Help - Search - Members - Calendar
Full Version: Drawn to CG
Hash, Inc. Forums > Featured > Work In Progress / Sweatbox
Rodney
Andreas Deja recently returned from vacation and posted a pretty cool set of animated crowd sequences by Milt Kahl. These were cut from Disney's 'The Rescuers' but luckily are still available to us. In only seven seconds Milt conveys so much material I simply had to study them. Amazing stuff.

But... I also wanted to dig deeper into some areas of A:M as well... 2D analysis tools/rotoscoping being two of those things... and so this oddity is the result. This analysis consisted mostly of quickly placed five point patches (the circles) and a few four pointers (picture frames, nose and bench) with the whole sequence re-lit in post via the Merge and Mix Post Effects in A:M Composite. I just wanted to play with A:M Composite...

There is a whole lot more I want to study from Milt's animation but my curiosity is satisfied for now. I can better appreciate how he pushed and pulled and turned things around. I do expect to return to this one.

If you haven't visited Andreas's blog recently you should check it out.

Believe it or not... this was also something of an investigation into a possible idea for use with the Rear View community project.
Rodney
This one has me excited. (It's the little things like this that keep me going)

Jamaal Bradley (animator at Dreamworks and instructor at iAnimate) posted this animation from Lady and the Tramp (animated by Milt Kahl) a few months ago:

http://vimeo.com/26587301

Note the use of the cycle there at the beginning. Six or seven cycles of Tramp snoring away... slightly modified on the sixth and seventh iteration.
Awesome.

I need to track down the finished scene to see how much of that cycle survived in the final film.
Rodney
Here's an example of how A:M can be used to clean up drawings for use as rotoscopes as well as with more traditional animation.
Some folks don't have a decent line tester to view their animated drawings with and A:M is more than up to that task.
Using Post Effects the line color can be tinted a different color too... faded out... inverted...
It might be easier to list what you can't do. wink.gif

Dropping the quality setting from High Quality to Medium in A:M's 'Save Animation As' dialog shaved off about 6MB without adversely impacting the line quality. A change in size would likely cut the file size in half again.
NancyGormezano
I am curious as to

1) what software did you use to capture the original vimeo clip of Lady and the tramp?

2) How are you using post effects? I assume you are doing a composite (assuming with a tga? exr? mov? sequence)? of lady and tramp. Can you show a screen capture of your project/composite hierarchy?

I always gets confused by the order in which one does a composite, specifically how to add/combine hierarchy of post effects. Very confusing to me, so I am missing something critical in understanding the workflow.
Wildsided
Nancy, It was probably screen capture software like Cam Studio or ZD Screen Recorder
Rodney
Nancy,
I started to respond but realized that a full explanation was getting a little out of hand.
I'm trying to put together something concise (and useful?) at the moment.

To briefly answer your questions so you don't have to wait...

QUOTE
1) what software did you use to capture the original vimeo clip of Lady and the tramp?


I used Camtasia to capture the sequence. While it is expensive I highly recommend the program because you can use it for a whole lot more than just capturing image sequences.

QUOTE
2) How are you using post effects? I assume you are doing a composite (assuming with a tga? exr? mov? sequence)? of lady and tramp. Can you show a screen capture of your project/composite hierarchy?

I always gets confused by the order in which one does a composite, specifically how to add/combine hierarchy of post effects. Very confusing to me, so I am missing something critical in understanding the workflow.


I'll do you one better than that. I've attached a Composite file that you can open in a text editor or A:M.

Opening the file in A:M will be problematic from two angles however in our classic 'turn a problem into a feature' approach read on and you'll quickly be post effecting your own imagery.

The first problem you are likely to run into when trying to import the Composite file is that there is no clear place to import the file from in A:M. I swear it was there a moment ago but... it's missing now.
No problem. From the folder where you save the composite file Drag and Drop the composite into A:M.
Note: You don't have to drag and drop to the location where Composite files/images normally are located in the PWS. You should be able to drag the file anywhere the cursor indicates that drag and drop is an option. Try it. It's fun.

Important: Since you don't have the image resource at the location the Composite file references, it will ask you for it. If you cancel out of this you'll likely crash out of A:M so... do not do that. Select an image even if it's not the image you ultimately want. Unwanted images need love and affection too. Then change the image inside A:M later to the correct image as desired.

I would be interested in hearing how the attached opens for you.


Rodney
I want to return to the main issues that keep people from using A:M composite.

QUOTE
I always gets confused by the order in which one does a composite, specifically how to add/combine hierarchy of post effects. Very confusing to me, so I am missing something critical in understanding the workflow.


I'm going to write this toward the complete newbie... hopefully we'll get some more folks to use A:M Composite this way.

