First go at all composite shots

mark_emark_e Website User Posts: 190 Just Starting Out
edited February 2014 in Practical Filmmaking
I've been using hit film for a while but mainly to augment vegas and a bit of tracking, screen replace, keying,  roto tidying up shots etc.
Along side I've been gradual working through the tutorials, think I have done them all now!, this is my first go at putting something cohesive together, used the first 3 scenes from voyager intro as my inspiration :)  Hopefully haven't let Axel and Simon down and have paid attention! particularly proud of my sun and the fog was a nightmare and I had no idea how hard it is to get a nice camera movement but I think I got there in the end ish I keep seeing bit's I missed now.  Thx for the tutorials!
I've watched it too many times now to have an objective opinion any more! happy to get some feedback, would like to do better next time as always!
http://www.youtube.com/watch?v=iXN9t1gDPZ0
thanks
Mark

Comments

  • Triem23Triem23 Moderator Moderator, Website User, Ambassador, Imerge Beta Tester, HitFilm Beta Tester Posts: 18,262 Ambassador
    Nice work! I see Capt. Sheridan took the scenic route to Corianna 6 after his last visit to babylon 5. ;-)
    OK, in the first of the three shots, I'm loving the flare (solar flare, not lens), and angle. I think the outer edge of your star's photosphere is standing out a bit too much as a clean edge, and I think you need to slow down the individual particle movement a little bit--it's a tad fast and it's not reading to me as quite the proper scale for something as massive as a star. Maybe some glow, too?
    Third shot--I want to do a shot myself echoing the Voyager opening credits shot where Voyager flies through a cloud of gas and dust like you've done, so let me share with you a couple of things I intend to try myself: I want to attach a small deflector cube to the front of my ship to push the particles from the path of my ship. I haven't worked out the mass, friction and bounce settings yet to get the particulars to "slide aside," instead of bounce away like a ball hit by a bat, but that's just slider tweaking. I'm also going to put a long, skinny cubical turbulence force in there and attach it to the rear of my ship to create a wake. That might be fun for you to play with, too.
    Second shot I have no notes, other than to say nicely done, and I love your space background plate.
    Camera moves in general--I'm hoping to do a tutorial on this, but it's three or four down my "tutorials I want to make wishlist.", but here's a few things I've tried to improve my camera moves in Hitfilm:
    Obvously, Hitfilm doesn't have true bezier curves. What I often do is create a multi-point rig for my camera...
    Let me use this as an example:
    http://www.youtube.com/watch?v=yFoJRGkIcxw
    Now, this shot has some issues (the v3 is WAY better), but this one has a fairly advanced camera movement. Basically, the camera is moving on the z axis through the asteroid cloud and, occasionally, slewing up, down, left and right around rocks--what I did for this shot was to create an array of three points, close together.The camera is parented to "Camera Z" point. Camera Z is a simple, two-keyframe motion--start and end. Camera Z is parented to "Camera Y" point. Camera Y is controlling the up and down translation of the camera. Camera Y is parented to "Camera X" point. Camera X is doing the left and right moves. First I set the camera Z motion, then started adding keyframes on Cameras X and Y as needed to slew around. In this animation the moves are still too fast to feel correct for the shot, but, by using a separate point to drive each axis, I get a lot smoother movement than if I tried to do all that motion with a single point.
    If I'm adding camera rotation, I'll usually create a rotation point for the array, (This "Tripod Point" would be parented to "Camera Z") and place it about 20 pixels behind and about 20 pixels above or below the camera. The reason I do this is because the majority of tripods, cranes, jibs and stabilizers to not rotate around the nodal point of a physical camera, where most CG cameras DO rotate around the nodal point. Offsetting the rotation pint of the camera a little bit makes camera rotations feel a little more real, since it more closely approximates the way a real camera would move.
    Remember--Hitfilm's camera orientation feature should only be used to point your camera on frame 1 to start your animation. orientation is locked between 0 and 360, and will always move between oreientation values along the shortest possible distance--this can lead to unpredictable behavior. All camera rotation animation should be done using the individual axis controls for x, y, and z.
    When rotating the camera on multiple axes at once--let's say the camera is panning left and tilting up at the same time--offset your keyframes so that each rotation axis starts and ends on a different frame. Again, think of what would happen if you were doing this shot on a tripod. No camera operator is going to do a perfect pan and tilt at the same time. Either the pan or tilt is going to start just a fraction of a second earlier and end later. Also, look at trying different interpolation values for each axis. using "smooth in" for the pan and "smooth out" for the tilt is going to look different that using "smooth" interpolation for both!
    Consider shifting the "alignment" (it's on the top of the viewer pane) from "local" to "world" or "view" when adjusting camera moves--"World" reads all axes relative to absolute position in 3D space. "Local" reads all axis relative to the selected object(s)/layer(s), and "View" reads all axes relative to the current position of the active camera. Often, shifting to "View" orientation when refining your camera can make it a lot easier to control the move.
    If you want to get fancy (which you don't need for a space shot, but I'm bringing it up, anyway.) you can even build a virtual dolly/crane for a camera: Create a point for the Dolly base. create a point for the base of the crane arm a little above the "Dolly" point and parent "Crane Base" to Dolly. Create a third, "Camera Mount" point for the far end of the crane arm and parent that to the Crane Base. move your camera close to Camera Mount and parent that. Now you can use your Dolly point to move your crane, the Crane base point to rotate the crane arm on the x and y axes, and adjust the Camera Mount to point the camera at it's target--this point rig will behave as a traditional crane, since we're mounting the camera at the end of a virtual arm and the arm on a virtual dolly, which will yield much more realistic camera movements for "ground-based" animations--again, because we've set up this rig to mimic how a physical camera would be mounted.
