PDA

View Full Version : Express



Glendalough
07-01-2008, 11:28 AM
Started this in 8.5 just before I got 9 about a year ago. The HVs were taking too long to render on an the old machine back then.

Hoping the way it's intended to come together will now materializes. It will be trimmed and the HVs can be upgraded plus the new photorealistic motion blur will look better, but still a lot of other things need to be done to bring it up to standard.

Any CCs or ideas very welcome.

Animation (9MG):

Express (http://www.jamespowersjr.com/jp-sr/train-h264.mov)

CMT
07-01-2008, 12:37 PM
Pretty cool! I like the slick look of the train and the camera angles.

StevieB
07-03-2008, 07:06 AM
Same here! And that camera angle vs. rotation of the train you have going on the last part really stood out (in a good way ;) ), great work!

Glendalough
07-03-2008, 03:34 PM
CMT & StevieB:

Thank you both for the compliments and encouragement!

Glendalough
07-12-2008, 06:42 AM
Having a bit trouble working at this size. First the computer would crash on render until I put the segment memory up to 64 from 32. Wonder how many people are doing this size nowadays in Lightwave?

Now the ground mist has partially gone away or something (?) and the HVs are taking long enough to render on an 8 core with 4GB, about 15 to 30 mins.

Anyway, this part, where the train goes by, is the crux of the animation so must bring in more HVs (dust etc.) and interaction, just hope the render times don't go through the roof. Wonder is 'pixie dust' any good? Have never tried it.

StevieB
07-12-2008, 11:18 AM
I wish I knew more about HV's so I could help you but I myself am still working on learning Character Animation and Arch Viz so I have had little time/reason to start experimenting with them. Sorry :(

archijam
07-13-2008, 05:57 AM
Something sticks to my eye: The fin on the front of the train really needs more subdivision, it's really jagged, otherwise a beautiful design ..

For the HVs, try searching for 'Particles' here (http://www.newtek.com/forums/showthread.php?t=77002).. not too expereinced with this. Pixie dust should be better for the sparks I think ..

Glendalough
07-14-2008, 06:20 AM
StevieB and archijam : Thanks for the writing!

Sub divided, smoothed out that bit on the engine.

Can't work out what has happened to the ground fog. Same file, it shows up in 8.3(?)

Having trouble with getting the HV Sprites in 9.3.1, they just don't seem to have the depth of earlier versions, either go too threadbare and thin, or else, when beefed up, they become very solid. There was a change in this version and others have noticed this as well:

http://www.newtek.com/forums/showthread.php?t=86066

Posted a 3 sec animation (1.6Mb) full size, the H264 compression is pretty amazing though a bit of banding in sky as this is not the very highest quality.

Train Full Size (http://www.jamespowersjr.com/jp-sr/2ndtrain.mov)

bobakabob
07-14-2008, 05:34 PM
Powerful composition and great modelling! Looks akin to the style of the Iron Giant!

Mr Rid
07-14-2008, 06:27 PM
Having a bit trouble working at this size. First the computer would crash on render until I put the segment memory up to 64 from 32. Wonder how many people are doing this size nowadays in Lightwave?

Now the ground mist has partially gone away or something (?) and the HVs are taking long enough to render on an 8 core with 4GB, about 15 to 30 mins.

Anyway, this part, where the train goes by, is the crux of the animation so must bring in more HVs (dust etc.) and interaction, just hope the render times don't go through the roof. Wonder is 'pixie dust' any good? Have never tried it.

I always render smokey voxels at half the res of the rest of the frame. The higher res you go with voxels the hairier and sharper they look
which doesnt help smoke/steam, and of course you can cut render time to 1/4( the difference between 640x and 1280x is 4 times as much area to render). Render half res and scale it in comp and the scale interpolation will inherently soften smoke.

To thicken sprites, you might try a gradient in the opacity based on texture density (or texture value), and or in the luminosity to simulate shaded areas. I would also suggest enveloping the dissolve as well as size over the first frames of birth so they dont appear to pop on suddenly.

More particles or a larger size and probably a higher amp hypertexture at the beginning can minimize the HV puffball syndrome. Velocity Translate can also help hide the puffballs as the hypertexture moves thru the particle according to the direction it is traveling.

Sometimes a Speed gradient in the size or velocity can thin/stretch and thicken/shorten particles appropriate to behavior.

A little vortex wind can help break up the motion of the particles in an organic way.

Mr Rid
07-15-2008, 01:58 AM
I...I would also suggest enveloping the dissolve as well as size over the first frames of birth so they dont appear to pop on suddenly..

I didnt state that well. I meant to use a gradient in the dissolve so voxels ease in, not to use the actual envelope graph.

Glendalough
07-16-2008, 02:31 PM
Powerful composition and great modelling! Looks akin to the style of the Iron Giant!

Thanks! The Iron Giant looks pretty interesting, must have a look at it.

