-
It’s All Connected — Clip Relationships and The Magnetic Timeline Paradigm
-
[alban egger] “But the abiltiy to have different layers in different modes (ripple or not) on different tracks and the benefits of the magnetic timeline where every clip is dragged towards the center and beginning of your primary storyline cannot be reproduced easily.”
I think many of us who dislike the magnetic timeline would say this moving to the center behavior is a problem, not a benefit.
The reason I don’t like it is because it completely ruins spatial context for understanding how the ripple edit is affecting clips downstream. Ripple itself isn’t the problem – I use it all the time. Rippling in a traditional timeline is easy to understand because the spatial cues are consistent. Time is absolute and external so rippling on either the right or left side of a clip always pushes or pulls program up or down from the left.
With the magnetic timeline, spatial cues are relative and inconsistent. Rippling from the clip right pushes or pulls program on the left, but rippling from the clip left pulls time zero into the program. This might make sense in a design for small screen devices like tablets or phones, but on a large monitor I think it’s unnecessary and disorienting. I think it’s a design mistake.
[alban egger] “it comes down to one’s preferences. A timeline is constructed of both, clips in relation to each other and cips that have to have a relation to a certain time in the timeline. So IMO none of the two paradigms are perfect, therefore it depends on your preferences and projects as to what works.”
I agree. I truly believe there’s a UI middle ground we haven’t seen yet. Done right, it would take from the best of both the traditional and magnetic worlds.
[alban egger] “In the magnetic timeline you can try more without risking to mess with your overall project, because the cliprelations, which basically are your story, stay, even when you break the timerelation.”
In FCP7, I select sequence, hit command-d, and easily experiment on my duplicate sequence without risk to my overall project. I just don’t see what the big deal is.
_______________________
David Lawrence
art~media~design~research
propaganda.com
publicmattersgroup.com
facebook.com/dlawrence
twitter.com/dhl -
[David Lawrence] “Did the EMC let you move clips horizontally?”
No. As I recall, and it’s been over 10 years, there was only 1 video track, it’s just that if you pasted in new clips they “covered” but did not erase underlying clips, until you selected them and merged them down. If you “uncovered” a clip you could of course then move it around the timeline like any other clip. This feature gave you the ability to “move now and decide later” which I have assumed is one of the benefits of the magnetic timeline.
Herb Sevush
Zebra Productions
—————————
nothin’ attached to nothin’
“Deciding the spine is the process of editing” F. Bieberkopf -
[David Lawrence] “Even though we perceive audio and video through different senses, they are both time-based media, and are represented and manipulated and very similar ways in a 2D graphical environment. While they each have unique properties and possibilities, in general, I don’t believe they need to be treated as differently as many people think.”
The difference is that in video the top layer covers anything below it, while with audio multiple layers merge but do not cover. This seems like a significant difference, operationally, to me.
Herb Sevush
Zebra Productions
—————————
nothin’ attached to nothin’
“Deciding the spine is the process of editing” F. Bieberkopf -
[David Lawrence] “I unlink and relink audio and video so frequently, I have a dedicated button on my mouse for just that function.”
In this we both agree. One of the wonders of this forum is learning about all the different workflows various editors are using; it’s an antidote to the often solipsistic nature of the work. I am amazed when I hear of editors who rarely unlink their audio, I have been assured they are amazed at those of us who consider it like playing piano with oven mitts on. Since I’ve always globally unlinked everything after creating my assembly, I didn’t even realize that their was an”un-link” toggle button in the timeline till Jeremy told me about it.
[David Lawrence] “having sync markers makes it easy to bring it back if needed”
Which is why I won’t consider an NLE that doesn’t have this feature, leaving both X and PPro out in the cold for me for now.
Herb Sevush
Zebra Productions
—————————
nothin’ attached to nothin’
“Deciding the spine is the process of editing” F. Bieberkopf -
[Herb Sevush] “The difference is that in video the top layer covers anything below it, while with audio multiple layers merge but do not cover. This seems like a significant difference, operationally, to me.”
True, but that’s more an agreed upon convention than a technical and representational requirement. When we stack audio tracks, the convention is to mix rather than cover. When we stack video tracks, the convention is to cover (or composite) rather than mix.
In both cases, time-based media is represented by 2D objects with duration corresponding to object length. We edit by manipulating these objects in space. Even though the media types are different, I believe the objects that represent them can share many tools and behaviors.
_______________________
David Lawrence
art~media~design~research
propaganda.com
publicmattersgroup.com
facebook.com/dlawrence
twitter.com/dhl -
[David Lawrence] “True, but that’s more an agreed upon convention than a technical and representational requirement. When we stack audio tracks, the convention is to mix rather than cover. When we stack video tracks, the convention is to cover (or composite) rather than mix.
