WIP: Apache Helicopter Attack Sequence - A PySwarm V0.4 Study
FractalDimensia
Posts: 0
Purpose
After spending a lot of time running rather simple test simulations on PySwarm, I decided to try my hand at creating a more realistic and complex animation sequence. The purpose of this thread is to describe the steps I am going through to create the animation sequence, and to illustrate how I am using PySwarm V0.4 to:
• Work with large scale scenes
• Generate realistic, scaled BOID motion - with specific focus on speed, banking, and grouping
• Guide a group of flying objects (BOIDs) through a terrain using a target helper set with a motion path
• Manipulate several groups of BOIDs within a single simulation sequence
• Replace basic BOID models with more detailed models
• Link NLA tracks and other objects (e.g., cameras, particle emitters) to BOIDs
It is possible the PySwarm script may need be to modified to handle some of the simulation requirements necessary to produce this animation. A new version of PySwarm will be released at the conclusion of this thread with those new features.
Taking a page out of Steve's (SciFiFunk) book (just not to such a grand scale :) ), I will post regular updates to this thread as I develop this sequence - describing the steps I am taking, any problems I encounter and any work-arounds for them, links to results (video sequences), and current/upcoming work. I will also make my .CAR files available for download at certain points in this process.
Since I am rather new to Carrara animation, another objective is for me to learn more about how to create animation sequences. I hope you find this thread interesting and useful.
Of course, all thoughts, comments, and suggestions to improve this study or the animation results are always greatly appreciated!
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Table of Contents and Thread Links
Animation Sequence Posted 11/11/2013
http://www.daz3d.com/forums/viewreply/477913/
Storyline
Setup
Sequence Duration
Cameras
Flight Path
Animation Specifications Posted 11/11/2013
http://www.daz3d.com/forums/viewreply/477919/
Terrain Specifications
Helicopter Specifications
Development Strategy and Current Status Posted 11/11/2013
http://www.daz3d.com/forums/viewreply/477921/
Step 1 - Basic Scene Construction Posted 11/11/2013
http://www.daz3d.com/forums/viewreply/477931/
Landscape and Water
Apache Helicopter BOIDs (simple model)
Helicopter Flight Path
Step 1 - Basic Scene Construction Posted 11/11/2013
(continued)
http://www.daz3d.com/forums/viewreply/477941/
Attractor Speed and Animation Timeframe
Step 2 - Helicopter Fly-In (Animation Stage 1): Test Run 1 Posted 11/13/2013
http://www.daz3d.com/forums/discussion/32239/#479349
What seems to be working well
Not added or tested yet
Issues with this test (fixes needed)
Current work
Step 2 - Helicopter Fly-In (Animation Stage 1): Test Run 2 Posted 11/16/2013
http://www.daz3d.com/forums/viewreply/481768/
What seems to be working well
Not added or tested yet
Issues with this test (fixes needed)
Current work
Step 2 - Helicopter Fly-In (Animation Stage 1): Test Run 2 Posted 11/17/2013
(continued)
http://www.daz3d.com/forums/viewreply/482513/
Link to video demo
Current work
Step 3 - Helicopters Fire Missiles (Animation Stage 2) Posted 11/18/2013
http://www.daz3d.com/forums/viewreply/483571/
Concept Description for the Animation
Modified LANDING Rule
Applying PySwarm to the Missiles
Step 3 - Helicopters Fire Missiles (Animation Stage 2) Posted 11/21/2013
(continued)
http://www.daz3d.com/forums/viewreply/486472/
Missile Specifications
Test 1 Setup
Test 1 Results
Problems Encountered
Next Step
Step 3 - Helicopters Fire Missiles (Animation Stage 2) Posted 11/26/2013
(continued)
http://www.daz3d.com/forums/viewreply/489872/
Test 2 Setup
Test 2 Results
Problems Encountered
Next Step
Step 3 - Helicopters Fire Missiles (Animation Stage 2) Posted 11/26/2013
(continued)
http://www.daz3d.com/forums/viewreply/489898/
Test 3 Setup
Test 3 Results
Problems Encountered
Next Step
Animation Sequence
Storyline
Five Apache helicopters fly through a mountainous jungle to take out an enemy stronghold. The helicopters fly in fast and low, using the jungle river as cover against enemy radar and thick mist against visual sighting. The grouping of helicopters must remain tight during the sequence. When the helicopters reach their target, they hover momentarily to acquire their targets, and launch a number of missiles that strike the enemy fortification. The helicopters then turn and fly out of the region, continuing along the jungle river.