As near as I can tell the required components of a Composite are:
- Composite container
- Image(s)
- Post Effect(s)

Steps to an effective composite
Note: There are other ways to composite images in A:M. We are discussing A:M Composite here. One of the benefits to using A:M Composite over other methods is that you can bypass the rendering process and save a lot of time re-rendering images. To save time: Render once. Convert and Composite.

1. Right Click in PWS and create a new Composite container
2. Create a Post Effect (any effect will do but I often use Mix)
3. Open/Add your imagery in A:M (If you've just rendered a sequence it may show up as available but it's better to use images your brought into A:M)
* Once you get a setup you like... save your Project. If something goes wrong you can then return to that saved state again.

Note that at this point we haven't actually created a composite. We are just set up for success. Composite container. Post Effect. Images. Saved Project. We are set.

- Drag and Drop the Post Effect onto the Composite
- Because this is a Mix Effect it'll have two image buffers which await files to be dragged/dropped onto them.
- Underneath the Image container should be a subimage container. These are the nodes you drag and drop into the image buffers.
Note: If the subimage containers do not show up something has to trigger them. Saving the Project might work but often just clicking around in the PWS will as well. This may be a screen refresh issue but I think A:M is just being resourceful in what it is drawing to the screen and only wants to show this container when you need it. Once A:M knows you need it... it will appear.
- Once the images are in the buffer containers I tend to click up at each stage of the hierarchy to refresh the thumbnail image at that level. This clues me into the fact that the settings at that level have properly effected the imagery. It can also clue us in when something is amiss... for instance, if no tiny image shows up at all then a setting may have been missed. Double clicking on the Composite container will of course open the final result for viewing in the main window.
* Here I like to open the Properties panel and Time Line. This will be needed if you are animating effects over time even if dealing with only one still frame it can help to see how the effect is being set. For animation it shows you where your effects have been keyframed.
* If you want to add another image into the composite there are several ways to do it but I like to add a second Mix Effect. This use of the same effect more than once is key to really diving deeply into compositing. It's also useful for effects like Blur which by themselves will only produce a limited blur effect but when nested under additional blur nodes can increase a blur to the nth degree.

Okay... we've played enough for the moment. Time to save our Composite (name the files something useful and number them incrementally... MyFirstComposite001.cmp... for instance). Save the Project file too if you feel the need.

But saving isn't the most important thing. The results of the composite was what we were after and even if we lose everything we need that! So, Right Click on the Composite container and select 'Save As Animation' and save your newly composited image, movie or sequence of images.

Note: If currently using 64bit A:M the .MOV option will not be available. For compositing/converting to .MOV format you'll want to launch 32bit A:M. Run both at the same time if that helps.

Note: When saving to .MOV format the default compression (codec) will be 'Animation'. This is no compression and will produce large files. You may want to choose a compression such as H.264. When you click on 'Compression' in the Save As Animation dialog it may take a little while for the settings to appear.

More later?
NancyGormezano
QUOTE(Rodney @ Oct 19 2011, 10:55 PM) *
I would be interested in hearing how the attached opens for you.


Thanks Rodney for your explanation and sample composite.

I was able to open your composite: Only by dragging it into A:M, as you pointed out (saw no way to import it directly). And If I selected any image other than a .mov type when asked for the file, A:M crashed.

I think right there is another hint to why composite isn't used as frequently as it could be, in that it isn't as robust, nor consistent in behavior as other features. It probably never had a good shaking out in A:M reports.

Similarly, I could follow your steps for starting a new composite (and as well have read Holmes tutorial). The trouble starts when one wants to experiment, change the ordering of mixing, add & delete nodes etc - Then all sorts of funnies start to happen. Many times, one has to start from the beginning.

However, after fiddling, and then getting a stable fixed post processing combo/sequence of effects, one can see that the use of the composite feature can have benefits. It's the finally getting there that can be frustrating to new & old users in trying to use this feature.

Thanks again! It did get me to relook at this feature once again.
robcat2075
I think A:M Composite was initially created so A:M users would have some way to use the new lighting buffers that OpenEXR renders could have.

Other compositing apps were either very expensive or not up to speed on OpenEXR.


If you can make it crash, AMReport it!
Rodney
QUOTE
Similarly, I could follow your steps for starting a new composite (and as well have read Holmes tutorial). The trouble starts when one wants to experiment, change the ordering of mixing, add & delete nodes etc - Then all sorts of funnies start to happen. Many times, one has to start from the beginning.


