[James Roth]

DAVE: “By the way, how much of this project is green screen? What’s the estimated running time when finished?”

This project is 1 hour long. 20 minutes of it features silhouettes walking around interacting w/ titles, animations, gesturing, emoting, etc. Another 20 minutes of it is 3D animation made in the same art style (big red 3D letters, 3D skyscrapers in red). Another 20 minutes of it is simple filler animations (2D text in 3D space and the like). The silhouettes are mainly used in this type of stark red/white art style, I’m focusing on setting up some killer animations to make the whole thing pop.

This shoot was a test for the real shoot w/ a professional model. I’m considering backlighting the model this time for extra fidelity, lighting the cyc wall better to eliminate shadows and noise, and using tracking markers. This is all shot in slo-mo.

My #1 concern at this point is…

SHADOWS!
Getting great-looking shadows and/or reflections is really crucial to grounding the silhouettes in the whiteroom and making sure the art style is “minimalist,” not just “plain.” I’ve downloaded and used VC Reflect and it works great, but due to how reflections actually work, the model’s feet are spaced apart on the Z axis in real life, which VC reflect doesn’t mimic since all it does is flip the 2D footage. Her feet and the “shadow feet” have a lot of problems joining up when she walks from the side, when she spins on her heels, etc. (Look at the last image to get an idea of how reflections would be tricky). I’m really using the skew and angle features in VC reflect to the max, but even keyframing the shadow angle/skew angle so the feet always meet up with her shadow mid-walk ends up in a very unnatural and wobbly looking reflection. Add in perspective shots like looking down on the models feet at a 45-degree angle and it gets really tricky, really fast. If someone here could suggest a way to create this look, or something similar, I would be forever grateful.

Plain moving disc shadows (animated masks and/or blurred shape layers) aren’t really cutting it, especially when she walks, because her legs/feet don’t affect that type of shadow. As you can see, the above shadows are “reactive,” that is, the ground shadow darkens when feet move over it. Coupled with the reflect this really sells the 2D/3D hybrid look on an otherwise plain 2D silhouette.

Any suggestions on how to eliminate the green-screen noise that accumulates by the model’s shoes is also greatly appreciated. Since her shoes are touching the green screen directly, I end up with black areas that inevitably end up as grain in the final key. An even bigger problem is spot shadows appearing between her shoes as she walks, as the studio shadows criss-cross one another. Any effect or adjustment I could make to eliminate this would be great. And also, why the heck does exporting to H.264 make my whites and reds so dark (also shown below).

Thanks guys!

[James Roth]

Thanks for the replies guys!

DAVE: “May we additionally assume that you took great pains to shoot a properly-lit, properly-set-up chroma key to give yourself the best chance of success? It would be a shame to shoot in 6K and work with just a wrinkly ol’ green bedsheet and a couple of borrowed lights.”

Yes, the green screen portion is not really the issue. I have taken all the steps to fix Keylight in AE including talking with Adobe customer support, rolling back to previous versions of After Effects, fiddling with GPU settings, etc. I think my system just crashes because it’s 6K. Exporting it as a still image sequence out of Premiere Pro takes eons. What I want to know is if there’s a way to convert the 6K so that it doesn’t crash After Effects and retains 6K quality; this is because I can easily zoom in the 6K and retain the quality. I’ve keyed using Keylight before, it’s this RED footage that Keylight doesn’t like. If I export it as a TIFF or PNG sequence, won’t I lose the ability to zoom in and retain complete fidelity? I could just zoom in beforehand in Premiere and hope I don’t cut too much footage of her walking close to the sides of the frame. I quickly realized I should have put Hollywood-style tracking markers on the green screen but hindsight is 20/20 I guess. Maybe on the next shoot.

DAVE: “For something that simple, an animated mask would do to make a shadow. Pretty easy.”

How exactly? Masking out a spot shadow under the model’s feet and tracking it across the frame? Maybe I’m missing something but I can’t see that looking that great, especially on shots like the mode’s feet where she’s walking and I’d have to match two shadows to each foot as she walks.

DAVE: “This is why you should pull the key in AE. Keylight’s good, and you have additional tools like Refine Matte, which help with details like hair.”

Premiere Pro’s Ultra Key has about sliders to refine the Transparency, Highlight, Shadow, Choke, Midpoint, etc. I’d definitely love to get Keylight working but in the meantime how should I handle my workflow? Key in Premiere, zoom it in for max scale, and export as TIFF sequence? That would take days of nonstop rendering with all the footage I shot. I could export as H.264, but I just want to make sure I’m not messing with the quality since this footage is going through AE a few times and then back into Premiere for the final edit; I don’t want all that shuttling to result in a loss of quality, esp. seeing as exporting this 6K footage is resulting in darkened whites and washed out colors as shown above.

What is VC reflect? I’m also not familiar with what Richard means when he says to duplicate the silhouette and use Track Camera on top. What I’m really interested in is finding a way to eliminate that grainy shadow that occurs under the model’s shoes as she walks; the green screen just gets too dark to key it out properly and short of having tiny gnomes with lights under her shoes on the next shoot I’m not sure how to clean up those shadows without destroying some other part of the silhouette (like the hair) with aggressive use of the sliders.