Glendalough
07-16-2008, 02:33 PM
I always render smokey voxels at half the res of the rest of the frame. The higher res you go with voxels the hairier and sharper they look
which doesnt help smoke/steam, and of course you can cut render time to 1/4( the difference between 640x and 1280x is 4 times as much area to render). Render half res and scale it in comp and the scale interpolation will inherently soften smoke.

To thicken sprites, you might try a gradient in the opacity based on texture density (or texture value), and or in the luminosity to simulate shaded areas. I would also suggest enveloping the dissolve as well as size over the first frames of birth so they dont appear to pop on suddenly.

More particles or a larger size and probably a higher amp hypertexture at the beginning can minimize the HV puffball syndrome. Velocity Translate can also help hide the puffballs as the hypertexture moves thru the particle according to the direction it is traveling.

Sometimes a Speed gradient in the size or velocity can thin/stretch and thicken/shorten particles appropriate to behavior.

A little vortex wind can help break up the motion of the particles in an organic way.

Thanks for all the tips!

The comping idea sounds good but may not work too well in this case. Am using radiosity, and reflections from the HVs are being cast over the objects, adding to the render times of course. Know what you mean about the voxels looking hairy and too sharp.

Am working on the dissolving in to avoid the popping with the sprites, have set most gradients to particle age. Think the Density is set to fade out to avoid the popping at the end. Get the Density and Opacity confused. Think the Opacity is like transparency and Density a sort of volume weight thing (?) .

Wondering if I should try some different textures or would this make little difference with Sprites? Have got to go over all the docs again carefully.

Am testing at the moment by removing all the objects and turning off radiosity so I can get a fairly quick render as you need the whole sequence (180 frames) to see what is happening. Don't really know if there is much point in using Viper as accuracy seems to be in question with a lot of people (?)

Am using Velocity Translate. Must try the vortex wind when the voxels get a bit more sorted out. Thanks again for writing.

Mr Rid
07-17-2008, 12:14 AM
...
The comping idea sounds good but may not work too well in this case. Am using radiosity, and reflections from the HVs are being cast over the objects, adding to the render times of course.

The idea would be to make the emitters in the train layer 'Unseen By Camera' so they still trace. Then in your separate voxel layer, make all the objects Matte so the sprites are isolated- also enable Flare2Alpha under Image Filter Processing for your sprites to render alphas.

But yeah, radiosity with HV is what is greatly increasing the render time. I dont know that it is necessary to enable volumetrics in the radiosity. But I suggest making Rays Per Eval 36, and see how it looks with Interpolated enabled to drop the render time down. I dont go above Ray Recursion of 2 for reflections. LW defaults to 16 which is twice as high as I have ever needed for even dense refraction.

Not sure why your sprites are so bright when it doesnt seem like any illumination source is motivating it(?). But you could just stick a linear or other light in the middle of the sprites (set to exclude the sprites) to sim the same effect of the sprites softly lighting the surrounding without volumetric radiosity.

Glendalough
07-17-2008, 01:44 PM
The idea would be to make the emitters in the train layer 'Unseen By Camera' so they still trace. Then in your separate voxel layer, make all the objects Matte so the sprites are isolated- also enable Flare2Alpha under Image Filter Processing for your sprites to render alphas.
...
Not sure why your sprites are so bright when it doesnt seem like any illumination source is motivating it(?). ...

Wow, thats pretty smart about making the emitters 'Unseen by Camera', wouldn't have thought of that for a while! Think I'll experiment with this on the next project which will have no end of HVs.

For the moment this is coming together somewhat, it finally twigged about the sprites being too bright. Had the luminosity too high (as to organize lighting just for the sprites for this night scene, which is of an ambient type in real life, would have been way too complicated and render intensive on top of everything else).

Will post something soon when it goes on a bit further. Am using all objects silhouetted or Matte so it renders quickly and can gauge some idea on proportion. Have to get some new group of emitters going at the exact second the train passes, to push a whole lot of particles across the track in a swirling and eddying motion (dust and stray smoke).

Glendalough
07-22-2008, 08:24 PM
A test of the HVs with movement maybe getting OK. Surfacing needs work, used dented rather than turbulence but not quite right.

HV Test (http://www.jamespowersjr.com/jp-sr/vox.mov)

Glendalough
07-28-2008, 06:25 PM
Photoreal speed blur and some new hvs...

2MGs:
Speedblur Train (http://www.jamespowersjr.com/jp-sr//train2.mov)

Glendalough
07-28-2008, 07:15 PM
'Free Plugin' -Competition Hysteria

Glendalough
08-01-2008, 07:36 AM
Medium Compression H264 version of final entry:

ExpressH264(10mg).mov (http://www.jamespowersjr/jp-sr/ExpressH264(10mg).mov)

Glendalough
08-01-2008, 07:49 AM
Mistake above, right link here:

ExpressH264(10mg).mov (http://www.jamespowersjr.com/jp-sr/ExpressH264(10mg).mov)