Triem 23's "Hey, I'm Working on THIS!" page (Update: Eagle Transporter Flyby Test Jan 31, 2018)

1356736

Comments

  • Thank you, gentlemen.
    @ Simon. Yeah, I'm going to revisit the invasion using what I learned from the other tests. I was debating using mobile emitters for the lasers and explosions but I decided against it for a couple of reasons.
    In that particular shot, where every saucer is firing down at the planet and every bolt is going to impact on the surface, mobile emitters would work fantastically, but I kind of built that rig to repurpose into other shots--Let's say I have TWO fleets of ships firing at each other. I decided to keep the lasers separate from explosions for shots that miss. The theory being that I would either have ships taking hits/exploding being "hero" passes using independently animated models, or that, for a whole bunch of ships exploding, that I'd bake the explosion into the texture source layer. Also, keeping the explosions independent from the lasers lets me change seeds if I want, to simulate secondary or chained explosions, or large explosions caused at ground level by, say, tactical nukes or something.
    Mobile emitter spawning lasers and explosions is likely more efficient, but independent emitters is more flexible.
    That said, when I revisit that shot I'm going to rework the two-emitter laser/explosion setup into a single mobile-emitter simulation. On the other hand, we'll see if the mobile emitter version renders any faster than the two-emitter version.
    Also, once I'm done doing tests and have these techniques solid, I'm going to spring for the proper Dalek battlecruiser from Turbosquid and add in individual Daleks and 13 TARDISes*.
    All hail Peter Capaldi's eyebrows!
    (There's that line in "Day of the Doctor," "The Daleks know something's happening! They're increasing their firepower!" Yeah, what's more likely is that The Daleks said "THE DOC-TOR IS DET-ECT-ED! THE DOC-TORS ARE DETECTED!" Then a billion-billion Kaled mutants collectively soil the insides of their casings.)
    *Tardises? Tardii? Tardes? Tardis? What's the proper plural? (Joke--I know it's TARDISes.)
  • The more important question is, what is the collective noun?
  • Triem23Triem23 Moderator
    edited December 2013
    A Team of Tardises?
    A Time of Tardises?
    A Trick of Tardises?
    A Tintinnabulation of Tardises?*
    A Trek of Tardises?
    Dunno. But, speaking of "Trek," I do know a proper collective noun is "A Search of Spocks."
    *I like "Tintinnabulation of Tardises," as "Tintinnabulation" is a fun word to say, and, while wheezing isn't quite bell-like, a tintinnabulaton of Tardises would be something to hear.
  • Triem23Triem23 Moderator
    A test, trying to generate a plasma globe in Hitfilm.
    http://youtu.be/Dz6s_qEPZBo
  • Triem23Triem23 Moderator
    edited January 2014
    Two in one night? Amazing what one can do when a client isn't making you miserable!
    Over on Simon Jones's "Tarns Test" thread, user "Lars H" asked if there was a way of making particle clones that are less random in placement.
    I did a test using Atomic Particles to generate a regimented grid of clones, and it does work... Sort of. I mean, after seeing Axel's "Walter Mitty" test, I knew it would.
    Setup is basically the same--throw your greenscreen footage into a Particle Source comp, do your key, and embed that comp in your main comp.
    The color of the plane you put Atomic on tints the clones--so make this layer white.
    Use a 3D point as a transform center and rotate 90 degree on the X axis to put the clones on a "floor."
    Make the Particles size 100, turn any particle variation to 0.
    Because Atomic has a maximum particle size of 100, you have to get your camera pretty close to get a clone to fill the frame--but Atomic seems to use the full resolution of the source layer for particles, because I'm not getting massive artifacts when getting up close to the camera.
    For some reason, my source particles became vertically stretched in Atomic--this was fixed by adjusting the horizontal scale in the Particle Texture composite.
    Finally, Atomic automatically billboards particles, so a camera move like the one I've used in this test is going to give away the game--but this is true of video-source particle clones with the main particle sim. The camera move in this proof-of-concept test is more to make it very clear that the clones are arranged in a grid.