I've attached a photo of a real Apache helicopter to this post to give an idea of what I want to ultimately produce.
Setup
The final sequence will be shot in 3 stages:
Stage 1: Helicopters flying in towards target
Stage 2: Helicopters hover and fire missiles; missiles strike their targets
Stage 3: Helicopters depart
Sequence Duration
I am planning for the final animation sequence to be approximately 5 minutes long.
Cameras
I will use the following conical cameras during the sequence:
1. High altitude camera
2. Front camera - moving in front of the helicopters, with its path set by PySwarm
3. Back camera - flying along the same path as the attractor behind the helicopters
4. Stationary camera - repositioned along the path to show "fly by" animations
5. Stronghold camera - positioned within the stronghold
Flight Path
The helicopters' flight path will be assisted with a Target Helper object as an attractor. The helicopter speed will be set to be approximately that of the attractor.
Animation Specifications
To create as realistic a scene as possible, accurate dimensions of the terrain and helicopters is important.
Terrain Specifications
• Dimensions: 100K feet x 100K feet (18.9 miles x 18.9 miles)
• Type: Mountainous with thick jungle vegetation
Helicopter Specifications
The following helicopter specifications came from WikiPedia.
• Model: AH-64A
• Dimensions: Length - 58 feet; Height - 12.7 feet; Rotor Diameter: 48 feet
• Crew: 2 (tandem)
• Max Speed: Approx. 150 knots max (172 miles/hour or 250 feet/sec)
• Speed in Sequence: Approximately 100 feet/sec (68 mph)
Distance Travelled by Helicopters: 150K feet (28.4 miles); total time: 28.4 miles / 68 mph = 25 minutes
Development Strategy and Current Status
Step 1 - Basic Scene Construction
Before digging in to creating detailed models of the scene (e.g., helicopter, landscape, water, enemy stronghold), a few concept studies are needed to verify PySwarm will produce the desired results and to adjust the various PySwarm parameters to produce the desired results. After creating basic models of scene components, I will conduct two concept studies.
I completed this step on Nov 10, 2013.
Step 2 - First Concept Study
The first concept study will look at stage 1 of the sequence (the helicopter fly-in). This step will focus on setting camera locations, positioning and adjusting PySwarm parameters, and adjusting the attractor's path.
I completed this step on Nov 17, 2013.
Note: There are still a few modifications needed for stage 1, but I will resolve these later in Step 5. These modifications are:
1) Add a banking delay. This will require a modification to the PySwarm script.
2) Adjust the flight altitude in the 30 seconds the helicopters are flying over land. This will be handled by running the PySwarm script on the final sequence in phases. (I'll post more on this later.)
Step 3 - Second Concept Study
The second concept study will look at the sequence's stage 2 - launching a number of missiles from multiple moving platforms (the five helicopters). This step will focus on creating realistic missile launch patterns.
Currently under development.
Step 4 - Update Helicopter Model
Once the PySwarm script and parameter settings are producing good results, the next step will be to:
• Replace the base helicopter model with the one to be used in the final rendering
• Add modifiers to the helicopters for rotor motion
• Retest PySwarm against the modifications
Step 5 - Final Rendering
In the final step, I will:
• Modify scene settings (lighting, mist, clouds)
• Add vegetation to the landscape
• Modify the river water
• Add the enemy stronghold
• Add particle emitters to the missiles
• Add explosion sequences
Then render the final sequence.
Step 1 - Basic Scene Construction
The purpose of this first step is to construct the basic object models for the animation sequence. Object details will be added at a later time.
The following objects/components were constructed and added to the basic scene to be able to perform the first concept study (next step):
• Landscape (no details)
• Water (no details)
• Helicopters (simplified base model)
• Helicopter Flight Path
• Several cameras
Landscape and Water
I created a basic mountainous landscape (only the main one) and waterways. Only the main landscape is identified. Also, identify the location of the enemy stronghold. (See the attached screenshot - Figure 1.)
