Editing scenario

[Richard Herd]

Bill’s convinced me about the databaseiness of X. So it seems pretty simple to assign database relationships to audio track/channels GUI.

[David Lawrence]

[Jeremy Garchow] “What’s most important about this is performance.”

Actually, I disagree. What’s most important about this is efficiency. When you have efficiency, you get performance for free. Throwing bigger, faster, hardware at bloated, inefficient code is usually the wrong solution.

[Jeremy Garchow] ” see you haven’t played around with Dynamic Link a bunch in Pr and AE yet?

Bloat is one thing, how it handles bloat is another.

FCP 10.0.4 (in my experience, perhaps not yours) handles the bloat much better.”

My experience on my machine is that it has the exact same bloat and compound clips problems as 10.0.1. It’s identical. Every change to a compound clip adds that compound clip’s data weight to the project. As you compound compounds, it increases geometrically. Why?

I haven’t played with Dynamic Link because I haven’t installed the full CS Suite yet. I’m waiting to get my license, then I want to uninstall CS5.5 and do a clean CS6 install. I’ll do some tests and let you know after I do.

I did test relinking. It works but it’s not as good as FCP7 and not anywhere near as good as FCPX. It may not work for your needs but you should test it and see. I imagine this is an area Adobe is getting plenty of feedback and will improve.

One thing I did test was recursive nesting, the kind of thing that brings FCPX to its knees. This was clearly a fake scenario but I was curious what would happen. I duplicated the project, then nested a sequence, then bladed it 25 times then nested that, then bladed that nest 25 times, etc, etc. About five levels deep.

No change in performance. Just as snappy as ever. I saved the project and checked the size against the original.

Original project was ~4.6MB, Bloat test project was ~7.2MB

Try that in FCPX and tell me what happens.

Now there was a penalty. When I reopened the project after saving it. it took about 15 minutes to reconnect media. I got a beach ball but instead of quitting, I checked the activity monitor. Sure enough, both cores on my lowly laptop were running full throttle and completely over-maxed. I just let it grind and eventually the project found everything and I was back in business. On a bigger, faster, machine, I’m sure it would have gone faster.

Again, try this in FCPX and watch what happens.

I really wonder sometimes if the FCPX project database architecture is fundamentally unable to scale.

The bloat problem should be priority number one on the engineering radar, because it makes the program fundamentally unstable. Autosave trouble is just a byproduct of this.

These problems are the kind of thing that if left unaddressed, will cripple FCPX’s chances in the major leagues.

They really need to fix this.

_______________________
David Lawrence
art~media~design~research
propaganda.com
publicmattersgroup.com
facebook.com/dlawrence
twitter.com/dhl

[Jim Giberti]

[David Lawrence] “My experience on my machine is that it has the exact same bloat and compound clips problems as 10.0.1. It’s identical. “

I’m just running 10.0.4. I didn’t want to update till I got a couple of projects out that were underway.

I’ve worked on two spots in 10.0.4 and find it amazingly slower, balky, freezing the skimmer for seconds at a time. If this is a performance update, I’m pretty unimpressed.

[David Lawrence]

[Jeremy Garchow] “Not a great way, or an unfamiliar way?”

Not a great way. For all the reasons we discussed in the thread about DAWs.

With music, an external frame-of-reference for time is essential. Tracks give the ability to hold this external time reference for multiple channels.

This is what Jim’s template accomplishes via multiple secondaries locked to frame one. It’s very clever but it should be built into the program. Multiple primaries could achieve the same thing.

_______________________
David Lawrence
art~media~design~research
propaganda.com
publicmattersgroup.com
facebook.com/dlawrence
twitter.com/dhl

[Alex Hawkins]

[Jeremy Garchow] ”
This is completely dorky and boring, but when I test an NLE, I’ll do an edit, archive the project to LTO, restore the LTO, and reconnect the media on a different machine. I will also start a project on a laptop/local storage, move it to our networked storage in the office, and reconnect.

