It Can't Be Done! Feb 2014 Nominations...

[robcat2075]

Is there something you think can't be done in A:M that it ought to be able to do? Is there something you'd like to see figured out and explained?

 

Perhaps A:M does it already but no one has tried it yet!

 

Here is your chance to nominate a topic or task that you would like to see someone figure out how to do with A:M. I will choose one and make a WIP thread out of it in February 2014. The nominator of the chosen topic will receive some fabulous prize, possibly a copy of one of Martin Hash's "Memoirs" books. (Did you know there's three of them?)

 

 

My Rules:

 

-Task will be something reasonably appropriate for a full-featured 3D app like A:M. For example, "Word processing" would not be appropriate even if Blender happens to do that.

 

-Task won't require new programming of A:M. I need to use A:M as it is, as-shipped.

 

-Topic will be something reasonably appropriate for one person to do. "Make a feature length film" is beyond my scope.

 

-Some nominations may be referred to already-existing how-to-do-it information. Those are not winners.

 

-"Finish those damn New-User tutorials" is already on my to-do list. No need to mention that.

 

-I am the sole judge and task-chooser of this contest and sole judge of whether I have accomplished the task and even if i don't figure it out, that's OK.

 

Make your nominations below. I will pick one before February starts.

[fae_alba]

Something I've been batting around in the dusty corners of my brain:

 

Be able to apply a "script" that describes the actions of a character (enter stage left, walk quickly to point "A", turn towards camera, sneak to point "B", etc.

 

Sort of like instructing an actor on how, when, where to hit their marks.

[Rodney]

Don't let Robert off the hook so easily.

That one is too easy.

 

Think bigger!

[Fuchur]

Don't let Robert off the hook so easily.

That one is too easy.

 

Think bigger!

 

- A gas "simulation"-like animation (some of the stuff other software can do with pyrocluster / cloud-generators. There have been quite nice things on the forum already like that... but not moving as far as I know.)

- A light paint-animation (glowing growing spiral-like lines which will fly around. (have a few ideas how to do it... but I am not sure if it really will look well)

- Do something like this:

(this one is not too hard, but it looks cool)

 

See you

*Fuchur*

[fae_alba]

Don't let Robert off the hook so easily.

That one is too easy.

 

Think bigger!

 

I'm not talking about throwing down a path...I'm thinking more along a dope sheet approach.

[robcat2075]

I'll note that the book prize will not be the Kindle edition but the actual hardcopy version of Martin's book (presuming it remains available in hardcopy).

[Rodney]

I'm not talking about throwing down a path...I'm thinking more along a dope sheet approach.

 

It's cool. I'm following you.

Full automation of animation using text and such (or if it can be demonstrated... something even better than that).

[fae_alba]

I'm not talking about throwing down a path...I'm thinking more along a dope sheet approach.

 

It's cool. I'm following you.

Full automation of animation using text and such (or if it can be demonstrated... something even better than that).

 

 

That be the idea.. Straight from script to storyboard to animatic....

[robcat2075]

You may need to describe the intended workflow a bit more. Literally converting screenplay into animation sounds like what animators do already, no?

[robcat2075]

These are all interesting ideas, none-the-less.

 

Keep them coming. Eventually someone will suggest something i can actually do!

[Simon Edmondson]

May I suggest:

A fisherman, in a boat, on rough waters, in the rain. Catches something he can't land but would like to land him ?

 

 

The weather,

The sea, the reflections and colours and transparencies the dynamic (?) forces involved, ...

 

Back to my simplicity.

regards

simon

[pixelplucker]

Don't let Robert off the hook so easily.

That one is too easy.

 

Think bigger!

 

I'm not talking about throwing down a path...I'm thinking more along a dope sheet approach.

 

Now that would be a nice way to animate. If you could plop down a null of some sorts and use it as a goal seak for the character. Goals then have the script applied and a timing.

 

Very similar how many game engines work with AI.

[fae_alba]

Don't let Robert off the hook so easily.

That one is too easy.

 

Think bigger!

 

I'm not talking about throwing down a path...I'm thinking more along a dope sheet approach.

 

Now that would be a nice way to animate. If you could plop down a null of some sorts and use it as a goal seak for the character. Goals then have the script applied and a timing.