• Water as a plane at z=0
• Landscape of size 100K feet x 100K feet (18.9 miles x 18.9 miles)
• Set height of landscape to create a river basin
Apache Helicopter BOIDs (simple model)
I selected the Carrara-provided Apache helicopter model for shooting the animation sequence, with the intent to make some modifications later before the final sequence is shot. This model was selected to make the resulting .CAR file publically available. I adjusted the dimensions of the Carrara Apache helicopter to match the specifications described above. (See the attached screenshot - Figure 2.)
For the concept studies (next two steps), I constructed a simplified (base) model to match the size and shape of the final model. (See the attached screenshot - Figure 3.)
Note: PySwarm makes it very easy to swap out models for a simulation. (For example, one of the air vehicles from the DAZ store would make for a great sequence!) You just have to follow the basic naming conventions required by PySwarm.
Helicopter Flight Path
I added a Target Helper to the scene as a PySwarm attractor to guide the helicopters along a specified attack path. I then set the attractor to Motion Path animation, and laid out the flight path for the helicopters to follow. (See the attached screenshot - Figure 4.)
Step 1 - Basic Scene Construction
(continued)
There was a need to do some basic calculations before the PySwarm script parameters could be set for the first concept study. Here are some details related to these calculations.
Attractor Speed and Animation Timeframe
Given: From the helicopter specifications above, I know I want the BOIDs to fly at approximately 100 feet/ second. Since I am using a Target Helper to guide the BOIDs to their destination, it's motion must also be at that rate. Too slow, and the BOIDs will fly into and around the attractor; too fast, and it will leave the BOIDs behind.
To set the motion path for the attractor, I followed these steps:
a. Set the Stage 1 simulation to be 4 minutes long (purely a guess!)
b. Moved the animation scrubber to 4 minutes
c. Adjusted the target helper's motion path to where the helicopters need to be when they reach their target
Now, I need to computer how fast the attractor is moving along the motion path
d. Measured the position of the attractor at the start of the simulation
e. Measured the position of the attractor at 1.0 minute into the simulation (an approximate straight-line path to this point)
f. Compute the distance traveled using the X,Y coordinates at both positions. Result was approximately 17,000 feet.
g. Compute velocity of the attractor - 17,000 / 60 seconds = 290 feet/sec.
This is nearly 3 times as fast as I want it to move, so I stretched the time for the path to be 3 times as long, or 12 minutes.
I then repeated steps c. - g. to get the following:
• New distance - 6,441 feet
• New velocity of attractor - 6,441 feet / 60 seconds = 107 feet/sec
With this calculation, I will set the Apache helicopter BOIDs' max speed to 107 feet/sec to track behind the attractor.
Sounds very cinematic! Hint! Hint!
http://www.daz3d.com/forums/discussion/31940/
;-)
You hint is kind of funny for me.
I was thinking how to work on PySwarm and try to submit something for this month challenge and I came along a similar idea. Mine was something from Star War, probably the battle on the snow planet in the 'strike back' episode (mainly because the snow speeders are not that difficult to model using the spline editor :-) ).
Hi FD,
You're tackling a nice project and one Carrara is well suited to do nicely.
I'm impatient to see what you'll come up with.
You hint is kind of funny for me.
I was thinking how to work on PySwarm and try to submit something for this month challenge and I came along a similar idea. Mine was something from Star War, probably the battle on the snow planet in the 'strike back' episode (mainly because the snow speeders are not that difficult to model using the spline editor :-) ).
That would be a phenomenal render! I'd like to see what you come up with. :)
If I can find a way to work in a single image render into the competition, I'm there! Been thinking about it already, EP.
This is going to be so interesting to follow... :)
I hope it remains interesting!
Step 2 - Helicopter Fly-In (Animation Stage 1): Test Run 1
I have completed the first test run for this simulation sequence. The primarly objectives of this test run were:
1) To test PySwarm’s ability to create reasonably realistic animation for the helicopters stage 1
2) To test various camera locations and types
3) To test animation motion and speed
You can watch the resulting 5 minute video (low-res render) here:
http://www.youtube.com/watch?v=f_ygjF98UnQ
What seems to be working well
1) Several (3) PySwarm V0.4 defects were identified with this first test run (thus, the delay getting the test run complete). The script was patched to resolve these defects. A new version of PySwarm (V0.4.1) will be released in the next day with these fixes. The PySwarm thread will be updated with the release of the new script.