In both cases, time-based media is represented by 2D objects with duration corresponding to object length. We edit by manipulating these objects in space. Even though the media types are different, I believe the objects that represent them can share many tools and behaviors.”
Agreed, although I will just note that the convention is universal – I can’t think of any NLE that doesn’t work this way, although as per this discussion DAWs treat audio differently. But you are correct that you could set the default opacity of all video above track 1 to 50% for instance, that we don’t is convention.
Herb Sevush
Zebra Productions
—————————
nothin’ attached to nothin’
“Deciding the spine is the process of editing” F. Bieberkopf -
[Herb Sevush] “Agreed, although I will just note that the convention is universal – I can’t think of any NLE that doesn’t work this way”
Yes, and I’d also note that these conventions usually become universal for good reasons. For example, there may be others but I can only think of one NLE where the default edit mode for the timeline is ripple.
_______________________
David Lawrence
art~media~design~research
propaganda.com
publicmattersgroup.com
facebook.com/dlawrence
twitter.com/dhl -
[David Lawrence] “[Herb Sevush] “The difference is that in video the top layer covers anything below it, while with audio multiple layers merge but do not cover. This seems like a significant difference, operationally, to me.”
True, but that’s more an agreed upon convention than a technical and representational requirement. When we stack audio tracks, the convention is to mix rather than cover. When we stack video tracks, the convention is to cover (or composite) rather than mix.”
My thoughts will help to clarify some of what Franz asked me to elaborate on. The way FCP7 stacks video layers is how Fairlight stacks audio layers on a single track. When I look at an NLE, I see a video track with multiple layers. Top layer plays but you can create an interaction with a lower layer by adjusting opacity, dissolves or keying.
Audio clip stacking on a single track works the same in Fairlight although the only interaction is to cross fade to lower layers. So to me an NLE is a single video track with clip stacking so why is then audio not able to clip stack? So Herb, David is right that they are both just media objects. It is the convention in NLEs to treat them differently.
So imagine how powerful and complex an edit system would be if you could have multiple video tracks with clip stacking plus multiple audio tracks with clip stacking and then patching to enable downstream mixing of elements. No need for video nests. Of course you would need to have track based FX as well as clip based FX.
Think of how grading and opacity layering could be much more powerfully applied to a bunch of clips on a track, interacting with another track via track based processing with complicated clip based layering with stacked clips on a track. This could be a way to get the power of node based control with track based spatial layout.
I think X makes a mistake by failing to see the power of track based spatial arrangement by exploiting the power of clip stacking on a track combined with track based processing. Taking the mixing desk analogy for processing and organisation and applying it to the NLE seems obvious to me. Add to that metadata management and I see that a track based paradigm will be far more powerful than any trackless system that only relies on software relationships that are not spatially presented and only clip based processing.
The grouping that Franz wanted to know more about is the ability to temporarily select clips on specific tracks (single or multiple) via temporary in/out markers. In Fairlight that is called range and can be toggled on and off. There is also the ability to permanently group objects which is more like the way metadata and roles works. This is the mode that I find least useful for the same reason as David L. That is that relationships change. The changes happen all the time during and edit so predefining them without a way to quickly unlink or relink is limiting, particularly in dialog construction where sync comes and goes.
-
[Franz Bieberkopf] “One aspect to all the track conception discussions which has been very evident but not as directly discussed is the different meanings that tracks have for audio and video. Audio tracks are very much modelled on the audio mixer (though only Premiere Pro seems to have taken this analogy in a more developed direction). Video tracks seemed to have arrived as an analogous counterpart to audio tracks, also partially from the multi-deck broadcast & ganged film flatbed analogy, and partially from the Photoshop layers analogy. Video tracks have characteristics of both Photoshop layers and the organizational grid of a mixer, without quite being either. “
As you say this it occurs to me that I don’t think of video tracks as tracks at all, but as layers, and usually discuss them aloud that way as well. I don’t think I ever use the layer metaphor for audio.
-
[David Lawrence] “1) If a traditional timeline had intelligent groups and a clip layering/overlaping mode such that the editorial results of a clip collision scenario were similar to FCPX, would that give you the benefits of the magnetic timeline?”
Some, but not all.
More specifically, you’d get collision overwrite avoidance in all cases, and reduced manual pre-maneuver selection requirements in some, but I think that many here would argue that magnetic reflow is one of the primary benefits of the magnetic timeline, and consistent and predictable reflow depends on having every single relationship in the timeline defined (in other words, having every clip connected to something).
Sidebar: I think that FCPX’s “ripple” is different enough compared to our traditional understanding of “ripple” that it’s confusing to use the same term — that’s why I’m calling what FCPX does “reflowing” instead of “rippling.”