The A:M Composite we are working with is a first generation implementation so some problems can be expected. I have found however that following an optimized workflow produce highly favorable results. As A:M is not a compositing application like Nuke, After Effects etc. I'm pleasantly surprised when I see it keeping up with these apps. As Robert mentioned, the compositor was primarily designed for use with EXR and should be expected to perform better with that file format. I've mostly been using other file formats in an effort to learn more about how Post Effects work with web-friendly image formats.

Our setup of A:M is likely to have something to do with our results as well. For instance, yesterday I wasn't using OpenMP and so I thought I should enable that. The next time I worked in A:M Composite I experienced some screen redraw issues and a noticeable slow down (odd because the rest of A:M sped up dramatically!). Of course this might not even be related to OpenMP... but I immediately thought it was.

As I find some useful combinations in A:M Composite I'll share them. smile.gif
Rodney
Here's a fun extract from an old Popular Mechanics write up on Hanna Barbera animation.

You can read the whole article via Google Books.

Link found via Jerry Beck over at Cartoon Brew.

At the end of the article you can read the following from 51 years ago:
QUOTE
"Also new this fall is the Flintstones... It's a show intended for grownups," muses Bill Hanna, "but I suppose the kids will be pushing their parents away from in front of the screen so they can watch it too!"


The article discusses some of the cost saving methods Hanna Barbera was using at that time.
Of course all these years later they can still apply to CG too.
Rodney
In another topic we were talking about Pose Sliders and modifying models...

Here's a animated sequence originally shared by Bill Alger through Google+ that demonstrates an equivalent process via Stop Motion.
Just replace the Guy animating Coraline with the Pose Sliders. wink.gif

Either way the 'trick' isn't seen in the end product.

(Click on the image to begin the GIF animation)
Source: http://animationanomaly.com/page/3/
Rodney
I had some uninterupted time to work in A:M today and I'm posting this mostly as a reminder of some of the things I learned/relearned today. Not much to look at but there was a lot of experimenting behind the scenes.

- Modifying characters (I've long wanted to set up a short tutorial/exercise on modifying one of the standard A:M Characters and believe it or not I started modding a cat based on Thom (I was going to call him Thomcat). He started looking a little like Andy Mulia's character from Werewolf and Firefly so I went in the direction of a wolf instead... I had fun thinking about Andy's characters again)

- Rigging (I broke a lot of things in an attempt to explore the idea of a minimal rig for ultra-simple cartoon characters... the goal being to pose/animate a character just by moving Nulls around the screen. I had two minor successes... mostly accidental... and hope to continue pursuing the idea.)

- Animating ( I did a lot of animating while testing/retesting bone constraints... fun but nothing worth sharing here. It was fun seeing Wolfy's Right Arm and Legs move to where I wanted them to be just by moving a Null around the screen). David and Mark would laugh at my ignorance here. Someday I'll rig something!

- Matcap Shaders (It'd been awhile since I played with them... I think the sketchy shader was one of Robert Holmen's) Be careful... that shader stuff can get addicting...

- A:M Composite (the Compositor did not like me very much today. I was obviously trying to do too much too fast and not waiting for the Compositor to process the images... never a good idea)

- Lighting (I Tested out a lot of different things but the use of Rim Lights was something I'd never spent much time on and I like the effect it had on the fishibng pole's string).

Other stuff I can't recall by looking at the images...

The idea of country critters and going fishing was somehow appealing to me. I don't fish... I figure this guy (wolfy) does. I should take it up because even the thought of it is relaxing.

All in all, silly stuff, but it's always fun to play in A:M! smile.gif
Vertexspline
Hey Rodney--glad you had alot of time to play with Animation Master-------Now that's time well spent!. Thomcat ---smiles.
Rodney
The following is the general workflow I was going to present for consideration to Steve over in his WIP on 'Pharmaceutical Man'. My goal was to put together a small video illustrating the process.. I'm working on it. In the meantime here is the general outline for everyone's perusal and commentary. The goal of the workflow is to limit the amount of keyframe adjustments while maintaining a set of clean and easily readable keys in the Timeline/Channel Editor. The fewer the better IMO.

This workflow is something I am trying to lock down myself and I do occasionally discover new angles to the approach. For instance what works really well for one rig may not work at all for another.

