodypart). In the animator, all you have to do is move some morph-sliders. The drawback is that using morphs for rotations is only a poor approximation and very limited (no hierarchies).
sascha
Site Admin
 
Posts: 2792
Joined: Thu May 20, 2004 9:16 am
Location: Austria

Postby rjh » Thu May 26, 2005 5:37 pm

Hey Guys,

Thought that I would jump in and lend my 2 cents worth since I have been animating exclusively through morphs for some time now (without a UI exporting rib files with Hamapatch and then using an external "rib-tweener" program). As has been discovered, there are many difficulties animating this way.

morphs has the unhappy effect of flattening out the mesh


and

Individually, the morphs were just fine. But when I combined them, they caused the arm and hand to flatten out


This is why there has to be several morphs to approximate a rotation and lessen the flattening effect. Also, as Sascha has pointed out, layered morphs must built upon each other. IMHO, Jpatch has implemented the morph technique excelently. It allows an over-lap between complimentary morphs that lends a "smoothness" to the animation. And I still feel strongly that the seperation of morph channels is a key feature just for this reason. Careful attention has to be paid to the creation of the morphs in the modeling phase, and its purpose in the animation must be defined prior to the morph creation stage. For example, I have a single character with several "morph target libraries" modeled for him. Some are generic, and some very scene specific.

Again, just my 2 cents worth ... =>

I have become accustomed to this methodology, and recognize that it may not be for everyone. And don't get me wrong, I cann't wait until I get to use the bones implementation!

All in all though ... great job! ... the more I use it, the better it gets. When creating poses for my characters, I found that the pivot point can be positioned prior to rotation (saves me many headaches!). Also, the right-click to change between "bicubic" and "linear" interpolation in the Animator is a life saver as well. I'll post links soon to the projects that Jpatch has enabled me to develope ... I will need feedback! ... One cannot operate effectively within a vacume forever. Thanks again!

Rob

BTW: I almost forgot ... CONGRATULATIONS!!! on the IRTC win; very well deserved!
rjh
 
Posts: 179
Joined: Thu Dec 30, 2004 10:33 pm

Postby sascha » Thu May 26, 2005 6:46 pm

Thanks Robert!

layered morphs must built upon each other

I'll later add a feature to do that automatically - Right now morphs are only defined at their 100% value. The new feature woul enable you to define targets at any values, e.g. at 25%, 50%, 75% and 100% - This makes it easier to use the morp in an animation as you onle have 1 slider to control several consecutive morphs.

About bones: I've started to work on the tools redesign I mentioned before. It's a bit more work than I expected (the undo/redo things must be adapted as well), but I think I've found an elegant solution and a first test with the move tool looks good.
It's one of those changes below the surface and once it's done you'll not see any difference :? But on the other hand it is a prerequisite for future features, so there's no way around it...

Btw, David, I didn't release the source of the latest version, so if you need it (e.g. to include the latest Inyo version) just let me know...
sascha
Site Admin
 
Posts: 2792
Joined: Thu May 20, 2004 9:16 am
Location: Austria

Postby dcuny » Thu May 26, 2005 8:31 pm

Sascha wrote:Btw, David, I didn't release the source of the latest version, so if you need it (e.g. to include the latest Inyo version) just let me know...
If it's not a problem, that would be great! I've made a couple changes to Inyo since the version I sent you, although I still haven't reconstructed most of the work that I lost after accidentally formatting my hard drive. I also enabled JPatch's passing back the parameter string on the textures, which lets me do ad-hoc sorts of effects.

But if you're busy, don't worry about it. I won't really worry about rendering effects until rather late in the animation. I've still got to model a character, and get a good grasp of the animator module. That's likely to keep me busy until you get back from vacation.
dcuny
 
Posts: 2902
Joined: Fri May 21, 2004 6:07 am

Re: IRTC results - Screensaver

Postby mstram » Thu Sep 15, 2005 3:19 am

sascha wrote:First place, yippee!!! :D


Hi Sascha,

Can you give me an idea of how you did the screensaver in your animation?

I'm curious both about how to generate the patterns, and secondly how to have a "sub animation" embedded in the main anim.

Did you do it with one program, or post process with something like Premiere?

Wondering if it's possible to do with just pov-ray ?

Mike
mstram
 
Posts: 4
Joined: Tue Aug 30, 2005 12:00 pm
Location: Toronto, Canada

Postby sascha » Fri Sep 16, 2005 3:43 pm

I wrote a little program that generated the screensaver animation (a lot of png files, like screensaver0001.png, screensaver0003.png,...).
In POV-Ray it's just an ordinary image map (mapped to a plane) - it simply uses the file "screensaver" + frameNumber + ".png".

This works with all kinds of animations, as long as you have the individual frames available as image files. You can use mplayer to do this - I did some tests with the clerk watching "Terminator II" by using the scene the T1000 morphs out of the floor - but as the IRTC rules don't allow to put matirial copyrighted by others into the animation I went on with the screensaver animation.
sascha
Site Admin
 
Posts: 2792
Joined: Thu May 20, 2004 9:16 am
Location: Austria

Postby squirrelhavoc » Fri Oct 14, 2005 12:11 am

sascha wrote:I did some tests with the clerk watching "Terminator II" by using the scene the T1000 morphs out of the floor - but as the IRTC rules don't allow to put matirial copyrighted by others into the animation I went on with the screensaver animation.


You should release a new version with those features put in.

I was looking around the IRTC site, and there was some really old info on prizes, and I was curious if you received anything for your work.
Squirrel Havoc

We live as we think, very very slowly....
squirrelhavoc
 
Posts: 180
Joined: Tue Jun 28, 2005 11:17 pm
Location: Oklahoma, USA

Postby sascha » Fri Oct 14, 2005 8:29 am

You should release a new version with those features put in.

A good idea. It would require chaning the first two shots - e.g. showing the clerk watching TV (lit by a flickering blue light), then showing the monitor with the Terminator II movie scene - just after the T-1000 morphs out of the floor in the movie, the rusty impostor starts morphing out of the museum hall...

I had a plan to re-render it in HDTV resolution - here are two images at 1280x720 and 1920x1080 - but I think the "shaders" are of too low quality for that resolutions. E.g., the tiled floor looks "flat" in the first image, so I decided that it would be too much work. Maybe when I've more time...

I was looking around the IRTC site, and there was some really old info on prizes, and I was curious if you received anything for your work.

No, I didn't receive any prize. The contest is run by volunteers, and I think they don't have any sponsors right now, so there are no prices other than fame and glory :wink:
sascha
Site Admin
 
Posts: 2792
Joined: Thu May 20, 2004 9:16 am
Location: Austria

Postby squirrelhavoc » Sun Oct 16, 2005 11:12 pm

I need help, how did you pull off those subtitles? Im looking for a way to add some to my animations, without the use of post processing with paint shop pro or gimp
Squirrel Havoc

We live as we think, very very slowly....
squirrelhavoc
 
Posts: 180
Joined: Tue Jun 28, 2005 11:17 pm
Location: Oklahoma, USA

PreviousNext

Return to General discussion

Who is online

Users browsing this forum: No registered users and 4 guests

cron