Sven H wrote:Every clip that has been reframed with the input sizing controls in the color tab, will now behave incorrectly. But not only sizing is affected by this issue, but power windows aswell. Roto, masks, whatever are now completely offset by a certain percentage. It looks like that percentage is tied to the mismatch of the resolutions. Meaning if there is a height difference of about 80% then the masks will be off by 80% aswell.
Unpopular opinion:
don't do that.
My tactic for workflow in multiple resolutions is this:
1) export a flattened textless render in full-res for the entire show, using a visually-lossless codec like DNxHR 444 or ProRes 444.
2) create (let's say) three separate projects, each in its own aspect ratio.
3) import the flattened full-res file to each project, make the project settings identical to delivery (vertical for social media, pan/scan for 4x3, whatever you're doing) and reframe as necessary. Position text in a way that's optimized for each format.
4) within each project, add the timeline to the Render Queue. Now, using the Option ("3-dot") menu, Show All Projects, and you can kick off all the renders from a single project.
5) upload the final renders to wherever they need to go -- Broadcast, Snapchat, Facebook, YouTube, whatever. Done.
We've used this workflow for a couple of years now and it solves many problems. We avoid nested timelines for many problems (particularly audio), and we also avoid mixing aspects within the same project with the possible exception of documentaries, which are kind of a specialized case. Many, many documentaries use 1.33 archival footage, and it's an ongoing discussion as to whether we present that in the original aspect ratio, or blow it up for the new show aspect ratio, or if we go somewhere inbetween. I can make each argument convincingly, but it's a create call. We change sizing as a Timeline Sizing-only grade with Static Keyframes, and it works perfectly. If we had to deliver in multiple aspect ratios, it might be tricky but doable with some work.