    Hope this is helpful, and keep up the good work! :-)
  • mark_emark_e Website User Posts: 190 Just Starting Out
    edited February 2014
    Thanks for the good feedback Triem,
    I hadn't realized about the local and global camera positioning options! that would have made my life easier i spent ages getting those camera pans smooth, was using what i learnt from the roto keyframing less is more! did make some simple rigs and good tip about not stopping the rotation etc all on the same key frame I derived that and started doing that right at the end just because it looked better it wasn't a conscious thought but it makes sense now, I'll get some smother tracks in the future.
    Scene 1 I agree sun looks a little harsh there is some glow but perhaps it's a bit too subtle /   getting lost in compression I added a fair bit of grain to get rid of some banding, I'm going to play about with some sort of displacement layer I think I really want to make the whole shot look hot some sort of heat haze I think would look good.  I couldn't make my mind up on the rotation speed :) the big flare and the sun I did with atomic particles / animated fractal noise to drive the burn map in the procedural fire, made grey scale ribbons in atomic particles flattened it and use that with some warp and used that as well, was getting massively distracted playing with that :-)
    Scene 3 I'm annoyed I dialed it all down to much I had rigged the ship along the lines of how you described if you look closely you can see the fog getting moved about and sucked in behind which looks very cool when it's more active a bit like a proper vortex with an attractor and turbulence but I was struggling to get the balance and avoid particles getting stuck so at the moment there are too many going through the ship to really notice it, however I have sussed it I think I had an epiphany in the gym this morning :)
    At the moment I have it set up like this because I wanted to keep the overall global movement of the fog with out pinging stuff around as you described and avoiding stuck particles.
    attractor behind ship 
    deflectors left and right angled with angled turbulence deflectors as well
    no front deflectors just a dettractor in front and then a separate point particle simulator that's yellowish animated in front of that to give the particles around the ship the colour of the yellow lights.
    I forgot about the lifetime property of the particles for the second system and I bet if I put the deflectors in the mobile force fields rather than behind them I won't get stuck particles.
    So I'm going to re enable the front deflectors this will give a clearer path behind where you can see the fog getting sucked in at the moment it goes through as well as around the ship as I have them off.
    Move the forcefields over the deflectors
    add an additional particle simulator in front of the front detractor with smaller particles with less mass and friction to emphisise the turbulence vortex behind the ship but then use lifetime to increase mass and friction near the end so they don't go pinging off around the main container.
    If that works and I have a bit of separation like there's a shield around the ship then I want to work out how to do some sort of simulated electrical interaction with some of the fog particles but I'll cross that bridge when I get to it!!
    it's quite addictive!
  • Triem23Triem23 Moderator Moderator, Website User, Ambassador, Imerge Beta Tester, HitFilm Beta Tester Posts: 18,262 Ambassador
    edited February 2014
    Sounds like those changes will work. Turn friction down to 0 in your particular sim. That helps avoid "stuck" particles whem slamming deflectors and detractors through existing clouds. I hadn't thought of detractors for "navigation shields" good idea. I did notice the subtle turbulence, but, as you noted, the current iteration was dialed down too much.
    Combining fractal noise and atomic to generate a burn map is clever. I have played with fractals to drive burn maps, but not atomic!
    When you're tweaking particle forces and dynamics, are you using preview mode? Think Simon Jones often suggests that. It speeds up working, and it can ve easier to see what's going on with the white dots, compared to the full sim.
    The "world/local/view" toggle can also make object animation easier, since "world" locks the object axes to the master axes, itcert makes object positioning easier, although I tend to lock animation positions before I worry about animation rotation.
  • SimonKJonesSimonKJones Moderator Website User, HitFilm Beta Tester Posts: 4,450 Enthusiast
    Lovely. I tried a similar shot of a ship passing through fog a while back and got OKish results - I love Triem's idea of using a turbulence field for the ship's wake. 
    Triem's basically covered the feedback I would have given - though it's worth mentioning that any VFX test containing B5 stuff gets extra points from me. ;)
  • Triem23Triem23 Moderator Moderator, Website User, Ambassador, Imerge Beta Tester, HitFilm Beta Tester Posts: 18,262 Ambassador
    Cuz, you know, Babylon 5! I will fistfight to defend the thesis that the EA Starfury is the coolest fighter in sci-fi. Vector thrust,inertial physics for the win!
  • mark_emark_e Website User Posts: 190 Just Starting Out
    This is more how I was imagining it, getting there.
    Did what I said I was going to do so, detractor in front, turbulance left and right, attractor behind, about 8 deflectors in front and with the secondary particle system size start at 0 quickly upto full size then tail off to 5% as it tails off slow the speed down.
    Probably around 100,000 particles in each frame.
    Next steps, offset the ship back a bit to give clear separation from the front
    Extend the animation a little to capture more of where it comes out of the fog, loving that
    Reduce the secondary particle system a bit, got a little carried away there!
    See if I can add some sort of displacement type shield in the front and I'm thinking mobile emitters exploding on contact with a plane to make some sort of electrical static type effect (ish perhaps sort of)  

