Animating vectorized sketches, point by point

Wondering how to accomplish a certain animation task? Ask here.

Moderators: Fahim, Distinct Sun, Víctor Paredes, erey, Belgarath, slowtiger

Post Reply
Amets
Posts: 6
Joined: Thu Sep 14, 2006 2:08 am

Animating vectorized sketches, point by point

Post by Amets » Fri Oct 06, 2006 8:12 am

Hey there! I've scanned 5 views of a character's face in order to open them in anime studio and then vectorize the image welding points for each part of the character's face; nose, eyes, hair, etc.

The idea is having one view in frame 0, then in frame 5 the other, then the front view in frame 10, and then the simmetric views of the first two in their respective frames to make a head turn.

It's a tedious work I've got to admit I've been trying to make it look well like for 3 hours but always end in the same thing, due to the change of the points in the other frames the original aspect of the character's face always ends all messed up, and some parts end up white, in fact, is so messed up that I can't longer recognize what points I'm translating or why are they affecting the color of the area where they are.

So, after 3 hours of useless work I wanted to ask you guys if this what you normally do to achieve the head turn, I already read the last thread about head turning, but my question is more specific, translating the points is the only way to achieve this, or is there another way?

Thanks a lot!
aphid
Posts: 8
Joined: Sat Apr 01, 2006 6:03 pm
Location: UK

head turns in anime pro

Post by aphid » Fri Oct 06, 2006 11:43 am

Hi, I have used switch layers for head turns,make like 5 heads like you did, if you work in the switch layer you keep the same number of points. It would be up to you how often you changed the frame, maybe three frames or so. The more positions you make the smoother it looks. Not sure if I explain this well. I would show you, but I don't know how to put things on here! It is quite a lot of work making the switch layers.
Aphid
User avatar
Rasheed
Posts: 2008
Joined: Tue May 17, 2005 8:30 am
Location: The Netherlands

Post by Rasheed » Fri Oct 06, 2006 12:35 pm

If you don't know how to do head turns, I suggest you start with a simpler example, instead of a traced set of heads. Once you understand how it works in a simple example, you can apply that knowledge to a more complex model.

It is really much better to invent your own method for a particular animation problem than trying a method that someone else recommends.

Having said that, I really like switch layers. However, for a convincing effect, you need to animate on ones (1 keyframe per frame), which means a lot of drawing in case of a complicated model. So my advice is to keep your model as simple as possible, but detailed enough for what you want to accomplish.

The trick is to make your model "animatable", which means you don't have too much trouble animating your model. This often means you have to try some animation with your model and modify it accordingly, in several iterations, until your model is useful enough, but still close to the original sketches.

I hope this helps.
User avatar
slowtiger
Posts: 5489
Joined: Thu Feb 16, 2006 6:53 pm
Location: Berlin, Germany
Contact:

Post by slowtiger » Fri Oct 06, 2006 1:54 pm

Convincing head turns are among the most difficult tasks in animation, especially for a beginner. You could tackle down the problem if you split it into several smaller parts:

Do the head turn in drawing on paper first. Do it often. Flip the drawings, or scan them in and let them run in sequence so you can control the movement. Is the volume consistent? Do face features move with the head, or do they float around?

Start with a simple sphere with a nose first. Don't proceed until it really turns. Only then add eyes, and other features.

Have a look into good animation source books, like "Animator's survival kit".

Rethink your attitude towards head turns. Does the character really has to turn the head? How far? At which speed?

Many recent animation series do without any head turn, the just flip from front view to profile. Anime often feature head turns of just some degrees, which can be achieved by just shifting the facial features ever so slighly. Classical animation often combines the turn of the head with a change of expression, at least with a blink of the eyes, to hide any flaws in the bigger changes. Also they nearly never do the turn horizontally, intead the middle position of the turn shows the head slightly bowed down. See "Animation: the Illusion of Life" (there's a great list in there which sets any number of inbetweens in a head turn in relation to an expression!) or Preston Blair's book for reference, or study classical animation frame by frame.