    Still--it works, and, since Atomic uses layer alpha to determine the edges of the grid, Atomic Clones can be placed in very very specific locations.  If I wanted to make, say, a triangular wedge of clones, I would create a composite shot with a plane, mask it to a triangle, then embed that in my particle comp. Bingo!  A wedge formation of clones!
    I assume disperse could be used for some randomness--using a control mask lets me limit the axes of dispersion, correct?
    http://www.youtube.com/watch?v=frYx8xD4oPw
  • Nice! That works well. One of us needs to take the time to make a proper crowd/army/something shot where we put the effort in to get it looking real, rather than these abstract tests...
  • Triem23Triem23 Moderator
    I have a plan....  ;-)
    In fact, it will incorporate both cloned me AND cloned spaceships.
    I'm just waiting for a day off where I have time to shoot proper elements, instead of just using that stock dancer from VideoBlocks.
    The plan is to just shoot on my block. Set up my Chroma Pop on one side, shoot myself several times for variety, strike the greenscreen, perform the camera move, then set up the greenscreen on the other side of the camera and shoot more elements--that way the lighting will be mostly correct. I might go ahead and shoot more greenscreen about halfway thru the camera move. (If you're looking down at the camera and the move is a 180 degree pan, picture groups--and separate shot elements--for left, center and right, relative to camera position)
    The plan is the camera pans 180, counter clockwise with the clone action being noticing what's above an looking up. Camera tilts up to show the fleet of flying saucers. Saucers begin firing. Camera tilts down to show clones dying en masses. Part of the theory here is that for the final segment--clone massacre--that I can extend the particle source and incorporate the beam weapons hitting clones into the texture source.
  • Triem23Triem23 Moderator
    Particle sims continue to be fun!
    Particle sim Asteroid field. Total time from start to render and upload, about 4 hours. This includes downloading models and a background and creating the titles.
    Should have spent more time finessing the camera move, but I fast-tracked this as over in the community project thread, they're debating using particle sims to do far and mid-field asteroids, so i figured this could double as a proof of concept test for them.
    http://youtu.be/yFoJRGkIcxw
  • That's cool, though I'm not quite sure about the lighting. The asteroids seem a bit too well lit, almost. I'd also say that this technique is best for adding detail to distant asteroid belts, rather than the close-up stuff towards the end of the shot.
  • Triem23Triem23 Moderator
    Probably true, although the lighting could be refined, (there are two point lights, one white, one violet. They really should be directionals, with, maybe a third and fourth light at low intensity, picking up colors from the background) and the rock models are all pretty low-poly (only about 1-2k polys each) with fairly low resolution textures (about 512 x 512) so this would also look better if I'd used higher-quality models that had some real details on them; craters, and such.
    I rushed this shot a bit, as I leave for an out-of-town job in the morning, and I wanted to get this up for the Community project guys as proof-of-concept.
  • Triem23Triem23 Moderator
    An adjusted version of the last shot, using a simplified camera move, adjusted lighting and higher resolution models.
    http://www.youtube.com/watch?v=1262QLFf0CQ
  • As I mentioned over on the community project thread, that second test looks vastly better. Lighting is hugely improved.
  • I'm sure that a lot of it has to do with the use of higher res models but like Simon says, it's amazing what a change in lighting can do :)
  • Triem23Triem23 Moderator
    Thanks, both! Now see what parallax and better music can do! :)
    http://youtu.be/s0RfzNKMbFg
  • The parallax has definitely made a difference - pulling out the different layers made me feel like I was more 'leaning into' the scene as opposed to watching it distantly. 
    I think the thing to concentrate on now you have realistic looking meteors, is to produce a more textured planet in the background. I think possibly the edges are a little too clean and the surface of the planet very smooth in comparison to the surrounding debris.
    What do you think? 
  • Triem23Triem23 Moderator
    Oh, probably true. That particular planet texture I created in Photoshop in less than a minute, and I've been recycling it excessively. In the Flying Saucer Fleet it's pink, with the asteroids, it's yellow, and, in the ringed planet I'll be rendering tomorrow, it's green.