2) Front, back, and stationary cameras look good, and provide good angles for the stage 1 video sequence.
Not added or tested yet
1) 2D Limit – Helicopters are currently set to fly at the same altitude. Test run 2 will provide a limited altitude range for helicopters to fly within (under 100 feet).
2) High-altitude camera will be added later.
Issues with this test (fixes needed)
1) Banking Delay – PySwarm V0.4 works by adjusting the banking (roll) based on the current heading adjustment/change. In real life, with many “BOIDs” such as helicopters, there is a delay between the banking and the heading change. This first test run reflects the need to modify the PySwarm script to add this delay (as a parameter).
2) Speed Check – The helicopters do not look like they are flying at a speed of about 100 feet/sec (60 miles/hour). They look to be flying at ½ to 1/3 that.
3) Altitude – I want to drop the flying altitude from about 230 feet (currently) to about 100 feet.
Current work
The focus of the next test run will be on:
1) Making the banking action more realistic by modifying the PySwarm script to include a delay between banking and heading change.
2) Adding slight altitude variations in the flight of the helicopters
3) Checking speed again to verify it is working correctly.
All feedback (thoughts, suggestions, and ideas) are greatly appreciated!
Thanks,
FD
Step 2 - Helicopter Fly-In (Animation Stage 1): Test Run 2 Problem
SInce it has been a few days since I completed the first test run, I thought I would post a brief update on the second test.
Very briefly, I encountered a problem, and I still have not figured out the source. I turned off "limit_2D" to get some vertical motion in the helicopters, but for some reason, they are refusing to cooperate. PySwarm seems to be correctly working, in that it is varying the altitude, and it seems to be sending the right altitude settings when it tells PyCarrara to create the keyframes, but the altitude (z) value is somehow getting ignored or overridden.
An interesting (likely related) observation I made is that I cannot seem to be able to reposition the helicopters' altitude (z transform value) in the scene manually using the Move select option, selecting a helicopter, and attempting to drag it to another altitude. (More peculiar, I can change the x and y transform values!) The helicopter snaps back to its original position. (See attached screenshot.) Even more interesting, I can manually type in a new position in the Motion/Transform for a z value and the helicopter will move. But if I then attempt to move it manually (by dragging it), the helicopter again snaps back to it's ORIGINAL z value, not the one I typed in.
I am hoping to discover the source of this problem this weekend when I have time to give it due attention.
If anyone reading this has an idea what might be cause of this, I would appreciate your thoughts! Thanks!
It's weird
A constraint would prevent from moving in the Z-axis, but it would be the same for both typing value and moving the arrow. It would also prevent the move, not generate a 'snap back'
Can you share the scene file ?
I'll post the .CAR file for download tonight. It would be great if you or others could take a look at it!
Thanks!
I figured there was a 95% chance that the problem was related to something really dumb I had done. Finally was able to think through the problem and discovered the source. It seems I had attached a Track Modifier to each of the helicopters that was forcing them at a certain z value. I have no clue how that happened. :red:
So I am hoping to post something this weekend on the progress I've made.
It's nice you found out. Spontaneous generation of track modifier, althrough well known to every Carrara user, is a mystery I've never been able to untangle :-)
I'm looking forward to see your next video.
:-)
Step 2 - Helicopter Fly-In (Animation Stage 1): Test Run 2
I have now completed the second test run for this animation stage (1). The primarly objectives of this test run were:
1) Add altitude adjustments and lower altitude of helicopters from about 230 feet to about 150 feet
2) Camera adjustments
• Increase separation between tracking camera (front camera) and helicopters
• Add a high-altitude camera with a motion path
• Reposition stationary camera along the path at four locations on the river bank
3) Improve helicopter separation
4) Tweak helicopter roll, bank, and pitch
5) Check speed and size dimensions of helicopters
6) Add enemy fortification (rough model)
The detailed parameter settings for the first two test runs are shown in the attached screenshot.