Doing that head turn in AS in a way that it is ready to be interpolated is much construction work - not easy, and very likely to fail miserably the first dozen times. As long as you are still gathering experience, using the switch layer trick is the best and fastest solution, because you can do a single vector layer face for each of your drawings without having to bother about the number of points and such (remember to not use "interpolation" then!).
User avatar
Rasheed
Posts: 2008
Joined: Tue May 17, 2005 8:30 am
Location: The Netherlands

Post by Rasheed » Fri Oct 06, 2006 2:40 pm

Here's another useful tip for beginners.

An easy way to create your drawings of a head turn with pencil and paper is to create a head shape out of clay (or plasticine) and stick a nose on it. Put the head on a pedistle (a blob of clay) and put that on a turn table (or a plate as a make-shift turn table). Turning the turn table, you can view the head shape with nose from all sides and draw it.

The next step would be to add eyes, ears and a mouth to your clay model and try to draw that in a head turn, frame by frame.

Use a big clay model, simple to draw shapes (sphere-like objects), and sketch big.

A next step could be to use life video and study that in slowmo, or even frame-by-frame, and draw from memory. In life video you will see some motion blur, that increases with faster head turns. You will need to find a way to show this motion blur in lines. There are as many ways as there are artists.

I haven't tried this myself, but I know that some animation studios have used similar methods (and others) in the past.

The point is that you should put a vast amount of time and energy in being prepared before you start to animate.

Furthermore, as slowtiger has pointed out in another thread (Walk cycles using The Animator's Survival Kit), using the computer to draw animation sketches can hinder your creativeness. "Greykid Pictures" wrote it like this (see this message):
We design our characters outside of Moho, just a normal sketch. Then we import the image and build the models with Moho, (using the drawing and paint tools) The we add the bone system.

We build very comprehensive models, taking into account all of the motion we'll want to produce with that character.

The reason why the animation works so well is because we believe in thorough preperation. If you plan everything out carefully the model can do pretty much anything.
That's good advice from a great animator.
Amets
Posts: 6
Joined: Thu Sep 14, 2006 2:08 am

Post by Amets » Wed Oct 11, 2006 12:22 am

Ok I've finished the head turn, I made it moving the points strategically every 5 frames, making the head turn less than a second long.

The next sketches were drawn, scanned and then imported in Anime Studio Pro, I put the first view in frame 0, the second in 5, the third in 10, and so on, being the fourth and fifth views symmetries of the first and second ones.

Image

Image

Image

Image

Image


Image

(The head turn of the animated gif should be less than a second long, don't know why adobe imageready rendered it this way)


This work took me like 4 hours, maybe because it's my first head-turn attempt, but I don't think that if I become a pro in this software some day, it will take me less than 2 hours.

I'm believing this isn't the kind of style of animation you use normally to do a head turn. First because of the time needed, second because of the imperfections you've got to fix after completing the move of the points in each frame (it's like if the program loses its memory and doesn't remember where the points were in the previous frames) and third because of the overall product.

Some of you replied saying that switch layers could do it, what is that? I know it's a bit lazy from me to ask you this, because of all the tutorials and threads that should be about the subject around here. But I just wanna know that if this style of animation wins versus the time, complexity and quality of the "move-points" style.

Thanks a lot for the replies :wink:

P.S.: if anyone wants the .anme file to see what I did just say it.
Bones3D
Posts: 217
Joined: Tue Jun 13, 2006 3:19 pm
Contact:

Post by Bones3D » Wed Oct 11, 2006 2:45 am

"Switch layers" is basically a method where you simply jump cut over a series of individual images without trying to tweening them using point morphs.

For example, lets take a look at the head sequence you created:

Image

If you were to create a series of layers, each with one frame of the sequence above, you could then use those layers to create an animated head movement over time by swapping out one of layer for the next layer in the sequence. (Usually, this is done by setting all but one of the layers to invisible in the layer properties.)

Here's how it would look when finished:

Image

While it's not as smooth as a fully tweened point morph, the motion itself is still implied pretty convincingly.
8==8 Bones 8==8
User avatar
7feet
Posts: 840
Joined: Wed Aug 04, 2004 5:45 am
Location: L.I., New Yawk.
Contact:

Post by 7feet » Wed Oct 11, 2006 2:49 am

Just a bit on why the .GIF is slow - each frame of an animated gif will be displayed for a specific amount of time. Imageready defaults to .1 seconds per "frame", so you get an effective frame rate of 10 frames per second. To get it up to speed, you need to adjust the time for each frame. As an example, setting the frame time to .042 will get you to something close to 24 frames per second. I use animated gifs rarely, and I don't know if there's a way to set another default, I usually end up setting each frame by hand.