    (Although it IS intended as a gas giant.)

    I think what I may try is animating some fractal noise and adding a sphere effect. Since the sphere effect doesn't react to comp lights, create a second layer, half black, half transparent, put that over the planet sphere in the main comp and rotate the dark half-sphere in place to be the shadow. Then I can try animating the planet textures.
  • Triem23Triem23 Moderator
    One more experiment with the same old technique... In this case I used the particle sim to create a ring system.
    Same basic particle clone technique with a few changes:
    Emitter was a quad, perpendicular to the planet and rigged to a point at the center of the planet. The emitter was rotated around the planet in the first second of the animation, dropping particles as it went to make the ring.
    Because the ring required so many more particles than the asteroid field, the particle sources were simplified--instead of six different texture sources, there are only two--additionally, shadow casting was turned off in the texture composites--we're not getting close enough to fill the screen with a rock, so I don't think anyone would notice.
    Particle texture comps were dropped from 800x800 resolution to 200x200.
    The planet itself was hooked to a camera gimble like the particle texture sources. Once the rotation was matched, I embedded the planet comp in the main comp, converted it to a 3D plane, then positioned it at the center of the ring. The planet comp was set to align towards active camera. This enabled the rings to surround the planet and occlude "As you would expect." without having to worry about masking a 3D model within the comp.
    I also went ahead and created a more detailed planet texture for this one.
    http://www.youtube.com/watch?v=ICZVzG_Axks
  • Lovely!
    For the asteroid field, it struck me that one missing element is a general sensse of microdebris/dust between the asteroids. Currently the space around the asteroids all feels a bit too clear.
  • Triem23Triem23 Moderator
    Very true. But the current iteration of the shot is near the limit of GPU performance on my machine. Thinking I might create a noise texture of tons of little dots and attaching that to the emitter to try and fill in the gaps. (align to motion, not billboarded?) Unless you have a better idea?
  • Wow - gorgeous colours there :)
    I still feel like there is a difference in the planet and the surrounding space which is preventing it from looking really "realistic". It's possibly the lighting but mostly (after watching it maybe 5 times) I think what's wigging me out is the difference in the colour blue between the space and the planet. 
    Your planet, although looking great as a gas planet and with the textures, is just a bit too aquamarine in comparison to its purpley-blue surroundings and so looks a little detached. 
    But anyway, that's just an extra. 
    To have put together what you have is amazing - I especially like the camera movement where you draw in towards the planet at the end of the shot. Starts to create an awesome sense of scale. 
  • Very true. But the current iteration of the shot is near the limit of GPU performance on my machine. Thinking I might create a noise texture of tons of little dots and attaching that to the emitter to try and fill in the gaps. (align to motion, not billboarded?) Unless you have a better idea?

    When you say you're nearing the GPU performance limit, how do you go about determining that?
    I agree with Simon about the clarity around the asteroid belt.  The camera movement is great! 
    As the camera moves around the planet changing in elevation the shot begins to illustrate increasingly more depth.  It's only at the 10s mark when everything is aligned almost straight on with the camera that it feels a bit off.  Otherwise as usual this is looking good! 
    If you were to change out the background for a more subtle one, I'm curious how that would affect the shot.  There's something about it that pulls me away from the focus of the asteroid belt and planet. 

     

  • I really like the Particle Planetary rings video! maybe tone down the background a little bit though. The colors are REALLY extreme at this point and don't seem very realistic. Something I try to look at every time I make a space scene is 2001 A Space Odyssey or the first 2 Alien movies. Space looks best when its harsh, but yet still having dark/dramatic lighting. Thanks! Good job!  
  • Triem23Triem23 Moderator
    When you say you're nearing the GPU performance limit, how do you go about determining that?
    I agree with Simon about the clarity around the asteroid belt.  The camera movement is great! 
    ...
    If you were to change out the background for a more subtle one, I'm curious how that would affect the shot.  There's something about it that pulls me away from the focus of the asteroid belt and planet. 

     


    Well, I know I'm pushing the limits of my GPU because with this shot and the third iteration of the asteroid field push (the one with parallax), I became familiar with "Hitfilm Error 200: Your GPU does not have enough memory."
    That's a good hint. ;-)
    (also @ Simon) The ring system has way way more particles than the earlier asteroid field: For the rings, I ended up reducing my texture sources from 5 to 2 and my particle count kept coming down as my particle scale kept creeping up. I think I know how to fill in this ring more with some dust and micrometeors, and I'm going to try that a bit later.
     


     


    Wow - gorgeous colours there :)
    I still feel like there is a difference in the planet and the surrounding space which is preventing it from looking really "realistic". It's possibly the lighting but mostly (after watching it maybe 5 times) I think what's wigging me out is the difference in the colour blue between the space and the planet. 
    Your planet, although looking great as a gas planet and with the textures, is just a bit too aquamarine in comparison to its purpley-blue surroundings and so looks a little detached. 
    But anyway, that's just an extra. 
    To have put together what you have is amazing - I especially like the camera movement where you draw in towards the planet at the end of the shot. Starts to create an awesome sense of scale. 


    Oddly enough the planet texture is actually purple. I used a green diffuse color on the model to cool it a bit, and the three lights in the scene are a light green (That's the sun, as you can see in it's flare), a violet and a blue, sampled from the colors of the background image. I see where you're coming from on the colors not sitting right, however, so I'll revisit that when I get the ring filled in.
    Oh, Kristie, that planet texture is unofficially dedicated to you. The previous planet texture is one that I've used on all my space shots since May (Hey, when doing tests, you reuse elements, right?). So I figured I'd change it up... I have an idea for using the Sphere filter to use an animated texture. Except the Sphere filter doesn't respond to scene lighting....but I MAY have a workaround for that. We shall see... (that might be a v3 of this shot. For v2, I just want to fill in the ring gaps and adjust color....)
     


     


    I really like the Particle Planetary rings video! maybe tone down the background a little bit though. The colors are REALLY extreme at this point and don't seem very realistic. Something I try to look at every time I make a space scene is 2001 A Space Odyssey or the first 2 Alien movies. Space looks best when its harsh, but yet still having dark/dramatic lighting. Thanks! Good job!  