 

Very similar how many game engines work with AI.

 

Exactly! If you could, while writing the script, put down timing markers (even just keywords) that can be read in (perhaps via an expression) tie it to a character and you have, at the very least, a beginning point for blocking out the animation.

[robcat2075]

To automatically go from script to animation without more animating... you'd need to already have an enormous library of animation to cover all the possibilities that might be called for in a script.

 

Every possible walk and at every possible speed. Every possible run. Every possible start, every possible stop. Every possible angle of turn, every possible sharpness of the turn. Every possible head turn to every possible direction. Every possible reaching of an arm to touch something in every possible direction and every possible height. Every possible sitting in a chair and every possible getting out of a chair. Holding things... every possible book, every possible tool, every possible food... There's just no end to it.

 

In-game animation works because they have limited you to certain behaviors in certain settings, settings that demand many of the same things happen many times.

 

If you're in a combat game you probably can't make the character stop and do a spontaneous MacGiver thing with some object in the set, unless that was foreseen by the game designers and built that specific possibility into the game.

 

Tell me if I'm not understanding the concept.

[Rodney]

I was deep into developing a solution to the Script to Animation method but it broke down at a critical point.

Here's the basic flow:

 

1. Write script

2. Place script online

2A. Alternatively, write the script while it is online

3. Tiny elves read script

4. Tiny elves animate

5. Loop back to script until all script elements are animated and approved by

6. Done

 

The problem was that I was having a really hard time keeping the script online due to intermittent connectivity.

There was also the problem of the tiny elves mostly be available to work only during off peak holiday cycles.

 

 

Back to the drawing/rendering board.

[Wildsided]

Damn animation elves, always so darn unreliable.

[robcat2075]

An elf i could make.

 

Keep discussing!

[Rodney]

Damn animation elves, always so darn unreliable.

 

Not so much unreliable as unappreciated (and underpaid)

I think it has more to do with lack of fiber optic cable which could take a few years to install.

In the meantime we are looking into renting a few satellites but the auraoaoara boreaelaise... ororo spectaculitis.... you know... the northern lightis thingamajig with the magnetical fielding that does that interference at the most inopportune times.

[Wildsided]

We could sort the aurora problem with a handsome prince or possibly a flock of seagulls...But that would mean running....running so faaaaar awaaaaay

[fae_alba]

Robert, I'm not thinking of building the final animation from the script, only as far as the storyboard or perhaps animatic stage at best. So think along the lines of move to point a, position camera (based on script keyword instructions), render a still, move on to next. So there wouldn't need to be any action cycles to deal with.

 

 

Aaaaand Action!

[Rodney]

, render a still,

 

Hey now... that's a whole new ballgame.

Now you are talking some kind of macro language.

 

Very doable as well but... very different than:

 

Be able to apply a "script" that describes the actions of a character (enter stage left, walk quickly to point "A", turn towards camera, sneak to point "B", etc.

 

Sort of like instructing an actor on how, when, where to hit their marks.

 

I'm just sayin'...

[GAngus]

Well, in reference to… "something you'd like to see figured out and explained"…

 

A nice 'step by step' showing how to combine one's own digital footage on an A:M background, and/or the inverse, of an animated A:M model appearing in ones' own digital (photo?) background.

 

I know this has been presented before, but the examples are few and far between.

[robcat2075]

Here are the nominees so far...

 

- Be able to apply a "script" that describes the actions of a character... fae alba

- a gas simulation Fuchur

- A light paint-animation (glowing growing spiral-like lines which will fly around. Fuchur (can you explain this one a bit more if it isn't the next one?)

- something like this

Fuchur

- A fisherman, in a boat, on rough waters, in the rain. Simon Edmondson

- The weather, The sea, the reflections and colours and transparencies the dynamic (?) forces involved Simon Edmondson

- how to combine one's own digital footage on an A:M background, and/or the inverse GAngus

 

There's still lots of time to make suggestions for things you want to be able to do in A:M but don't think can be done.

[pixelplucker]

I think the script idea could be implemented. You have the core components in AM already such as newton. There would have to be some sort of classification in movement types such as run, walk cycles control movement from point to point, facial expressions in another. For it to work the program would have to go through a simulation pass much like the cloth sim does. Tricky part comes in when you have obstacles such as stairs. Does the user need to put in another goal and another set of target?

