I was asked some questions in another forum so have decided to post some of the answers here in case anyone is interested.
What about asset management?
so with asset management, kinda not really, you could, just make everything talk to a database that pulls info from generation as well as other apps, we actually have an awesome automatic time tracker I didn't show that gets info from generation and other apps, and keeps track of how much time you spent on each shot, but it would be a similar system. we are a "comp" based system. so almost all the tools are based off of that idea. not every shot needs 3d not every shot needs mocha or sytheyes. but every shot needs a comp, so that is our primary asset. maybe the easiest way is to explain how we collect assets for backup. At the end of a show, we have a project Archive script. now all this works because everything is in the right place because it was all put there by a script. so we know where to find everything. And what is important and not. so the script goes through and grabs, most the 3d stuff, not renders and not sims. everything else doesn't take up that much space by comparison anyway, then it gets to the comps. it gets the final comp, and looks inside it to see what was actually used in the final shot. goes through and collects all those assets, and puts them, if not there already, in a precomp folder for that shot. then the script makes a new comp with relative paths to them, that means all the test renders you did, get cleaned up for you. only if the render was in the final comp is it saved. now the 3d assets are still there if you wanted to go back because they don't take up much space, but not the renders. same with all the old versions of comp renders, essentially if it's not the final or in the final comp and it's a big image sequence. we dump it for backup. it can always be rendered again if needed. but not worth keeping. so as far as asset management, we know what's in a comp, and what 3d assets are in it. and where it's track came from. but if an artist updated a 3d asset, our system has no idea and it's up the the artist to let the other 3d person know it needs re rendered. we have tools that make it easier for mdd swapping and the what not. but more work could go into this area. but honestly, with a small group not a big deal for us.
what about bigger studios?
the biggest deal with generation is it doesn't scale well, because of not being able to have multiple people editing. it's an all or nothing. so if one is in editing nothing else can be changed. if they said, per shot editing that would go a very long way to making it scale to bigger groups.
what about two people working on the same shot?
as for having 3d and comp working on something at the same time, not a problem, the comp script gets newest comp, the 3d gets newest 3d, so both tools work, the issues you get into is one of them has to be on bottom. as far as playing back in the timeline. but you can easily click on the higher level and playback that it just wouldn't go to review. unless you made it it's own shot. for example, ogre look dev, as a shot, then you render a 360 turn of it. well then that would go to review. or even just a frame. but it's based on shots. so you'd just say, show setups or scene setups are shots. it really hasn't been a problem.
how long did this take?
as for development time. really hard to say, it's been me and another guy Rob Field, we got generation a couple of years ago. neither one of us really did scripting or python, I had done one successful python script before and it was basic. and needed a lot of help to make it work. We got generation, could see the potential. so decided we needed to learn python. that was a couple of years ago. so we started banging our heads against it, and over the couple years it evolved to what you see. actually there's more I didn't show, would have taken to long, but if we had to start from scratch and do it now, with the better python ability, it would probably take the two of us, 3-4 weeks. it's not as bad as people think and Rob and I aren't as good as some
the hard part for us was, just learning python while also learning the python connections to each app. that's why even in the video some of the gui's are easygui, aka tkinter and others are pyqt. we went down the easy gui path first. actually the no gui path first, but then learned easy gui, then learned pyqt.
why can't blackmagic do what you did?
the problem is, every pipeline is different and everyone swears by their setup. So if someone brings tools to market. Say we did, well generation is included now, so no cost there. but what would it have to cost as a separate suite of plugins? because then you are talking people getting it and not wanting to work the way it works and breaking it immediately which would then cause mounds of support issues. if blackmagic wrote everything we did. same thing. everyone cries because. "Why does it save to the precomp folder? wouldn't it be better if X" we have refined ours as it has developed and I'm sure we will keep doing so. almost every show. we find a new need, a new tool that would make life easier and write it. last week, we have a client that doesn't use the client review system. just texts, "all is good" so I made a "batch change status" script. pops up a gui. you select what the status is, what you want it to be. and it does it for the whole project. 30 mins to write. and now we have it. But you do that times everyone's shows it becomes a pretty big undertaking and a support nightmare. We were just discussing how we would have to write/ rewrite all ours to make it more customizeable to any one's fancy. We could make it a lot more changeable. but at the end of the day it's the balance between, it all works because everything is done by it. it knows where everything goes, so where to find it. and If humans get involved and want to keep changing things or not following the rules. then it all falls apart.
that's why epp, was trying to be so open ended. but I don't think anyone knew what to do with it. We had already started down our road so never even looked at them.
so I get the problem and difficulty for blackmagic. the more custmized, the better the tool is, aka ours, but the less adoptable to everyone's needs. the more open it is, aka epp, the less useful it is to people.
What they can do is fix the current issues. the muilti user problems. and some other's I won't get into here. make it solid and polished.
Then they could decide what road they want to go down. if it was me it would be a slightly more customizeable our system
I'm clearly biased though, but it just works, and works well, I think that gets the small to medium guys up and going. and if you can't script it your self, then just take what you can get and stick to what someone else thought was a proper pipeline. the medium to bigger places are going to have scripters anyway.
hope this is helpful,
Chris
Fri Aug 14, 2015 8:54 pm
.
