Automated inbetweening

[Klaus Hoefs]

There are a few out for this by now. It all comes back to 2D-Mapping on 3D-primitives. One we had here is CrazyTalk, a few dollar app.
It also features gesture control, the setup (mapping) is done in a few seconds because all morphings and movements are predefined and can be adjust with sliders of animation. They have also a face capture module.
Here is a test we did with it.

CT_Test4.mov

(492.14 KiB) Downloaded 1640 times

And as it is obvious, same issues as ever with this technique.

[idragosani]

This kind of thing has its place, I think, for a specific effect (like some kind of creepy nightmare thing), but for general tweening, it starts going into that uncanny valley...

[Elodie]

Yes, but the angle of that turn is probably less than 30 degrees (and probably uses a 3D model?)

Nope, in that case, it's just a 2D drawing with the wrapping grid. But you are right, this kind of effect cannot work to turn a head or a character at 360°

[ZigOtto]

... What we see here looks like an animated motel sign silently glowing in the dark.

you've been too rude with him,
with your kind of harsh comment, once again, you've made a new and anthousiastic forumer escape
and probably definitively lost for our community !!!

apparently, he didn't appreciate his "secret-confidential killing" technic (cross-Warp + cross-fade)
to be detected and publically revealed to anyone reading this forum, ...and no one for "buying" his amazing inovation.

that's said, and to pick up this kind of CG inbetweening technic, I think there's room for improvements in our beloved tvpa,
specially around the warping grid feature (auto-detection of content with auto-fit grid, and so on ...)

[Paul Fierlinger]

Rude? What's so bad about a silent motel sign glowing in the dark? Not everyone can animate one of those so effectively...

[Fabrice]

Sykosan contacted me and does not want to answer this topic anymore.

Eh ??? ... ...

This topic was making so many people participating and debating, that's too bad we were only a few people to see the result of Slowtiger questions and challenge.
(it's also a bit impolite from sykosan to have removed the picture/animation : Slowtiger is the one who should have seen the result of the inbetweening process, more than anyone here ...)

You got 3 opinions and a question from Fabrice (how lucky he was to go to the forum so late !)

In fact, I was here reading the forum so lately because my train was arrived about 3 hours later than expected at the station, and also because I had someone who was expecting an urgent email/PM from me at the other side of the planet. (thanks SNCF + snow ... thanks ... same today ??? Oh well ... ok)

I thought I was unlucky yesterday because of the train, now I feel lucky to have seen the animation

Rude? What's so bad about a silent motel sign glowing in the dark? Not everyone can animate one of those so effectively...

Paul, If I use the comparison with the 2D animation process :
We have two keyframes : "being rude" and "say nothing" Please try to find some "inbetween". (it's a joke and it's said kindly, I hope it will be read in this way, since as french guy born in France, I'm not the best at doing jokes in english)

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

More seriously, I want to add something to my last post, since I have more time now to write :

IMO, anybody is free to use the animation technic he wants, but he then gets the results/renderings corresponding to the choosen technic.
The aim of my post was to point the required time to obtain the result, but I don't think it was a lack of interest nor a sarcastic post ( at least it was not my intention ... or I wouldn't have asked a question).
it's a rule, I always try to see the good and bad side of a technic/option/parameter ... isn't it the way a developper should work ?

Sykosan, I remember I wanted to post later today or tomorrow the same exemple as Elodie did (originally created by Peter Wassink in 2004 with the TVPaint FX-Stack technology, if I remember well).
Then I wanted to explain if possible that a complete turn around could be something complex or time consuming with such effects. That would have given an interesting discussion, but before posting a big post in english, I would have prefered to know the answer of my basic question.

Also, I wanted to post you a question about the eyes management : on the first keyframe drawn by Slowtiger, we can almost see only one of them, on the second keyframe drawn by Slowtiger : we could see the two eyes clearly.
In your animation, we could see a clear fade-in to make the missing eye visible.
So my question was : were you cleaning all the images by hand after using your process or could the process give better results with other parameters ?

but, at the end, I was about to sleep and I finally I just asked if you took a look at the FX-Stack, with the hope that someone could talk about the Wrapping Grid or other related FX like the Optical Flow, etc ...

http://www.youtube.com/watch?v=YrLnF7CQ8Ac (there's a link to the makers in the description)

Thanks Slowtiger for posting this !!
I really wanted to post this same exemple yesterday evening, but once again it was late and I couldn't find it easily on the web, so I stopped any long research.