    http://www.youtube.com/watch?v=xa-n3W7GV3k
  • Triem23Triem23 Moderator Moderator, Website User, Ambassador, Imerge Beta Tester, HitFilm Beta Tester Posts: 18,262 Ambassador
    Ok, now THAT is GORGEOUS! Once you offset the ship and finish the shield, you need to do a breakdown or tutorial of how you did your forces setup.
    For the shield, maybe an animated texture, 4-5 frames of lightning or plasma, and high friction, so it "sticks" to the deflectors?
  • OrangePekoeOrangePekoe Website User Posts: 478
    This is more how I was imagining it, getting there.
    Did what I said I was going to do so, detractor in front, turbulance left and right, attractor behind, about 8 deflectors in front and with the secondary particle system size start at 0 quickly upto full size then tail off to 5% as it tails off slow the speed down.
    Probably around 100,000 particles in each frame.
    Next steps, offset the ship back a bit to give clear separation from the front
    Extend the animation a little to capture more of where it comes out of the fog, loving that
    Reduce the secondary particle system a bit, got a little carried away there!
    See if I can add some sort of displacement type shield in the front and I'm thinking mobile emitters exploding on contact with a plane to make some sort of electrical static type effect (ish perhaps sort of)  

    http://www.youtube.com/watch?v=xa-n3W7GV3k


    Bravo Mark!  That looks great!  Great use of the particle sim, you've got some great control over your particle environment in this shot.  I like the idea you have of mobile emitters exploding on contact with a plane.  Can't wait for the next revision!!!

