[Terry Coolidge]

Hi Kevin,

Thanks for your thorough and thoughtful reply.

“is the clip you are working on edited at all, or is it just a single clip straight off the camera?”

I believe it was shot as 24p (stopping and starting the camera for multiple takes, I’d say), captured to disk as one long segment, and then split into separate clips when exporting out for me. I have 20 similar shots, and they seem to vary in their pulldown removal cadence. Some are WWSSW, some are SWWWS, some are WSSWW, etc. I don’t think any of the individual clips is “edited” if I understand what you mean. Each individual clip appears to be a single shot segment. From what I’m seeing, I am fairly confident that the pulldown removal cadence is consistent throughout each clip. There don’t appear to be any edits within a single clip.

Out of 20 clips, AE is able to provide a 3:2 pulldown removal guess for 17 of them. The other three give me an error beep after trying to come up with a guess. I don’t know what “24Pa Pulldown” means.

Is there any reason to not trust AE’s guess? I’d love to think that I could just trust AE, and then avoid using the three problematic clips. That seems like the easiest (laziest?) solution.

I believe I understand what you’re telling me in your post, but I am curious about what I should do about framerate for my final comp. Should I choose 23.976, 24, or 29.97? My final output will need to be 29.97 (interlaced) as it will be played back on an NTSC device, but what should my comp be as I assemble everything? Does it make a difference as long as I handle the final render properly? I’ll force render output to be 29.97 interlaced, but can I make my workflow less problematic by the framerate I select for my working comp? I’m just as happy to have a “24p look” for the graphics I’m adding, but I’m sure it’s fine to have the smoother (more video-y) look of 29.97 interlaced for the graphics I add as well. Maybe that preference is what drives the decision, but I’ll only start expressing a preference if it truly doesn’t matter. My first priority is to make sure I don’t get interlacing goofiness going on as a result of the way I choose to handle these green-screen clips. I did post a sample of what one of the frames looks like in my comp:

https://gallery.mac.com/terrycoolidge/100100

(It looks like this no matter what framerate I set the comp to… 23.976, 24, or 29.97)

I guess I’ll have to hook up a camera, dump this to tape, and then play it back on a TV in order to see if it’s working, right? Can you tell from this still image whether or not I should raise or lower my expectations that all is well? You talked about checking to see if it “looks good.” Does this look good to you?

Thanks again for your help. Any further information/advice you can provide would be greatly appreciated.

– Terry

[Terry Coolidge]

Anyone ever find out what the deal was here? I’ve just run into the same trouble. Pasting numbers trying to fill the composition, and at some point (when there are too many characters, maybe?) the numbers just disappear. For me it happens before I’m even half-way down the comp.

I’m using a brand-new, 3.06GHz Intel Core 2 Duo iMac. Could this possibly be a processor issue? Graphics card issue? What else could it be? Hard to believe AE gets maxed out so quickly.

[Terry Coolidge]

Shape layers and per-character 3D are two CS3 that come to mind right away. Also the interaction between AE CS3’s 3D space and the Vanishing Point “filter” in Photoshop CS3 Extended. But this is where I defer to folks with more experience. What’s the opinion? If performance is better AND there are new features, then there doesn’t seem to be much of a debate. But my main question was about CS3 performance on a G5 (PowerPC) vs. AE 7.0.

[Terry Coolidge]

Hi Kevin,

Thanks for the good information. Although now you’ve introduced a new element to the equation. I thought I would follow-up with a couple of bits of information.

I have a dual 2.7GHz G5 with 4GB of RAM. I already have Nucleo… but not Nucleo Pro. When rendering, I am able to more fully utilize the two processors already because of Nucleo. I had already intended to purchase Nucleo Pro when upgrading to an 8-core machine, but maybe I will now consider purchasing Nucleo Pro before upgrading to CS3. Would that still be your recommendation given these details?

I will eventually be buying both Nucleo Pro and CS3 as well as upgrading my hardware. But the hardware step may need to come later than any software purchases. I’m mainly just trying to determine what makes sense right now. I may be able to swing things so I can get both Nucleo Pro and CS3 now, but my curiosity right now is about whether I should think of one as more helpful than the other with my current configuration. Ideally I would buy both, as it definitely sounds like this will be an improvement over how I am currently working. But if I had to choose one vs. the other right now, which seems more helpful? New features (and performance improvements?) of CS3, or workflow/performance improvements associated with Nucleo Pro?