I am conducting final low-res renders of the revised sequence and will post a link to the video when it is completed (tomorrow sometime).
What seems to be working well
1) Several more adjustments were made to PySwarm V0.4. As with the first test run, the script was patched to make these adjustments. I have held off releasing a new version of PySwarm (V0.4.1) until these additional adjustments can be fully tested. The PySwarm thread will be updated with the release of the new script.
2) Front, back, and stationary camera angles are looking better.
3) Altitude adjustments are working.
4) Helicopter separation is better.
Not added or tested yet
1) Final helicopter positioning at the end of stage 1 - I will probably manually move the helicopters to get them into positions for the stage 2 (launching missiles).
Issues with this test (fixes needed)
1) Banking Delay (still no solution identified yet) – PySwarm V0.4 works by adjusting the banking (roll) based on the current heading adjustment/change. In real life, with many “BOIDs” such as helicopters, there is a delay between the banking and the heading change. This first test run reflects the need to modify the PySwarm script to add this delay (as a parameter).
Current work
The focus of the final (3rd) test run will be on:
1) Make final camera adjustments – positions, zoom, etc. (note - DOF will be set in preparation for the final render)
2) Test final helicopter positioning at the end of stage 1
3) Adjust helicopter altitude during land fly-over section of flight path (from time 3:30-4:05) - There is a section of stage 1 when the helicopters fly over (rather, at the moment, fly through) a land area. Will adjust the altitude flight path during this timeframe.
Again, all feedback (thoughts, suggestions, and ideas) are greatly appreciated!
Thanks,
FD
While waiting for the final test renders to complete, I did some upgrades to the Apache helicopter I plan to use in the final rendering. More work to do on it, but it's looking better.
I also did some basic calculations to discover it will take approximately 2 calendar months to render the final animation on a single computer. >:(
Somehow I doubt Octane Render will be ready in time to come to the rescue. ;-) It's time to start thinking through my backup options to get this rendered.
Step 2 - Helicopter Fly-In (Animation Stage 1): Test Run 2
(continued)
I have uploaded a new animation sequence for the second test run for the Apache Helicopter Scene's Stage 1 - Helicopter fly-in.
https://www.youtube.com/watch?v=cT3ycNY55-w
This spliced sequence is for test purposes only - camera locations, helicopter behavior, etc. But the sequence is close to how I envision the final shooting to look like. However, only about 3 minutes of time will be devoted to this stage in the final render.
This test now shows the fly-in from four different camera angles - front, stationary, back, and (a new) high altitude for about 6 minutes of the 12 minute fly-in path. The scene ends with the helicopters arriving with their target - an enemy outpost - in front of them, setting up for stage 2 - launching missiles. (I'll post more details on this stage soon.)
For more details on this test, please read the post above:
http://www.daz3d.com/forums/viewreply/481768/
Current work
At this point, I would say the resulting .CAR file is successful in doing what I had hoped it would, though a few more adjustments (noted in the post above) are still needed. All helicopter motion (position and rotation) is generated by the PySwarm script. The main issue I have left to work on before the final rendering of stage 1 is to add a banking delay between when the helicopters bank for a turn and when they respond. Because of the scripting change required, I have decided to delay working on this issue until after I have redesigned the Python script in preparation for V0.5.
Again, all feedback (thoughts, suggestions, and ideas) are greatly appreciated!
Thanks,
FD
So, what are your render settings? There are things I do when rendering a still image that I don't do when rendering an animation.
One thing I do that helps with a complex animation is to break it down into layers for compositing. I set my full scene up, check the lighting, position the cameras and then use Save As to save as many versions of the scene as layers of video I want. So, if I want three layers, the background layer will be saved as my-scene01, the middle layer will be saved as my-scene02, and the foreground as my-scene03. I then open each scene and either hide elements or delete elements not needed in that layer. If there is an element in the middle layer that receives a shadow from an element in the foreground layer, I'll put a shadowcatcher shader on the mid-ground element and leave it in the foreground scene.
That's just a quick example of my workflow. There are other ways to help speed things up, but unless I know the render settings or what effects you're using it would be super generalized.