Without getting too crazy, this represents typical workflow scenarios. Whatever NLE does this the best will factor heavily in to our decision.”

Mmmm boring, yes. Dorky, no.

A very good work practice to get into quite honestly. Should be more of it…

…er starting with us I believe. ahem!

Alex Hawkins
Canberra, Australia

[Steve Connor]

[Jim Giberti] “I’m just running 10.0.4. I didn’t want to update till I got a couple of projects out that were underway.

I’ve worked on two spots in 10.0.4 and find it amazingly slower, balky, freezing the skimmer for seconds at a time. If this is a performance update, I’m pretty unimpressed.
“

I’m not getting this at all, 10.04 is very fast on my system with no freezing and almost no crashes. Have you tried trashing prefs etc? Also you could try setting up a new user on your system and testing 10.04 with that, it just solved a problem I had with importing XDcam EX material on my system.

Steve Connor
“Sometimes it’s fun to poke an angry bear with a stickl”
Adrenalin Television

[Jeremy Garchow]

[David Lawrence] “My experience on my machine is that it has the exact same bloat and compound clips problems as 10.0.1. It’s identical. Every change to a compound clip adds that compound clip’s data weight to the project. As you compound compounds, it increases geometrically. Why?”

Hmm. Not me.

The bloat still happens, but it doesn’t sky rocket quite like before.

[David Lawrence] “I did test relinking. It works but it’s not as good as FCP7 and not anywhere near as good as FCPX. It may not work for your needs but you should test it and see.”

I have been testing and seeing.

We have lots and lots of projects on our SAN. These damn AVCHD and DSLR cameras rename all the files to the same thing. 0000.mts in the case of AVCHD, and 0000.mov in the case of DSLR (DSLR is a little better, but still not good). Let’s say you have 15 cards of AVCHD, that’s 15 clips of 0000.MTS, and that’s only one project.

Now, since Pr doesn’t give you a file path when reconnecting, you can really screw this up HUGELY if you happen to choose the wrong file path, or choose the wrong files accidentally. While we try to be as buttoned up as possible, I am not perfect, my coworkers are not perfect. Computers, on the other hand, are GREAT at managing massive quantities of data. Pr could be better here.

[David Lawrence] “One thing I did test was recursive nesting, the kind of thing that brings FCPX to its knees. This was clearly a fake scenario but I was curious what would happen. I duplicated the project, then nested a sequence, then bladed it 25 times then nested that, then bladed that nest 25 times, etc, etc. About five levels deep.”

What kind of clip and how long was it? Apples to Apples.

[David Lawrence] “Now there was a penalty. When I reopened the project after saving it. it took about 15 minutes to reconnect media. I got a beach ball but instead of quitting, I checked the activity monitor. Sure enough, both cores on my lowly laptop were running full throttle and completely over-maxed. I just let it grind and eventually the project found everything and I was back in business. On a bigger, faster, machine, I’m sure it would have gone faster.
“

This happens with X and 7 too. Bigger longer projects take longer to load.

[David Lawrence] “Autosave trouble is just a byproduct of this.”

Well, if you watch what’s happening, it autosaving after every blade. So not only does the performance go down, the autosave saves too much and grinds it all up.

Again, I have ran basically the same tests as you just did, but I don’t know how long your clip was and what format.

I made a 2 hour timeline of all kinds of things, compounded it, and split it up.

The Project went from 10MB to 200 MBs, it used to get in to the GBs. Once all the autosaves happened, it was running just fine.

I do test this stuff, and 10.0.4 seems to be running leaner.

There’s no question that X needs help here. I am not worried about the end of the world and the database is DOA. Are their kinks? Yep, just like any NLE, especially new ones.

Jeremy

[Jeremy Garchow]

[Alex Hawkins] “Mmmm boring, yes. Dorky, no.

A very good work practice to get into quite honestly. Should be more of it…

…er starting with us I believe. ahem!”

Everyone’s workflow is different. Perhaps you might not need this test, I don’t know.

Our projects come back, and when they come back, we need them to work in fairly short order.