In a game the ai has a fixed set of animations it calls back, big reason why you often see feet sticking through walls or feet sliding when one move translates to another.

 

I still like the idea because you could feasably use newton and cloth to simulate the reactions of a character such as running through the woods where branches could knock the arms and shoulders back and the character would correct itself after passing.

 

Putting this all together would definately be a daunting task but really amazing imo.

[robcat2075]

... you could feasably use newton and cloth to simulate the reactions of a character such as running through the woods where branches could knock the arms and shoulders back and the character would correct itself after passing.

 

But that is a "polish" detail whereas fae alba's request is for some sort of blocking/camera placement automation which would not encompass such a detail.

 

 

One limiting factor in this scripting ... where in A:M do you paste the text for it to work on?

[itsjustme]

One limiting factor in this scripting ... where in A:M do you paste the text for it to work on?

 

Dope Sheet?

[Fuchur]

Here are the nominees so far...

 

- Be able to apply a "script" that describes the actions of a character... fae alba

- a gas simulation Fuchur

- A light paint-animation (glowing growing spiral-like lines which will fly around. Fuchur (can you explain this one a bit more if it isn't the next one?)

- something like this

Fuchur

- A fisherman, in a boat, on rough waters, in the rain. Simon Edmondson

- The weather, The sea, the reflections and colours and transparencies the dynamic (?) forces involved Simon Edmondson

- how to combine one's own digital footage on an A:M background, and/or the inverse GAngus

 

There's still lots of time to make suggestions for things you want to be able to do in A:M but don't think can be done.

 

Lightpainting is more or less something like this. In real life you can use a long exposure time in your camera to create these effects. It is the look I am after here, not a still-image:

- http://digital-photography-school.com/wp-c...ht-painting.jpg

-

-

> more or less like the light tails which are following the dragon here at 3:44s (and after that > greatest epic music by the way )

 

See you

*Fuchur*

[Wildsided]

Always nice to see Bahamut blowing stuff up

[Rodney]

Lightpainting is more or less something like this. In real life you can use a long exposure time in your camera to create these effects. It is the look I am after here, not a still-image

 

I was experimenting with something like this in A:M about two years ago but I have no earthly clue were those project files were or even how to set it up in A:M at the moment.

Reason No. 5673 why I should follow my own advice and post my projects to the A:M Forum.

Then at least I could find them and pick up where I left off.

 

I believe in order to assist Robert in his effort we may need to pick very specific targets and point to it (otherwise we get what we get... well, we'll get that anyway but at least then it'll be more specific)

The problem with "Light Painting" is that not all FX we see is 'light painted'... much of it is compositing... and we get a similar effect as if it were lighting.

A classic example of this is the trails behind a fireball. Those are usually created by manipulating a very basic image that can be created by a variety of means (particles, drawing, still images animated/scaled/transformed over time). This effect is then blended/composited with the back plate to complete the desired effect.

 

I guess what I'm saying here is that the effect isn't created by a light in a 3D scene and therefore to me cannot strictly classify as 'light painting'.

'Light Painting' to me is actually using lights to paint a scene, environment, surface or effect in (more or less) real time.

As such light painting is quite straightforward although just as in the real world there is smoke, mirror and slight of hand that produce the effect in a 3D scene.

 

The image here... (http://digital-photography-school.com/wp-content/uploads/2009/11/light-painting.jpg) clues us into how one particular method of Light Painting works (as Fuchur mentioned... long exposure times).

For one of many ways forward consider Marcos (Xtaz) Rezende's MUFOOF technique.

 

This works the same as in the real world where an illuminated object moves quickly under the right conditions to have it appear as if it is almost a solid object... and in some cases fool the eye into believing it is actually a solid object. The light displays on the sides of buildings that has been all the rage in Europe over the past few years is a prime example of the latter.

 

I'd be glad to suggest some tips in this direction but it is possible Robert may want to select this for his "It can't be done!" project in February so I'll defer for the present.

I will note that a lot of blur and bloom gets thrown into the mix on some of the Final Fantasy shots I've seen.

Those additions are (relatively easily) accomplished via Post Effects in A:M (esp. v18 as A:M Composite appears to be working even better for me).