Sykosan : It's interesting to see, and there is a more interesting debate to open here with the community :
Should we develop more like in the youtube video ? Or should we develop more in your direction ?
What does the people think ? Wouldn't it make TVPaint less or more intuitive/interesting/appealling, if we have to separate many strokes on several layers rather than doing everything on the same layer, like on paper ?

that's said, and to pick up this kind of CG inbetweening technic, I think there's room for improvements in our beloved tvpa,
specially around the warping grid feature (auto-detection of content with auto-fit grid, and so on ...)

You are right Zig, and that was one of my intentions when I was asking about the TVPaint FX-Stack / Wrapping ...
We use to receive so many requests and advices : reading people debating is something helping us to decide the future developpement priorities.

[idragosani]

I don't understand why he deleted all of his posts. It was an interesting discussion, even if some of us were skeptical about it. It's always fun to discuss animation technique, it can certainly help drive feature development.

[slowtiger]

I can only guess he feels embarrassed now and therefore deleted all traces of his short visit here. Oh my.

I always thought one of the greatest features of the online community was their willingness to share - everything, from links and code snippets to complete tutorials or even revolutionary new concepts.

[ZigOtto]

...
I always thought one of the greatest features of the online community was their willingness to share - everything, from links and code snippets to complete tutorials or even revolutionary new concepts.

you're right ST, but imo he came here not so much to share his "revolutionnary new concept" but to sell it , that's why he wanted to keep secret the effective process, remember, he was talking about "signing a contract" with the dev. team, to get back royalties as discoverer, inventor. Don't get me wrong, it's not to blame, just another approach, if you have something to sell, you must find your potential clients, customers, and arouse their envies.
he posted his shiny videos just as a teaser and expected from our community a large and of one mind "Wow !!! you're a genious !... I would kill my mother for such a plugin !"
unfortunately for him,, we've very various people here, including some ones very sceptic and prompt in criticising when it goes to "automate" part of the creative process in animation.

[Paul Fierlinger]

Thank you, Zig O. for putting this interesting and disturbing thread into proper perspective.
Your skeptic,
Paul

[Jotokutora]

Hello everyone,

Here is another technology that would complement TVPaint. There is no real trick, is pure drawing interpolation. Yes there is a procedure to it, but not intrusive just a small adjustment on the workflow and that's it. The Trick is on the interacting toolsets.

Here are some test I did. Most of them are two key frames. On others one in-between was converted as a 3rd Key to improve on the final results.

I hop you like it.

PS. The technology still experimental and not for sale at the moment.

1. http://www.youtube.com/watch?v=Q0UN4UQtUx4" onclick="window.open(this.href);return false;
2. http://www.youtube.com/watch?v=CilfZNC_LGY" onclick="window.open(this.href);return false;
3. http://www.youtube.com/watch?v=7dpljmT4VKo" onclick="window.open(this.href);return false;
4. http://www.youtube.com/watch?v=Zansp3t1ur4" onclick="window.open(this.href);return false;
5. http://www.youtube.com/watch?v=augKVWbLzys" onclick="window.open(this.href);return false;
6. http://www.youtube.com/watch?v=sQSI2EBWV5E" onclick="window.open(this.href);return false;

[Fabrice]

All the question is how and how long to prepare the images to achieve the results. (same question we asked to Sikosan)

[Paul Fierlinger]

And would only work for a limited kind of unappealing, mechanical style of drawing --

[ZigOtto]

yet another question : can we set a curve path (spline) for the interpolated positions between Key-A and Key-B ?
or better, 2 curves, one for the position in space (x y z), the other for the position in time (speed-curve) ?

[idragosani]

Like with the previous poster and their new tweening technique, the samples are very short and mechanical looking and still look like morphing ala Flash or ToonBoom. How would you accomplish a walking character with this technique? Or imagine a baby kangaroo (giant mouse?) hopping around, with his head leading into the hop, follow through on the tail, squashing and stretching, etc. How would this be accomplished?