It’s why I put the boring tests in so that I know when the boring stuff needs to happen and fast, it can happen. And fast.

[Jeremy Garchow]

[Simon Ubsdell] “I guess this is true but implementing favorites as a tool for maintaining persistent I or O means that you effectively throw away the other benefits of having favorites in the first place – because now they can only be used for trying to retain IO rather than for their opresumably original purpose.”

Since you had to read the manual about three point edits, I’m curious, how much real time have you put in trying out favorites? I’m not trying to slam, I am sincerely curious on how much work you’ve put in to X’s organizational structure? I’ve put some work in to it, and I am still learning a lot.

Yes, you have to learn to use and optimize favorites. When you do, you will realize that perhaps keywords are better for your long ranges to hold on to, and favorites are shorter more temporary ranges (sound familiar?).

Keywords also have the benefit of overlap.

It’s very easy to click on a favorite, have that range selected, and then modify it for further editing either in the Browser, or even the timeline.

I also use markers in FCPX, to hold on to pertinent parts. The nice part about markers, is that they travel in the edit, i.e., they move to the timeline, in to the index, and are also text searchable.

So before you need to favorite a favorite, I think that you might need to figure out the myriad of ways that the Eevnt structure works, rather than complain that it’s broken. And if you don’t care for it, that’s OK too.

By reading the Smoke manual, I don’t care for it either. After trying a demo, I think I might like it.

I’m not saying X is perfect, it is far from it. And I STILL can’t put it in to normal, everyday production work.

[Simon Ubsdell]

[Jeremy Garchow] “I’m curious, how much real time have you put in trying out favorites?”

Well, I’ve just been editing an interview with Mr Complacent Certainty of Inflexible Opinions Inc. and I’ve already keyworded the thing to death, but there’s only so far you can go without keywording every single phoneme, unless you’re Bill Davis.

So I’ve started to use favorites to narrow down my selections and I want to start editing with one particular favorited clip that runs: “When all said and done, favorites are the answer to the problem of persistent ins and out. Period. There’s no argument about it.” And another favorited clip that goes: “OMFs and EDLs are a waste of time. Any idiot can see that.”

I’ve filtered my clips so I’m only seeing favorites. Clearly the point of favorites is to narrow down your selections below the level of keywords so this seems the strategy that Apple have intended for us to use at this point. Let’s try and remember that favorites are by definition sections of clips that I am planning to use within the edit, though the chances are very high that they will need further editing, but Apple know all about that don’t they, so it’s cool.

So, the bit I want from the first clip is that bit that goes: “favorites are”, so in order to get a persistent IO I have to unfavorite the clip because you can’t have a favorite within the range of another favorite and …

Ooops, what’s happened?

I can’t see my clip anymore. It’s disappeared. I’m going to have to hunt it down again to refavorite just the part that I want.

A minute or so later I’m back on track. Great.

Now I need to go through that whole process again because I want just the words: “are a waste of time” from my second clip.

Obviously I’m missing something again. But it does seem to me that given that favoriting is a process for narrowing down selections, not being able conveniently to select within those selections without actually destroying the first selection is, you could almost say, entirely self-defeating.

That’s what I’ve learned about the ineffable magnificence of the FCPX organizational structure.

But other than that, you’re absolutely right – everything is for the best in the best of all possible worlds.

[Jeremy Garchow] ” I am sincerely curious on how much work you’ve put in to X’s organizational structure? I’ve put some work in to it, and I am still learning a lot.”

Life is short and clients will insist on paying me exorbitant amounts of money to work for them so I tend not to have time to waste on exploring stuff I broadly understand already. I genuinely think I learnt as much as I need to learn (for now) within about the first three days – this really isn’t rocket science, much though some folks would try and have us believe that it is. Anyone who was familiar with iMovie before FCPX appeared will have similarly had no difficulty getting their head around the “organizational structure”. The apparent novelty is only in the eyes of those who weren’t already up to speed with iMovie.

Simon Ubsdell
http://www.tokyo-uk.com