Thanks again for your input. Much appreciated.

– Terry

[Terry Coolidge]

A follow-up question about using all 8 processors of an 8-core machine during rendering. I typically don’t render in the background. I do previews here and there obviously, but usually when I’m ready to do a final render I queue things up, hit “render,” and walk away. I’m now thinking about buying a new dual quad-core (8-core) MacPro, and I picture myself letting the new machine render using all 8 processors and having my current PowerMac G5 as a machine to do other work on while rendering on the MacPro. Are there reasons why I shouldn’t do this? The idea of having 8 processors and only using 3/4 of them come render time seems strange to me. But maybe you’re suggesting only using 6 so that you don’t cripple the system from being able to perform other tasks while rendering. A clarification would be appreciated.

I’m vacillating between configuring a new MacPro with 8GB of RAM vs. 16GB of RAM. I keep seeing the word “theoretically” used when talking about allocating memory for a multiprocessor rendering, and I’m wondering if the upgrade from 8GB to 16GB is worth the cost. The extra 8GB of memory is only going to impact performance when launching a multiprocessor rendering, correct? I’m not going to see substantial benefits when I’m not rendering, am I? Is there some way of quantifying how 2GB of memory per processor would perform vs. only 1GB of memory per processor? Would double the memory significantly reduce render times over the course of an entire rendering, or does it mainly just help here and there during rendering as AE loads up the elements to render? I’m sure it depends on the project, but can you provide some sort of real-world example to help convey the benefits of extra RAM? 8GB (or 1GB per processor) seems like a reasonable amount, but if twice that is really going to make a huge difference, then I think I would want to consider that.

Thanks.

[Terry Coolidge]

Hi bogiesan,

I wish it was that simple. My 3D comp has layers that rotate and stuff, so it is not just a 2D movie now. I’ll try to explain more clearly what I’m doing. I’m recreating one of those tri-sided rotating billboards. The billboard is made up of a bunch of vertical “stripes” that are actually triangular prisms. Different imagery on the three sides. When the panels rotate, I want to be able to see the dimensionality of them that is retained in a 3D comp. I’ve basically built one of these in AE’s 3D space by how I’ve positioned my 3D layers. I’m compositing this synthetic, 3-dimensional billboard on top of a static background plate. I totally need to maintain the 3D nature of my 3D comp, so basic corner pinning will not cut it.

[Terry Coolidge]

Let’s say I don’t…

😉

However, even if I did know specific dimensions, this still wouldn’t help me match the perspective since I would still need to play with 3D camera position, orientation, focal length, etc. etc. etc. I’m basically trying to match a square to a square. So the trick is getting the camera in the right spot to look at my 3D comp square so that it perfectly matches the square in my still photo (background plate).

[Terry Coolidge]

Every layer in AE has a “Blend Mode” setting that is set to “Normal” by default. Simply change the blend mode on a layer to one of the “burn” options. If you’re trying to specifically recreate the Burn tool from Photoshop, then you probably want to create some sort of grayscale (black and white) animation going on within the layer that you set to “burn.” Maybe try a white solid layer and use the “write-on” or “stroke” effect to create black brush strokes that then affect the underlying layer(s).

[Terry Coolidge]

You know, I started off this project by using the Write-On effect, but for some reason I switched to the Stroke effect. I’m trying to remember why I switched. Hmmmmm… Makes me wonder why there are two such similar effects that behave quite differently in how they achieve their effect. I wonder if there will ever be a time when I would want to use Stroke instead of Write-On. My guess is that I’ll discover that time inadvertently right when I’m up against a tight deadline.

😉

Oh well, it’s all better now. Thanks for the prompt response, Dave. Much appreciated!

🙂

[Terry Coolidge]

But my Brush Position is determined by a mask shape, not keyframes. The only keyframes are on the “End” property of the Stroke effect. Two keyframes… 0% and 100%… that’s it. Is there a way to convert the mask shape to key frames and then do the “Rove Over Time” conversion that you’re talking about? Or is there some other way to smooth out the way the Stroke effect moves along the mask shape?