    (Also @ OrangePekoe/Kristie) Hey, that background is a NASA photo! Nah, it's very very very colorful, isn't it? And there's a grade over the entire shot pulling saturation down! The bg I used is obviously processed false color, and I did some warping of the image in photoshop to try and get it to work as an environment map, but th wrap isn't quite right. I'll try something else.
    Kaveh, I always liked "Babylon 5" space. B5 tended, overall, to use a lot more nebulae and dust clouds in space shots than most shows and films (Partially, I think, to have a bright, colorful background to silhouette the black Shadow vessels)., but I'll totally agree that the overall color balance of this shot is very very saturated. ;-)
    Thanks for the feedback everyone!

  • Wahoo! A planet texture unofficially dedicated to me - a high honour indeed! :)
    Wow I am going a bit blind if I can't tell that the planet is purple - but I suppose it's the green lights which messed me up a bit. 
    Lovely NASA photo - but I do agree with OrangePekoe - it is a little too colourful. I'm looking forward to what you do to it next!
  • Triem... what kind of GPU do you have>?
  • Triem23Triem23 Moderator
    Next Version:
    http://www.youtube.com/watch?v=GMc9rJ84hiM
    MichaelJames: nVidia 580GTX M
  • Triem!  Great job, the second version - you've made some great improvements.  Is a single texture (dust/cloud) tied to each asteroid?  Can you make it spread out a bit further and hmm.. "smoother" from the asteroid?  The drop off is quite noticeable at some angles. 
    What I really appreciate is the stars light reflecting through the cloud/dust and the asteroids.  It's dampened and not as overpowering as the other version!
    Keep it up :D
  • Triem23Triem23 Moderator
    OrangePekoe--thanks!.
    The initial ring is a quad emitter, perpendicular to the planet. The emitter is attached to a point. The emitter point is parented to a point in the center of the planet. The first two seconds of the animation is the emitter point being rotated around the planet, dropping particles as is goes. These particles are rotating 3D models, gimbled like the saucers and asteroids tests.
    The SECOND (i.e. the changes) is a second emitter (actually on a separate layer), parented to the same points rig. This second emitter is a bit taller and a lot wider. This second emitter contains two particle systems. One is using the stock, built-in concrete debris, and is the micrometors outside the main ring. The second particle system is built-in Dark Cloud Smoke at about 64 alpha, normal blend,  using a color sampled from one of the rocks in the first ring.
    The entire setup of particles and planet is nested into another comp, all particles/Planet as black, sitting on top of a white plane with some blur--the becomes the luma matte for the flare and light rays.
    The low alpha of the dust made a nice luma matte that really sells the flare. I also changed flare from "Sun Glare" to "Orange Spikes" which I think is better for this shot.
    As noted above these last couple projects have been pushing my GPU (I get the occasional ERROR 200, your GPU does not have enough memory), and I've been unable to render this version at all as an mp4. However, I tried rendering as an image sequence, then reimporting to render the mp4. I think this is less processor intensive, and will allow me to add back some of the particle detail I had to drop--I'd like to up the particle count of the first dust ring and lower the scale, then dd a second dust ring with a different texture to ease it off a bit, because the transition IS a bit harsh. I'd also like to up my micrometeor particle count. I'd also like to add my texture BACK to the first ring--right now they are untextured rocks.
    I'm not going to be able to return to this shot for a while--I have two projects I MUST get back to (one of them has a hard deadline of Jan 30) instead of having fun in Hitfilm.
  • Very pretty! Occasionally the light flare remains behind the asteroids towards the end, though, which is a bit odd. If the flare is visible then it should be in front of everything else in the frame.
    Looks great, though, and it's a cool camera move to reveal everything. The general look of the shot is VERY B5. I half expected Narn heavy cruisers to jump in:
    http://youtu.be/f-uhi_qIgSU
    Great music in that sequence.
  • Triem23Triem23 Moderator
    edited January 2014
    Oh, Gorash VII, such a beautifully staged sequence, intercutting between G'Kar praying and meditating and the absolute annihilation of the Narn fleet. That "Holy Zarquon's singing fish!" moment when the Shadows reveal they can disrupt jump points. Great stuff. And did you catch that the bridge of the Narn Cruiser is a Virtual set?
    Followed by that sequence where the Centauri start slamming nickel-iron rocks at orbital speed in Narn--and the explosions and Shockwaves are visible from orbit... Until the clouds are boiling over the explosions, dimming them... That's a loooot of dust, stone, particulates in the air. That's condemning a world to decades of nuclear winter.
    That's the single scariest moment of B5.
    Big fan. I'll take a comparison as a compliment.
Sign In or Register to comment.