  • SimonKJonesSimonKJones Moderator Website User, HitFilm Beta Tester Posts: 4,450 Enthusiast
    Love the particles swooping in behind the ship, very cool. I wonder - rather than just an attractor on the back, if you had a couple of separate attractors on each wing, or maybe top and bottom, it might result in more of a swirling, curling motion.
  • mark_emark_e Website User Posts: 190 Just Starting Out
    Thanks everyone, I'm loving playing with this and some great suggestions, I wonder if I can some how animate a couple of the attracts to spin them and what that would do to Simons point, hmmm will have a play tonight after work and horses :)  have a bit of an insane idea involving the mobile emitters and lightning effect if it all works how I think it does and I can create an animated layer with lightning then use that as a particle source for the mobile emmiters hitting the shield (and if that works perhaps having a few in the fog to make it flash a bit it could be good!
    Of cause if that was in real space I'm sure it wouldn't move like that at all and the ship would've be instantly vaporized if it hit something like that at speed but I'm not worrying about that, it's starting to look cool ;-) 
  • RossTrowbridgeRossTrowbridge Website User Posts: 423 Enthusiast
    This is really looking great. I'd love to see a tutorial on it as well. What ship model are you using? It's definitely not a Whitestar, and is different from the Bluestar model I have...
  • mark_emark_e Website User Posts: 190 Just Starting Out
    Added shielding and interacted with fog a bit more at the front.  Didn't go for mobile emitters in the end did it with normal particle simulator.
    I lost a bit of the detail from the previous test I should have bumped the fog up a bit more around the ship but I've run my course with it is one now it is what it is :)
    My favourite bit is just as it exits the fog cloud where they trail fades out and flips behind it.
    I hadn't seen anyone do the shielding like I did that's in full 3d space around the model so you can spin the camera around it an it looks fine, don't know if that's of interest for a little tutorial as well or if it's old hat :)
    Need to take a break for a few days will get a tutorial together over the next week or so with a different scene that lends it's self better for a tutorial, ohh and I can do the twiddly bits as well like simon suggested you can just see them if you look close, faked it rather than physics but it looks cool you just can't really see it much in this shot but I'll do something like a ship hitting a single cloud and dispersing it for the tutorial and then you'll see it better.
    http://www.youtube.com/watch?v=dSDYTsjkRdY