A General Outline for Blocking:
- Where possible animate Pose to Pose (The specific action will dictate of course but generally Primary and Secondary Actions are animated Poses to Pose while Overlapping Action and Follow Through are animated Straight Forward)
- Where possible animate on adjacent sequential frames first (by keyframing all Poses on frames 0, 1, 2 and 3 so that A:M doesn't dictate inbetweens. This is important in the early stage of blocking because A:M doesn't yet know enough about what you are doing to just 'go with the flow')
- Spend the majority of time locking down these Key Poses (of all of the Poses one is generally the most important... the Storytelling Pose... the animation will often not read clearly without this Pose and all movements generally flow into and out of this Pose. Remember to design/maintain an especially strong Silhouette for this Pose)
- Stretch all the Keyframes out to their full Timing (Timing is usually dictated by the director or the dialog/audio).
- Sync the Key Poses with the Key Beats and Timing cues. If you've got an Xsheet filled out... use it and refer to it often!
- Snap all of these Keyframes to Frames (Sometimes they get positioned between two frames and this can cause problems)
- Peak all of the Keyframes in all the Channels (This builds some snap into the animation automatically for those you don't adjust/smooth)
- Determine where current Key Poses miss or do not properly Anticipate the Major Beats of the action and/or audio (adjust and then force a keyframe on each of those)
- Consider compressing the Keyframes back down to where there are no Inbetweens and refine these Poses (This places the keyframes back on sequential frames again such as 0,1,2,3,4. Optimally when these keyframes are expanded there are is more than one Keyframe for every four frames of animation)
- Turn on Onion Skinning and Refine the Blocking by looking for Arcs, Exaggeration, Line of Action, Weight, Opposing Forces, Straights versus Curves and Curve Reversals (Note: By focusing on the Extremes you are setting these up for success after the Blocking is over)
- Look for the amount of change in each Pose from one to the next (If there is a keyframe there generally there will be a major change in Pose)
- Expand the keyframes out again and make sure you've saved your progress at this point. You are (essentially) done Blocking at this point.
Step away from the computer and take a break. You've earned it. When you return to the animation the fresh perspective will have done you good.

The Key to Blocking is to keep things rough/raw yet capture the essence of the scene while conveying it's primary emotion. Where possible name the intended action or emotion. This will help you maintain the shot's focus.

Of course all of this is a lot easier to read than execute. wink.gif

I will be most appreciative of any feedback that either confirms this outline or suggests better workflows.
Rodney
With the various posts on modeling vehicles popping up lately I've had vehicles on my mind as well. Of course it's of a slightly more cartoony version (My favorite vehicles are usually the ones seen in the old black and white animated films) but still...

With additional emphasis being placed on Tintin pending release of the Spielberg/Jackson movie this has renewed interest in the world the character lives in. For instance, I found this collection of all the vintage vehicles used in the Tintin comic of particular interest:

http://dardel.info/tintin/tout.html

While these aren't blueprints or schematics it does represent a pretty nice collection of cartoon conveyances.
They try to match the view of the vehicle in the comic strip with a similar view with the real vehicle.

The image below... a sampling of references/drawings used in the comic strip:
Rodney
I recently posted a link to the public domain program ImageJ.
The program has the fortune of being associated with a good many utilities one of which deals with a process known as seam carving.
Many may have seen this feature in action via demonstration where images are magically transformed to alter over all size but maintain the main features within the image. Plugins for Photoshop make experimentation easier.

Looking into ImageJ allowed me to find a link to a java-based seam carving utility that I've noted for possible use in processing my images. The downside of the utilities I've seen thus far is that they can only be used to shrink images. I'll have to look further to see if an utilities are readily available for expanding images (inserting imaginary space between essential elements of an image).

Moving beyond these utilities I can't help but think of how this seam carving will someday work in three dimensions. I'd guess that someone already has a working application somewhere.

Here are a few examples where I've let the seam carver utility go to work.

In the first instance (from Disney's Black Cauldron) I specified that most of the critters left arm (screen right) should be removed from the final image. We do this by painting on top of areas to designate whether they should be kept or removed from the final image. In this case some pretty substantial changes are effected in the final image. Note that the character's expression has actually transformed from 'confused' to 'worried'. This is largely caused the horizontal stretching when both images were added together with the same width and not an effect of seam carving itself.

In the second instance (also I believe from Black Cauldron) a background is altered with instructions (overpainting) designed to maintain key features. Note that the fidelity and accuracy might not be adequate to maintain registration without some very careful calculations. For instance, if a character's feet were to seem to be firmly planted on a ledge in the original that contact might not be preserved in the altered image once it is rescaled. The original is on the left and the seam-carved results on the right.

In the third instance (from Richard William's book) I painted the areas in between the poses green (remove) and painted the character and wording below as red (retain). In this case the results are equivalent to moving each of the character poses and works together as if by hand (very similar to simply removing the empty space). It should be noted that because the entire image is processed there is some degradation of the characters and text even though those areas were specified to be retained. One way or another the program will fit everything into the required dimensions.

For more information about seam carving and to download the java utility used in this exploration visit the authors website:
http://www.ulfdittmer.com/imagej/seam-carving.html
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.
Invision Power Board © 2001-2014 Invision Power Services, Inc.