[James Roth]

Hmm, sounds complex, I’ll definitely try a lot of these options out. Funnily enough the pillars of that bridge sit on land and not water, the real headache is somehow creating an accurate silhouette of the landmass and skylines on both sides of the bridge. The model is standalone so it’s like making the Brooklyn Bridge without New York in the background. The pillars don’t have to sit on land (I can make an abstract shot) but the skyline in the background would probably ground several shots in the video much better than stark white. At the moment I’m using heavy fog to mask the absence of the city on either side. Continuing experiments, I’ll keep you guys posted, thanks!

[James Roth]

Thanks I’ll check those out!

[James Roth]

Thank you for your reply Ted! I tried some experiments with E3D but they didn’t turn out well, and I don’t own any 3D standalone programs, so I decided to try CC Sphere, this is the result so far…

The image wrapped around the CC Sphere is HUGE, 10800×5400 pixels. It withstands the extreme zoom, starting with the globe just a small dot on the frame, and then it zooms way close on the map, which works up to a point, but then it obviously gets pixellated when we’re at the level of states and territories.

The main problem I have is that I can’t get the animation to look more exciting. I need the stark white background so I can do an insane zoom from Radius: 10 to Radius: 4000 but the lighting and the smoothness of the sphere makes it look blah. I added some lens flare (correction: lenz flaer) and it helped a bit but I’m stumped on how to make the globe look more professional. I need to maintain this white/gray color scheme. CC Sphere will support me animating the wraparound, so I’m going to make a country fill in with a paint spill effect while I zoom in. I’m thinking of stacking parented images and masks for that “Extreme Zoom” effect but because the globe itself is shaded I don’t exactly know how to match the lighting when I make the cut to a flat image. Making the paint spill work with a 3D model instead of CC Sphere would be tricky I think; I have limited experience with 3D models, I have Element 3D but that’s it and no shader packs. I want the globe to look HUGE and the lighting to interact with the globe in a more pleasing way. Since CC Sphere is not a 3D image I can’t fiddle with camera lenses or anything like that. Any ideas? :\

[James Roth]

It seems to be a bug. I made a composition with similar (but not exactly similar) settings and there was no slowdown. I even reused the assets from the tutorial. I just want to make sure that my CPU is interacting with AE in a healthy way and there is no hardware problem on my end…

[James Roth]

windows_os.png

ae_version.png

I don’t know how many times I have to say it; I have not overclocked my CPU. I have not changed any factory settings in my PC whatsoever or voided any warranties.

[James Roth]

Okay, let’s define “slow”.

What does it mean to you: “not instantaneous”? An hour per frame? Someplace in-between?

Dave LaRonde
Promotion Producer
KGAN (CBS) & KFXA (Fox) Cedar Rapids, IA

—————————————————————————————————-

After Effects refuses to respond to any button presses, I put my cursor over buttons and there is no highlight, it stays like that indefinitely until I remove the camera layer. I can maybe manage to display a drop-down menu but it takes a good 30 seconds to do so. Windows says the program is not responding. Even hitting Undo takes 30 seconds before AE takes the camera layer off.

I didn’t OC my CPU, it’s factory unlocked. I have not touched any CPU settings, all I did was put it onto the PC and install Windows 10. I got it from NewEgg, it’s not a second-hand CPU or anything. If you’re saying it’s faulty that’s another issue altogether…

[James Roth]

10283_lesson11.rar.zip

I haven’t overclocked my CPU. My PC is a custom build and only a few weeks old but running solid. I think the CPU can go a bit over 4.0 GHz even at factory defaults when pushed to its limits, which obviously AE is doing for some reason. I have a great aftermarket cooler that could push an OC to at least 4.5 GHz (although people have clocked higher using the same CPU and cooler) and my system is well-ventilated so it’s not a heat issue.

The RAR of my Project is above, straight from the tutorial files. It includes a sample movie of what the finished product should look like.

As far as I can tell, AE is having trouble using cameras in conjunction with distant lights. As I increase the (z) value closer and closer to the tutorial-specified value of (z)-2000 AE gets increasingly annoyed and slows down until I literally have to force quit the program.

[James Roth]

I backtracked by deleting both light layers, and disabling the Casts Shadows in Material Options. Left with just the background image and text (in 3D) I added a camera. No problems so far. I added the first and second lights, went to Light settings and entered the values. No problem so far. I enabled Casts Shadows. Again, no problems. With both lights shining, the camera in place, and all the settings enabled, I tuned the final value; camera position. I adjust the last value, the (z) axis of the light. The slowdown happened again, as soon as I adjusted the (z) axis of the first light, the key light.

The location I am moving it to is (x)955, (y)-102, (z)-2000 as per the book instructions. Entering the last value, (z)-2000, immediately creates extreme lag. Is it my hardware not being powerful enough to handle a light at (z)-2000, or is it a bug?

My CPU is maxing out at 4.13 GHz even though task manager reads that it’s working at 15% capacity. ??