    The model was from foundation 3d, it's a nice model.
    http://www.foundation3d.com/index.php?categoryid=38&p13_sectionid=441&p13_fileid=1783
  • Triem23Triem23 Moderator Moderator, Website User, Ambassador, Imerge Beta Tester, HitFilm Beta Tester Posts: 18,262 Ambassador
    Yeah, that's a nice shield. Did you do that with mobile emitters hitting your deflectors or go a different route?
    I also like the camera move at the end after the White Star flies off screen, tilting down to the fog.
    Speaking of--fog's thicker in this version. Did you up particle count, texture size, or both?
    Rtrowbridge, I think that's a White Srar, personal transport version, as seen in B5 ep "522," "Sleeping in Light," unless this model is just an original interpretation of the White Star silhouette.
  • mark_emark_e Website User Posts: 190 Just Starting Out
    edited February 2014
    The shield I duplicated the model had it still parented to the ship rig, scaled it up so it was bigger than the ship, added white solid fill effect, animated fractal noise, waves, chromiance, paralax, displacement, glow and flicker changed blend mode to add then duplicated that again and hid it using it as the displacement source.
    Basically I used Simons UFO tutorial where he explains how to add glowing lights but rather than changing the duplicated model in the import I applied a blanket white fill as a base as an effect to base everything off then made it bigger.
    The additional glowing interaction in the cloud is an additional standard particle system with the inbuilt flame effect set to add mode with low alpha boarding turned off textures spinning stupidly fast 8 iterations / sec and motion blur and size and spin variation turned on for the particle system.
    I pushed the main particle count up for the fog layer ~15000 / sec and let it run for 5 secs before stopping it but there is global turbulence running and random texture spinning as well to keep my PC warm :-)  the ship fog layer was at 1000 / sec continuous with texture spinning and the glow layer was at 50/sec but very low alpha. then cut the particle system just as it exits the cloud so you get that nice exit, the force of the bounding box is not enough to stop you pushing particles out like that.
    5 plane deflectors, 1 global bounding box deflector, 1 global force, 2 local turbulence forces 1 detractor, 1 attractor 
    ~50 mins to render 16bit float for the whole lot to PNG then I put the grade layer and lens dirt on after, set the levels properly with scopes in vegas so youtube didn't clip the 0-255 output from hitfilm and added sound in vegas rendered out to ~20mbps for upload.  Adding grain really helps with the fog to reduce banding need 32bit float really perhaps the next version ;-)
    Obviously slowed down a bit and you have to be sensible managing when you turn the particles on as you are working but was rock solid never crashed once was really pleased.  Secretly I was just keeping adding more particles and physics to see when it went bang but I obviously wasn't trying hard enough!
  • Triem23Triem23 Moderator Moderator, Website User, Ambassador, Imerge Beta Tester, HitFilm Beta Tester Posts: 18,262 Ambassador
    50 min to render? Your GPU kicks my GPU in the head! If I were pushing around 100k+ particles, with physics, models, etc, my machine would take far longer than that!
  • mark_emark_e Website User Posts: 190 Just Starting Out
    Got a 3gig 780ti o/c works really nicely with hitfilm, and might have a bit of a generally over the top whizzy system that far exceeds my capabilities ;-)
  • MichaelJamesMichaelJames Website User Posts: 2,038 Enthusiast
    Mark what are the specs of the rest of your system?  I tried to do a particle simulator with 100,000 particles per second and it was painfully slow but fine for a while and now I can't even get Hitfilm to respond so I can disable it
  • mark_emark_e Website User Posts: 190 Just Starting Out
    Mark what are the specs of the rest of your system?  I tried to do a particle simulator with 100,000 particles per second and it was painfully slow but fine for a while and now I can't even get Hitfilm to respond so I can disable it

    Don't forget I totaled the amount of particles in existence, not set it at 100,000 per sec I might have to try that :-) I had ~100,000 particles in existence all being affected by global turbulence and local forces I didn't keep creating new ones except off the nose cone.  I obviously kept the particle count down until render time with that number of particles it would be slow enabling it but usable.  For really heavy scenes like that and the one I did the stupidly dense meshes :-) I got in the habit of turning the camera off making changes turning it back on again when I started pulling everything together, I tried to break the scene into smaller comp shots as well but I found that hard for both of those as I really wanted everything interacting with each other so ended up having duplicate environment comp shots for some of it and then just copy and paste into one with the camera off for the final shot, turn camera on and render.  I was pushing it a bit I think.  Normal usage it's fine.
    I've got a Del T7600 2 x 4 core 3.30gHz Xeon Nvidia 780-ti, 1300w power supply 16 gig whatever the fastest ram i could get at the time was.  I went for slightly lower number of cores but faster overall as I noticed that so much gets offset to the GPU, not sure if I should have go for 2 x 8 core but slower overall but this setup works fine for me and there's not a lot that max's out all the cores.  Have just got ssd drives and a couple of slots to swap out 1/2tb drives as I fill them up, didn't bother with raid or anything as most stuff i'm doing now disk access isn't the limiting factor.
     

  • SimonKJonesSimonKJones Moderator Website User, HitFilm Beta Tester Posts: 4,450 Enthusiast
    Bear in mind particles per second has to be taken into consideration with their lifetime.
Sign In or Register to comment.