Frank Engel wrote:The short version is that when you use a MediaIn node to pull media from the underlying timeline, what you are getting is post-edit-page footage at timeline resolution rather than original media resolution - in other words, you are working with the "output" of the edit page, already scaled/cropped to the timeline.
This is partially incorrect and somewhat misleading.
Regarding "footage at timeline resolution":
- This is true in the case of a Fusion Clip and Adjustment Clip composition, and a Fusion Effect/Transition. It is not true for a single clip composition, which one could argue is the 'default' composition type as it's the type you get if you put the playhead over a media clip and go to the Fusion page.
The MediaIn node of a single clip composition sees the original media at original resolution, and has access to its metadata.
Regarding "post-edit-page":
- Maybe you're referring to the image scaling as being 'post edit', but this is rather misleading as the Fusion page is always before the Edit page, not after. The Fusion page will not see any transformations, adjustments or effects made/added in the Edit page - with the sole exception of the Lens Correction effect, which does occur before Fusion.
The scaling to timeline resolution isn't considered part of the Edit page, but is rather part of the source media loading that Resolve does automatically. This graphic from the manual (p1352) shows the order of operations:
In order to get Edit and/or Color page effects to apply before Fusion - so that their impact is seen by a Fusion composition - nesting has to be involved. That is, making a Compound or Fusion Clip out of the source clip(s), then placing a Fusion composition on that container clip. Any Edit/Color adjustments done inside that container clip will then be seen by a Fusion composition placed on the container clip.
In contrast, a Loader node brings in media directly to the Fusion composition, thus you are working with the "original" media
Yes, the Loader node sees original media in all cases - but unfortunately it only works for images and image sequences. In the case of single-frame images (not sequences), the Loader node is vastly preferable to MediaIn because MediaIn nodes can't cache in RAM.
and not with something already manipulated by the edit page.
But again, all edit page effects on this timeline, except Lens Correction, will not influence a Fusion MediaIn node. Nesting needs to be used if one wants to change that order of operations.