Felix Telfer - Digital Media Studio

Part 2 of 2. Pretty much the rest of the playblasts I can find around the lab. Chronologically, the destruction playblasts come after these.

Part 1 of 2 - not sure if this will still be considered but it’s from ages ago, decided I should probably upload it all.

These are a few tests from the window scene, I was attempting to make a smoke plume for the windows - I couldnt quite nail it with forces, uniform fields were only working to a certain extent…

Something went wrong somewhere…

More fun and games…. about 8 hours ago. Procrastination?

Fun and games with fluids

Oh… and another motion tracking test. Couldn’t get it to track this one in 3d, which was concerning - I think the motion blur in faster frames throws it off, which means either hand-tracking, high frame rates or high shutter speeds.

Will probably just fake most of the handheld-ness with wiggle expressions in post… but maybe a few for real.

Something thats been annoying me in the background -DoF in post.

Turns out it’s just a custom shader you can assign to a new renderlayer - then you can just render it out and use it as a source for your lens blur effect.

As a note - it seems to need a simple matte choker in there to shrink the DoF layer a bit - otherwise there are white outlines in the more distant areas… not too difficult.

Sweet… now to plan what’s happening on paper… less than two weeks to go, methinks I’m gonna need a ‘farm of some sort =/

Mooooore smoke sims.

Sorry about the sims being in different sizes, I’m getting each one from a separate workstation so the perspective window is always slightly different if i’ve modified settings and forgot to hide them in the attr. editor.

I saw a quick demo of someone else’s on Vimeo using surfaces to emit from, so the last test deals with that. I think it’s the right direction to go for, as it seems to create better smoke formations, and more of them.

Will be simming some more tonight to see how it works with different attributes.

This is after up-rezing the sim - and simply playblasting it.

I can’t seem to work out why, but when pumped through mental ray the sim is somewhat smoothed out - I’m not sure if it’s a lighting, texturing, or rendering parameter issue so I’m just dealing in playblasts at the moment.

The main critique I’ve had is that it seems to fast - it doesn’t dissipate properly at the end, sort of like it’s moving into some sort of gradient mask of transparency.

The next tests play around with bouyancy, twist, swirl, and fuel to try and see if it helps at all.

So - Maya fluids. I’ve been working on these for a while but they take forever to playblast, so they’re all on the backburner while I do other stuff.

Ttextured smoke has been a big issue.

Earlier smoke tests using textured opacity maps turned out to be somewhat unrealistic - it seems to be good for stills but not motion, as the map is static and easily visible as a noise/turbulence overlay.

Here’s the first test after all of that - still way too smooth, but getting there.

Another attempt - I’m putting off doing scifi as this is far more interesting!

Slightly better but still jumpy.

From looking at the tracking details this time in MatchMover, it seems as though it’s mostly the bottom details that are screwing it up - the base of the windows isn’t defined well enough for it to track it nicely so the tracking points just jump everywhere. The top points are pretty solid which results in a wierd rotating camera effect.

Looking at the keyframes for the camera as well, it seems to be calculating that it’s moving around a bit - probably a side-effect of the above problem.

The solution - trying to track the window panes instead of the concrete this time - they seem to be defined a lot more clearly, especially at the corners.

Another motion tracking test, this time with 3D tracking on 8 points on a 3D model.

Interestingly enough it’s not quite as good in some respects, as it’s basing it on such a small amount of points - the data is pretty noisy.

That being said, I could simply create some geometry and have it appear correctly in perspective (the camera was oriented at the right angle this time, instead of having to orient the geometry like in the 2D tracking test)

I think it’s a matter of getting footage with a definite high contrast ratio on key 3D points - I’ll have to try again tomorrow to get some new footage with a Z7.

Onto scifi… sigh (hurr hurr)

Motiontracking test using Alias MatchMover.

Refs for next, 3D motion track:

Window size:

2.07m x 1.21m

Block Size:

.97 x .46m

There’s also a reeeally crude HDRI lighting setup in there, simply using a frame from the video footage. I’ve used a plane to cut off light from the base of the sphere temporarily. I reeeeeeally need to get a reflective sphere for a proper env map!

Some reference for smoke - some kind of junkyard fire in the US.