Newbie here (to C*, not compositing, Inferno user). I am evaluating a demo copy, a couple of questions occured, any isnights appreciated;
1) When image sequences are loaded into RAM, I am surprised how much RAM is actually used. I had expected ~1Meg per NTSC frame, but it is dependent apparently on # of images and operartors as well, so RAM goes fast. Is there any way to a) request only composited image to be loaded into RAM, or portion of screen to render. I ask because the 512Megs goes very very fast.
A related q would be does this process than mean that if changes are made on the fly, only changed layers are updated so playback interaction is sped-up?
2) When cloning from one frame in a sequence to a different frame in a sequence, is it possible to see the source in one viewport and the destination in another (so that I can see both what I am sampling and where I am painting it)? Similarly, if I select a source other than my target image, is it possible to see that in another viewport as well?
It would be great to a) be able to see whatever tool you are using in a viewport in the other viewports even if it is not neceassarily affecting those directly (think keyer, paint, etc.) Also, viewing different frames in different viewports would help alot as well.
1) You want to do File>Render to RAM. You can set quality, size, frame step, etc and cache much more for full-speed playback.
2) You can look at anything you like in a viewport, and you can have multiple viewports open at the same time. On the left side of the middle command panel there is a drop-down to change the layout from one screen to a 2x viewport, a 4x viewport, or a 4x filmstrip-style viewport. Just change the layout, click on a viewport so it's active, double-click on the layer you want to look at in the Workspace, activate the other viewport, double-click the layer you want to look at there, and work in either of them.
______________________ Brian Demong Publisher, Combustion Instructor Mesmer Animation Labs
1) As Brian suggested, you can use "Render to RAM" to only render the current operator (i.e. the composited image) into memory. But, here's a little background on how combustion works - it uses a dynamic caching system that brings all elements in and out of RAM as needed when it renders. So, when you render a composite, combustion "saves" the intermediate steps automatically up to the defined limit of RAM (which you can set in prefs, by the way). Once that is full, it flushes intermediate caches depending on a host of factors, such as time to render, priority in the stack, etc. In the end, watching the RAM meter doesn't necessarily tell you what can fit in memory if you're using the cache - it will still end up playing the final composite from RAM if you let it continue caching as normal, with no intervention on your part, because the image currently in the viewport gets the highest priority.
The main reason it does this is for interactivity - if it flushed the intermediate stages immediately, then changes made downstream would have to re-render every time. It also makes a huge impact on our background rendering, saving it from having to make extra render passes unless actually required to do so.
So sometimes it looks like RAM is being "used up", but it's not a problem, because combustion will manage the RAM by itself to cache the animation you're trying to play back.
2) You can see any point of the tree in any viewport - just click in the viewport and double-click on the item you wish to place there. You can also click on the item and hit the "Send Up" button to place it in a viewport.