When painting with light one must consider several factors the minimum of which is: the lights themselves, the surfaces the light is hitting (even if those surfaces are not visible) and the environment in between.

[Simon Edmondson]

...

Lightpainting is more or less something like this. In real life you can use a long exposure time in your camera to create these effects. It is the look I am after here, not a still-image:

- http://digital-photography-school.com/wp-c...ht-painting.jpg

-

-

> more or less like the light tails which are following the dragon here at 3:44s (and after that > greatest epic music by the way )

 

See you

*Fuchur*

 

Gerald

The images remind of the light paintings Picasso did with the photographer Roland Penrose.

This is just a guess but, could you use a small intense volumetric light, rendered with Motion blur against an alpha background, then composite that against the render of the model in action. A variation of that might work for the character moving by changing the extent of the blur and varying the light on that render, composited with the original ?

Pardon me if this is inappropriate.

regards

simon

[Fuchur]

Lightpainting is more or less something like this. In real life you can use a long exposure time in your camera to create these effects. It is the look I am after here, not a still-image

 

I was experimenting with something like this in A:M about two years ago but I have no earthly clue were those project files were or even how to set it up in A:M at the moment.

Reason No. 5673 why I should follow my own advice and post my projects to the A:M Forum.

Then at least I could find them and pick up where I left off.

 

I believe in order to assist Robert in his effort we may need to pick very specific targets and point to it (otherwise we get what we get... well, we'll get that anyway but at least then it'll be more specific)

The problem with "Light Painting" is that not all FX we see is 'light painted'... much of it is compositing... and we get a similar effect as if it were lighting.

A classic example of this is the trails behind a fireball. Those are usually created by manipulating a very basic image that can be created by a variety of means (particles, drawing, still images animated/scaled/transformed over time). This effect is then blended/composited with the back plate to complete the desired effect.

 

I guess what I'm saying here is that the effect isn't created by a light in a 3D scene and therefore to me cannot strictly classify as 'light painting'.

'Light Painting' to me is actually using lights to paint a scene, environment, surface or effect in (more or less) real time.

As such light painting is quite straightforward although just as in the real world there is smoke, mirror and slight of hand that produce the effect in a 3D scene.

 

The image here... (http://digital-photography-school.com/wp-content/uploads/2009/11/light-painting.jpg) clues us into how one particular method of Light Painting works (as Fuchur mentioned... long exposure times).

For one of many ways forward consider Marcos (Xtaz) Rezende's MUFOOF technique.

 

This works the same as in the real world where an illuminated object moves quickly under the right conditions to have it appear as if it is almost a solid object... and in some cases fool the eye into believing it is actually a solid object. The light displays on the sides of buildings that has been all the rage in Europe over the past few years is a prime example of the latter.

 

I'd be glad to suggest some tips in this direction but it is possible Robert may want to select this for his "It can't be done!" project in February so I'll defer for the present.

I will note that a lot of blur and bloom gets thrown into the mix on some of the Final Fantasy shots I've seen.

Those additions are (relatively easily) accomplished via Post Effects in A:M (esp. v18 as A:M Composite appears to be working even better for me).

When painting with light one must consider several factors the minimum of which is: the lights themselves, the surfaces the light is hitting (even if those surfaces are not visible) and the environment in between.

 

I am very sure this is not a light at all... do not confuse this: I am not after creating such a thing with a light. It is just about how the look can be created in 3d / A:M. No matter how.

I have several ideas how to do it (for instance with a snake-like glowing-geometry with bones in it attached to a path but that might be quite much work especially for many of them) or with a particle-emitter (but that one is tough, because the particles will very likely never glow like that and look as "full" as the tails in the video at least not if we are talking about sprites in A:M here.

 

I have several ideas how this could be archieved, I am just giving something that is hard to archieve here for Robert... I could give a unsolveable task, but where would be the fun in that?

 

See you

*Fuchur*

[Rodney]

I could give a unsolveable task, but where would be the fun in that?

 

I believe that is the type of task (the unsolvable) that Robert is seeking.

[Fuchur]

I could give a unsolveable task, but where would be the fun in that?

 

I believe that is the type of task (the unsolvable) that Robert is seeking.