Looking good, there, too.
User avatar
heyvern
Posts: 6964
Joined: Fri Sep 02, 2005 4:49 am

Post by heyvern » Wed Oct 11, 2006 3:57 am

If you can do a very good head turn using point motion in two hours...

... that sounds pretty good. I would expect as you get better at it you could cut that time down even further. It doesn't take me two hours to do the turn IN ANIME STUDIO... however the combined time of doing the sketches, refining and tweaking adds up.

Two hours is not... that long in the scheme of things. 4 hours may be pushing it a bit.

Another thing to consider... once the head turn is done... you can reuse it over and over. You can copy the basic motion and change the expressions etc. This cuts down on time later on.

I have found I don't need as many frames as I use to for a head turn. Using the switch layer option I can do a convincing turn from 3/4 to front with just one step... one for 3/4 one for straight ahead.

If I need additional steps to further refine the turn I can grab intermediate steps from with in the original and tweak those.

-vern
User avatar
slowtiger
Posts: 5489
Joined: Thu Feb 16, 2006 6:53 pm
Location: Berlin, Germany
Contact:

Post by slowtiger » Wed Oct 11, 2006 8:52 am

Well, it's done ... but I would rather call it a "proof of concept" than a finished head turn. If I were the director I wouldn't approve the scene unless the hair stops moving on its own ...
Amets
Posts: 6
Joined: Thu Sep 14, 2006 2:08 am

Post by Amets » Wed Oct 11, 2006 8:20 pm

Yeah, you're right Bones3D the switch layers work a lot better than the morph points. I haven't tried that myself but I could see that after I saw your head turn using the views of my character's face. The result is sweet.

If I could draw 5 more views for example, using the method rasheed mentioned (watching the head turn of a character's face using a clay model) and then applying switch layers on anime studio, the results could be amazing!

But that would take a lot more work to the one that draws rather than to the one that animates, wouldn't it?

thanks for the replies,

cheers
User avatar
Rasheed
Posts: 2008
Joined: Tue May 17, 2005 8:30 am
Location: The Netherlands

Post by Rasheed » Wed Oct 11, 2006 9:38 pm

A head turn in full animation is time consuming and will probably be created once and used several time. In the Simpsons I often see the same head turn of Homer, in which you can "admire" his dental work. I can imagine this head turn is recycled over and over again. Of course, I helps that grandpa Abe resembles Homer, so the same clip can be used with some modifications.

In Anime Studio, there is the shortcut of using 3D rotation, as you can see in the library animation Squeakly_10sec (File -> Import -> User Movies -> Squeakly_10sec). It has another look and feel than the method using drawn frames inside a switch layer, so it might not always be appropriate.

Another solution is to stage your animation so, that you don't need a full motion head turn, but rather a head flip and some rearranging of the face. In limited animation this will work just fine, as you can see if you study some of the modern TV animations on the networks.
Bones3D
Posts: 217
Joined: Tue Jun 13, 2006 3:19 pm
Contact:

Post by Bones3D » Wed Oct 11, 2006 9:49 pm

The sky is the limit as far as how smooth and detailed you want to get. Just be aware of how long it takes you to draw each frame if you're working on a deadline.

As for the point morph oddities, you can see it in the head nod demo I created a while back as well. I've been meaning to redo the demo using switch layers. (Though, I do like the convenience of having automatic tweening at my disposal.)

- Anime Head Nod
(Moho 5.3 Document)

The head object and the top hair both initially had weird bugs in them where some of the edges would intersect one another in odd places that eventually required correction with extra keyframes. For a lengthy sequence, this could consume a considerable amount of time beyond that which initially went into generating the original keyframes from your drawings.

With switch layers, you avoid much of this hassle, since you don't have to take the errors auto-tweening may create into account. In fact, switch layer frames don't even need to contain the same number of points per object, leaving the possibility of applying various special effects wide open.
8==8 Bones 8==8
Post Reply