A general comment, and not related specifically to PySwarm, but if you analyse most movies, each shot is usually only a few seconds long (if that!), with cuts to different angles etc. I think aiming for 5 mins for your video sequence is too long, you maybe need to plan it over perhaps 2 mins, which of course would shave 60% off your render time too!
Thanks, EP.
Animation is a new topic for me - Carrara or otherwise. I remember reading about your idea of compositing in other threads, and I already have begun looking into how that might speed up my render time. I should have plenty of time while running test renders to experiment, and since the primary focus of this project is to test PySwarm's capabilities, I don't want to lose sight of the objective.
Still, the project provides a lot of opportunity for learning, so I hope to capitalize on that as well! I'll be looking for more specific ideas and suggestions soon, I'm sure. :)
Good point, Phil. And thank you for the reminder! Once I have a reasonable amount of film footage (in test render mode), I will decide which parts to use for the full and final render. I would rather plan for a longer sequence and cut back than to plan it short and have to add more later.
Also, I would agree with the "few seconds" rule if it is for the "action movie" genre. TBH, I'm not sure "action" is the genre I'm going for. I was (mentally) playing around with the idea of ...
WARNING! POSSIBLE SPOILER ALERT!
a more subtle video/message. Something like - putting the fly-in animation to slow, melodic tune with majestic jungle shots, river pontoons and canoes, rice patties, villages on the banks, etc. that gently put the viewer into a state of serenity. And then at the last moment shocking the hell out of their serenity with an unexpected and violent attack. Then ending the sequence with the helicopters floating off silently into the mist.... Heck, now it sounds like a year-long project! ;)
Good point, Phil. And thank you for the reminder! Once I have a reasonable amount of film footage (in test render mode), I will decide which parts to use for the full and final render. I would rather plan for a longer sequence and cut back than to plan it short and have to add more later.
Also, I would agree with the "few seconds" rule if it is for the "action movie" genre. TBH, I'm not sure "action" is the genre I'm going for. I was (mentally) playing around with the idea of ...
WARNING! POSSIBLE SPOILER ALERT!
a more subtle video/message. Something like - putting the fly-in animation to slow, melodic tune with majestic jungle shots, river pontoons and canoes, rice patties, villages on the banks, etc. that gently put the viewer into a state of serenity. And then at the last moment shocking the hell out of their serenity with an unexpected and violent attack. Then ending the sequence with the helicopters floating off silently into the mist.... Heck, now it sounds like a year-long project! ;)
I like the idea - you can do so much to change the feel of a piece with different music! I still think that for a "production", you want to avoid having each shot lasting too long. And one of the benefits of doing the pre-render production that you are going through is so you only need to render what will actually be used. There is usually very little in an animation that gets fully produced and then ends up not being used, for this very reason.
A few quick comments regarding film sequence length and camera angles of scenes.
The exact length of sequences has changed over the years and the way various camera angles are used too.
Older movies will have more longer sequences and not as many cuts to different angles as newer ones.
You might want to check out the 3 part Chaos Cinema Series of Videos as you can see how the way action sequences have changed in movies.
http://vimeo.com/28016047
http://vimeo.com/28016704
http://vimeo.com/40881319
The creator of the videos has several others that are well worth watching too.
Jay, thanks for the video series! I started watching the first one, but have run out of time. So I will finish them tonight and post my thoughts relative to the animation I am attempting.
However, just as a quick reaction in contrast to the first part, I am reminded of a scene from Platoon (just came to me), which shows a completely different perspective. If you view the short 1 1/2 minute (relatively famous) snipit below, notice the length of the shots. And actually there is a scene cut off right after the last one that is longer / more dramatic. I would say the purpose of the shooting in this scene is not to convey action - action (mostly slow motion) is a means to the feeling being conveyed through cinema and music.
https://www.youtube.com/watch?v=k9EtS_qwiSw
I guess for me the bottom line is that "chaos cinema" has its place, though I personally do not like the style, and not all films (even action films) have to be done this way. I am not an adrenaline junky. ;) Much like all forms of art, each person has their preference.
Let me conclude that this whole discussion deserves its own thread! I know there are Carrara animators out there who remain relatively quiet in the forums, but a topic discussion such as this might engage others. I'd like to hear others' perspective on this, and how that influences their work.
Thanks!