 

The moving gas simulation should be hard enough and close to unsolvable... something equal would be water with ink disolving in it...

I do not really have a good idea on how to do it without using live-footage as a texture (which would be cheating at least if the footage would be a video-strip of ink disolving in water )...

 

See you

*Fuchur*

[fae_alba]

, render a still,

 

Hey now... that's a whole new ballgame.

Now you are talking some kind of macro language.

 

Very doable as well but... very different than:

 

Be able to apply a "script" that describes the actions of a character (enter stage left, walk quickly to point "A", turn towards camera, sneak to point "B", etc.

 

Sort of like instructing an actor on how, when, where to hit their marks.

 

I'm just sayin'...

 

I downgraded my expectations 'casue Robert said he couldn't do it (throws gantlet down in a theatric way).

[Rodney]

I downgraded my expectations 'casue Robert said he couldn't do it (throws gantlet down in a theatric way). rolleyes.gif

 

One of the obstacles of rolling out a scripted animation method is to know which path out of several to investigate.

It could be that more than one or a combination of several would work best.

 

Here's the basic approaches to scripting:

 

Plan 1 (Using what works already)

This is the Library of actions approach and the most likely scenario to get a working script editor

Advantages: Drag and drop scripting workflow. Unless desired, users would not have to type any of the script in. They would select from available options instead.

This is likely to be the most successful because it been demonstrated to work already... and works quite well. It does need to be fine tuned, optimized, documented, etc. for use with the majority of resources A:M Users will create.

 

This is a powerful approach and can produce rough animation very quickly and adheres to Martin Hash's philosophy, namely; "Reusability is the foundation of A:M."

Downside: Script elements need to be created/defined/declared before they can be used by a script. This is however easily offset by sharing modular elements of the script or restricting script function to lowest level activity (to account for differences in rigs).

Challenge Example: If the script were to say "Do the hokey pokey" the term 'Do' is understood as a request to execute an action.

The obstacle is the action and subsequent variables. Until created, the script has no way of knowing what 'the hokey pokey' nor the criteria for determining success.

 

Plan 2 (Enhance what works already)

This is the dopesheet approach which is the likely approach to script animation within a Model

That is to say that if the resource cannot be pose driven then this approach would be ideal.

Note: Plan 1 works externally as well as internally and as such can use most elements of Plan 2

 

The Wizard approach (Extending currently available tools)

This involves a longer term goal of mine which would begin with a minor update of the Font and/or AI wizards.

Phase 1 would be to create the initial scripting pathway (for font generation from an external textfile)

Phase 2 would be to create variables that all positioning and orientation

Phase 3 would be to extend the wizard beyond fonts to Models, etc.

The issue with this approach (other than it requires a considerable amount of programming) is that it might be better just to create everything in a language like Python, Qt or such in the first place.

 

Other options not considered here (such as using other programs in conjunction with A:M).

 

So Plan 1 and 2 are currently operational but would need to be refined.

The Wizard and Other options would most likely need to be programmed.

[robcat2075]

To me, it sounds a bit like the animation version of "make dragon"

[Roger]

How about "force a character through an arbitrary opening", say, a keyhole? Perhaps Thom is on one side of a door and a black hole is on the other side, sucking him through the keyhole (in reality the black hole would rip apart the door and everything else, but this is a special case where its gravity only affects the particles in Thom's body).

[pixelplucker]

... you could feasably use newton and cloth to simulate the reactions of a character such as running through the woods where branches could knock the arms and shoulders back and the character would correct itself after passing.

 

But that is a "polish" detail whereas fae alba's request is for some sort of blocking/camera placement automation which would not encompass such a detail.

 

 

One limiting factor in this scripting ... where in A:M do you paste the text for it to work on?

 

 

Instead of pasting text or trying to write a script, set tags and call them up in a drop down list. Similar to how the extrude tool sees named groups. The script would be essentially calling up a pose and/or action right?

Would something like that be feasable?

[Rodney]

set tags and call them up in a drop down list.

 

I believe Robert's point is that there is no place to do that without programming a new plugin/feature.

 

When going into the programming world 'it' (i.e. anything) can be done with the appropriate amount of time and commitment.

So, I'd say no to the feasability thing.

Coding new functionality for tags and dropdowns is not feasible.

