I'm working on a project which is one of those 2.5D things where you have a massive BG comp and smaller comps layered in front with distance on the Z axis for parallax while panning. The camera moves around mostly on X & Y with some zooming to look at all the details. The catch is, it's in 8K
Is there some trick to determining just how big that BG comp needs to be to not lose quality when zooming? I'm working with high-res scans of art so can get any resolution I need, but the dimensions are starting to get into those incomprehensible numbers and I don't want any more than I need for obvious reasons. It's a lot of slow panning through a "multiplane" sort of world.
Any other workflow tips from those of you working in 8K? This is my first foray into stuff this large. (Thank goodness for proxies!)
8k resolution would lag hard on just a single video. But your making a complicated project. You would need a beast of a machine to work on that project without the slow down...And the slower you move the longer it takes to finish. I have unfinished projects for that very reason
Not that I've done much work in 8K, but whenever I have done anything that includes something exceeding that size, I break it down to smaller comps or images. What I mean by this is instead of having one 32K image, for instance, I would have four 8K comps or images instead. That way I can turn off, or delete the ones that are not needed at a given time on the timeline. Also, I would have images at resolutions that are multiples of 2 that can be faded in or out when zooming in or out. So, instead of scaling a 32K image down to 8K I would have an 8K version of that full image for when it is needed. The images would be parented to one master null for manipulation. That's just my approach to overcome the limits of my computer's capabilities.
So, basically, the simple approach is to take that one humongous image and split it into 8K pieces in Photoshop, then scale the original to 50%, split it to 8K pieces and repeat (scaling to 25% of the original, splitting to 8K pieces etc.) until the image itself is 8K or less.
Or, reversely done, you can scale the original image directly to the multiples-of-2 size that fills the 8K comp and create higher resolution versions as need arises.
As for calculations of the resolution needed, the only way I know how to do it is to determine what the largest zoom that you need is and calculate it from there. Although with the described method you really don't need to. You can just pull another closeup image (of size 2 to the power of X) or a set of them from the original image.
Again, just one possible approach. I don't know how well it applies to what you are doing.