[Steve Bentley]

Ah, right: tailoring it to get good numbers. That would never happen in this day and age would it? You’ve been watching too much “fake” news! ( I couldn’t resist)

I’ve always been interested in why AE is as slow as it is. Don’t get me wrong, I’m a dyed in the wool evangelist, but compared to Nuke, Fusion, Discreet, and Sabre, its kinda pokey, both in rendering out and in getting about the interface.

Is this a function of legacy code that is just too hard to circumvent without “starting again”? Is it remnants of old Mac architecture left over from COSA days? Is it Adobe trying to keep it compatible with other Adobe products? By 2019 it should just be faster. Or is it just development dollar availability? Or perhaps developer apathy.

If you compare AE to the above 4 package in the following areas, it’s just fallen behind (or maybe always was)

Real time play back – not really real time if you have to wait for caching. When you get into hundreds of elements even the big boys have to cache. But on a simple BG, matte and element shot, we still have to wait for the cache.

Larger frame sizes seem to make the time lag exponentially worse. The big 4 do get slower but not precipitously so.

Scrolling down through the comp’s layers has never been quick and I’ve never understood why. (Pshop has the same issue)

Even moving mask points around. This used to be much worse and it has improved. You might not even notice a lag now in AE2019. Until, that is, you’ve spent time on any of the above 4 packages.

And added-in code or expressions runs interminably slow – you might not notice it until you have a ton of recursive stuff. (this may be a function of the compiler(s) Adobe has chosen)

All of this is not me complaining (ok maybe a little) but more a query as to why I always feel I’m driving a VWbug (that I love dearly) within a pack of race cars.

[Steve Bentley]

Hey Cassius,
I was feeling the same way but wanted some feedback to confirm.
I didn’t understand your comment:
But scoring higher on a benchmark that they created seemed a cheap tactic.

It really depends how you use AE – one should build a box based on that. As you say, using Element or not, raytracing or not, wanting to multitask or not, will all inform that decision. I have a feeling this was more of a marketing technique or a way to write off a hardware purchase under the guise of a youtube review, than a true “lets find the best solution”.

These days its pretty hard to find someone who is using AE and NOT using a 3D package in concert. If the shop is compartmentalized enough so that AE would be the only app that lives on a particular box – an “AE station” as it were – then that shop is probably using something else (Nuke, Sabre, Discreet etc)

The cost of that box they spec’d also seemed a little stratospheric for what you get. I mentioned that running the benchmark on our machines beat their numbers because we built ours for just over $3k CAN$! (so what’s that, $1.95 USD?) Bringing that box into Canada would bring it up into the $8K range CAN$. I could fabricobble a nice little render farm for that price that would wreak havoc with that benchmark.

[Steve Bentley]

Can you try a revolved object instead?

[Steve Bentley]

I remember something like this. I think the top and bottom of cylinders aren’t actually attached the way you think they should be. I just made one editable and beveled it with the usual Create Tools bevel and got wonky results. So if that’s the case once it’s poly’s, it’s anyone’s guess what’s going to happen parametrically. But normally, deforms work better on polys because the parametric objects (except the cube or plane) may have different geometry depending on a number of factors that can change, until the are made editable and fixed with a certain number of polys in fixed positions.

[Steve Bentley]

Its it well ordered? Like, all the nuts are called NutXXX or something? You can use the search parameters in C4D object manager to only show you those objects and then delete them all at once. Or you could order them by name in the structure list and do it there.
The search and structure list will ignore hierarchy so you don’t have to open every folded assembly.

[Steve Bentley]

Remember too. 10mm doesn’t mean much unless you know the sensor/gate size.

[Steve Bentley]

That looks like the back part of the hair is a separate layer, blended with the main body of the hair with a soft edge mask and then use something like ripple or wave or mesh distort or reshape to get the effect you want. This will only really work where, in the area of blending the waving layer with the static layer is all one color or so black it doesn’t show. You could do it with reshape and have the area of blend not move while the outer edges move and that would work if there is detail in your hair at the blend point.

[Steve Bentley]

Hmm, I’m wondering if the export path is the culprit. We have SW here and we use the STL export from SW to bring the model into C4D. The ASM format has all the information for say, a sphere to be completely smooth but doesn’t have any geometry perse. That format is really meant for G-code (or similar) generation, or for the SW viewer to generate real time viewings of the parts and cross sections.
While there must be geometry there or C4D would just not see it, (and of course that you can get it smooth in some renders) I’m wondering if adding in certain new objects interrupts the tessellation order of the SWASM in C4D.
Have you got the original SW file? Can you try STL? The great thing about STL is that you can tessellate it in SW to the mesh density you are looking for (that’s the “options” button just before you hit save).

[Steve Bentley]

Its all relative. With only one object, the position’s move is always relative to its parent. With only one object the “parent” is the world. So the Z will always be the worlds Z.
If, after you have rotated the object, instead of setting the position value with a node, just drag it along the Zmanually and notice the values of the object in coordinates box. Its not just the Z that changes because its no longer moving in the “worlds” Z alone.

But you can change the orientation of the cube’s axis. Use the Enable Axis modification. But this gets tricky because you will probably want to change it back later.

When it comes to rotating things it’s a good idea to use nulls any way (and usually Parent, sub parent, child). Because it’s so easy to get stuck with something called Gimbal Lock. Once you turn an object in two directions, there will be a point were when you turn the third axis the object will either not rotate or not rotate the way you expect – and there’s no way out of the problem. But if you keep each of the objects in the parent,subparent,child heirarchy and dedicate each one to a different axis, it will never bite you. It also lets you (especially with camera moves) tweak any animation subtly without upsetting the overall motion because the children are just along for the ride and will do what the parents do. This way the children can add their own motion to the parent’s.

[Steve Bentley]

And choosing your layout from the layout list (top right of interface) doesn’t shoot the panels to their right place?
I wasn’t having this issue on the PC in <=R18 but am now in R19,20. So I have to choose the correct layout each time. I assumed this was because I’m constantly traveling and having different second monitors in each location. I can hear C4D grumbling “Oh what now!?” each time I plug in a new screen.