Now if you know how to do this with A:M as it is... say on!

 

Edit: Note that technically this can be done via dopesheets but... you've got to set the tags (via creating a pose) and then enter those tags into the dopesheet (via the dropdown menu in the dopesheet).

So in that sense not only is it feasible, it can be done and just needs to be demonstrated.

[robcat2075]

The script would be essentially calling up a pose and/or action right?

 

And where does that pose or action come from?

[pixelplucker]

It would definately have to be a plug-in or added since it isn't there now.

[fae_alba]

Again, in my mind my script idea was not to "replace" the animation stage of any project. There is no way you can automate the art of animation (like how I worked TOA:M in there?) . My thoughts have always stopped at the animatic stage....And in reading the conversation so far (kind of hit or miss for me right now since I'm in NYC taking some training for work, in a winter storm no less!), I kind of agree that this would take 1) some serious off line discussion on what this might look like and 2) some external coding to accomplish. The dope sheet approach would be one way to do it today, as a POC though. Assuming that the models (or main characters) of a project are being developed at the same time as the script is being finalized (they should be) and at least the rig is in place by the time the final script is signed off on. Basic poses can then be created (perhaps use a naming convention of "animatic_nnnnnnn") and as the script writer can then (with a list of the poses) key them into the script. For each scene in the script, add tags to the scene header say like to call up the pose. Add in camera movements of and for timing.

[largento]

I'm coming in late on this, but going back to Robert's original request...

 

"It can't be done!"™ — Getting A:M to render a frame that has the high quality look of a Disney/Pixar production.

 

 

 

I'm not talking about matching the quality of the models, but achieving the high-end look of textures, colors and light.

 

Can A:M make a render that looks more like Toy Story 3 than Toy Story 1?

 

[robcat2075]

Can A:M make a render that looks more like Toy Story 3 than Toy Story 1?

 

I'm not doubtful that it can be done. If someone were to provide the models for such a scene that might make for a good WIP thread.

 

I'll note some things I see working in this TS shot

 

-lots of AO. Note the unnatural amount of AO surrounding the door knob in the upper left corner. Maybe it's supposed to be dirt?

-soft focus. Nothing is pixel sharp in this image. Woody is "in focus" but you still can barely read "SHERIFF" on his badge.

-DOF. The background is even a bit more out of focus.

-Bloom. On real film that bright window would bleed into the surrounding image. They are simulating that effect of course.

 

 

It's also entirely possible that this image didn't come straight out of the PIXAR render farm. The characters and background may have been rendered separately for ease of lighting and combined in post.

 

 

I can see getting all of that done in A:M.

 

 

 

[largento]

I, too, feel almost certain that there's compositing going on here, but I've never truly grasped the concept.

 

There's likely some post color timing that's done, too.

 

Maybe we could use some of the Oz models?

[robcat2075]

To convince people I think we'd to have the same models in the scene presenting the same textures and shapes that need to be lit.

 

This particular one appeals to me and I think would interest many people. Perhaps we can keep it in mind for a future installment of "It Can't be Done" if the models are not immediately obtainable.

[largento]

Okay, this is kind of a "wow" for me.

 

I went looking for info and remembered that I'd read that the theatrical release of Toy Story 2 had been on the old Technicolor Dye process. I wondered if there was a way to recreate the Technicolor 3 strip process using Photoshop. After some complicate tutorials and confusion, I discovered that starting with Photoshop CS 6, there's a simple layer adjustment called "color lookup" that lets you instantly add the 3 strip (or 2 strip, or one of many other color looks.)

 

Look what happens when I do that to the Toy Story 1 image (along with a little blur/soft lighten layering)

 

 

Granted, this is a low resolution image with lossy compression, but look at how instantly more vibrant it looks!

[robcat2075]

If I raise the color saturation on the top image I can get a very similar result. Saturated color was sort of what Technicolor was about.

[largento]

Admittedly, but it's clear that's part of what I'm seeing in the look I'm wanting. I just did it with some old renders of mine and it made me very happy. :-)

 

It seems to kill that pervasive "gray" feeling I always sense in renders and why I almost always add some kind of ambient color to my renders.

[robcat2075]

If it works it works, and there's no such thing as "cheating" in CG when it's all artificial anyway.