Before I get to question 2 about benefits, there’s a premise in the argument to address:
[David Lawrence] “One way to think of clip connections is as a special type of grouping. For example, in Premiere Pro it’s possible to select any number of clips on the timeline, group them, and they’ll move together when you move any one of them.”
I’m not sure that clip connections are a special type of grouping.
With FCPX’s clip connections, one clip is connected to another in terms of the latter; one clip is a parent, and the other clip is a child of that parent (anchored at a specific point on each side of the relationship). Moving the parent affects both the parent and the child because the child inherits from the parent. Moving the child affects only the child and not the parent because inheritance is one-way.
With Pr’s grouped clips, the clips are peers. Their anchoring is based on a common externality (absolute time); there is no single pivot point for that linkage. Once the clip is grouped, they are fundamentally treated as a single object, not as two objects with a hierarchical relationship.
The presence in FCPX of that movable connection anchor point means that reflow is contextual, and editorial maneuvers on one side of the anchor will have a different result than maneuvers on the other side. More on the ramifications of this difference in a moment.
[David Lawrence] “2) If “FCPX is “aware” of clip relationships in a way that traditional NLEs are not.”, how does that benefit the editor?”
I’ve gotten some pushback in related threads on my apparent obsession with clip relationships, so I want to go into a bit more depth about what I think a clip relationship means.
Franz pointed out in this thread some of the various different kinds of clip relationships. It’s a really great and well-thought out post, but I think it’s grounded in traditional/Legendary thinking.
What do you gain by parenting clips so that a specific point in time in of the child is related to a specific point in time of the parent? Context! It’s not important what timecode of a child links to what timecode of a parent; it’s important what’s happening in each clip at those timecodes. Correctly defining the linkage point of a clip connection is an editorial decision based on the content or meaning of both clips and the editor’s rational for defining one clip in relation to the other. This action encodes a little piece of meaning, a little piece of the editor’s motivation for making for that edit, that FCPX is designed to preserve through subsequent maneuvers.
There is no construct in a traditional timeline that’s anything like this. There is no way to connect a happening in one clip with a happening in another (thanks to FCPX, I can’t call them “events” like I wanted to). Context can be implied in a traditional timeline through an editor’s track organization, but it cannot be codified in the edit like this.
My answer to the question above about editor benefits is in Apple’s terminology. In FCPX, the primary storyline drives the edit. Secondary storylines and connected clips are linked to points in the primary storyline and follow along with those points as the storyline upon which they are based is changed.
The constructs in FCPX add a layer of abstraction to editorial. The editorial mechanism is very literally a storyline and not a timeline. When editing clips in, the editor makes a decision about whether that particular clip drives the story (influences the flow, is primary) — or whether it rides the story (is influenced by the flow of the story elements the clip is attached to, is secondary), as well as decisions about what events in the storyline come before and what come afterward. All editorial maneuvers, then, are either rearranging primary elements (which in turn drive the rearrangement of dependent secondary elements), promoting or demoting elements into or out of the primary storyline, or the assortment of traditional editorial operations like trims, slips, insertions, and deletions.
The benefit to the editor would be that the editor can concentrate on the story in terms of the story, without engaging in the editorial bookkeeping necessary to keep a traditional tracked timeline following a story. The challenge to a traditional editor is to stop thinking in terms of clip placement and start thinking in terms of structural relationship.
My variation on this question is this: what does this abstraction change? What drawbacks does it have? What control (if any) is lost to the editor with a magnetic timeline versus a fully manual tracked traditional timeline? What are the real ramifications (if any) of assuming a story-driven, A/B hierarchical edit?
Interestingly, there’s a place where this relationship model falls apart — the implied relationships between formally unrelated clips. Just as clips may have implied relationships in a traditional timeline by virtue of their placement, so can you end up with implied relationships in FCPX that you cannot codify with a single connection point.
When a connection defines the start or some intermediate point as the link between a parent and a child, the out point of the child is formally unrelated to whatever is underneath it in the primary. This relationship may be significant to the story, but there’s no way to encode it in the FCPX model; two connection anchors would confound the flow. FCPX preserves formally defined relationships during maneuvers, but the informally related / formally unrelated edits go along for the ride. This creates potential contextual conflicts that must be resolved with zero UI indication that such a conflict may exist. If you are depending on FCPX to manage your story for you, here is one case where it may fall short.
Walter Soyka
Principal & Designer at Keen Live
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
RenderBreak Blog – What I’m thinking when my workstation’s thinking
Creative Cow Forum Host: Live & Stage Events
Reply to this Discussion! Login or Sign Up