Positron Development (CVS & Win32 Release 0.1 Uploaded)

General Forums Blender and CG Discussions
41

it is true that for the Elephants Dream we at Orange could use another video sequence editor than Blender, if there was some reason to do so.

we actually used cinelerra a bit when doing early edits of the making of in October for the BConf"05 (LohnC did that little editing, he is reposinsible for the making of).

but as, especially with Peters patches now in CVS and will be in the 2.42 release, the Blender sequencer is robust enough for editing this short (and Peter’s 2hour long) movie. there is nothing principal in Blender code that hampers video editing functionality, i mean: if you dont work with 3d, you can just ignore the 3d view. but there surely are things to be improved in the code, and for a long time in the autumn i was actually looking at other codebases like the LTC framework that Jahshaka uses … and it would also be possible to make the sequencer just a UI to something like LiVES.

one strategy to develope larger / different things with Blender is forking, and in a way the 2.42 sequencer was developed in a fork (Peter’s private copy he needed in their Belgian studio to make a movie (but not 3d)). the Foundation supports that by giving projects cvs (in future also svn) repos etc. actually during 2005 Bassam here was using that fork, instead of BF, for working on the live edit.

but certainly there is still room for separate video editing apps also, and also in the open source / linux world there are now many cool libraries (like livido) and even frameworks / platforms that can be relied upon … and related to other areas like 3d, well, there is no limit to the amount of interesting things that could be fun.

so, good luck indeed and hopefully there is interesting code to look at soon :slight_smile:

~Toni

42

Here is a short list of some of the reasons why using Blender won’t work. Some could argue that some of these are not needed features, but here is where the decision has to be made. Do we start with the Blender codebase and then decide what features we can add to it, or do we start with the features we want, and build them on whatever codebase will work.

Core Features for Positron:

Central Data storage (all project and video data for the entire workplace is stored, or can be stored on a central server. Each workstation can run their own server if they want, but it must have the ability for multiple users to be working on the same project).

High Performance Scrubbing (Ever tried scrubbing through a MPEG2 video? FFmpeg is good, but still it’s scrub functions are slow compared to scrubbing through a image sequence like Positron does)

Render Farm Support (slave nodes work off of the Central Data Storage and render the projects automaticly).

No matter what you do, the NLE in blender will allways be “bolted” onto it. Think of this, when you output a video why do you have to hit “Do Sequence” and then “Anim”. That makes no sense for a NLE. NLE’s should automaticly render as needed, and then the user should have the option to output the file to disk as he feels fit. I can just see users wondering what on earth the hundreds of buttons in the buttons windows have to do with a NLE. Blender is hard enough to understand for a person learning 3D. You honestly think we should make a NLE person try and learn the whole interface just to learn how to edit a video?

Anyway, that’s my rant for today.

43

yep, video editing in Blender is currently practical mostly for people who use Blender for other reasons too.

LiVES also imports everything so that it makes it an image sequence, and it is great to have such immediate seek. and having a 160GB scratch disk in a video editing workstation, if you are working on several DVDs at the same time, is ok with modern computers. this could be of course an option in Blender sequencer too … and e.g. reuse code from LiVES or Positron for it (supposing positron will have a compatible license and reusable code :wink:

data storage can be done in basically any way with Blender, and there are many research-y projects about it going on in the dev wiki with occasional bf-committers discussions … things like DbBlender etc. and farms we obviously already use.

so basically i think you are wrong in this: “No matter what you do, the NLE in blender will allways be “bolted” onto it. Think of this, when you output a video why do you have to hit “Do Sequence” and then “Anim”. That makes no sense for a NLE.”

there is no reason to not enhance those things in Blender. but that can be my concern, and not yours :slight_smile:

~Toni

44

Hi!

Are there some testing builds around or still “just” version 1.0.0?

CUL :slight_smile:

45

To Non-Supporters: If you don’t support the project then why in the name of the alimighty are you even posting in this thread? It’s not really doing anything but discourage someone that’s trying to get some work done…

That being said, it would be REALLY nice to see a new version…

46

Probably because it’s far easier to criticise than to create. Good luck with your project.

47

To all those who are looking for a new version, it is comming.

Since the codebase has totally changed, the version should change as well. So, we’ll call this 0.1 (to make the critics happy). The old (1.0) version will be archived and removed from CVS eventualy.

0.1 will be out this thursday. It will work, although may not be all that useful. As it stands now, clips can be loaded, and displayed. Nodes can be created moved, and synced with the central server.

Up for tonight is the rendering pipeline, and GUI methods for connecting nodes. Up for Tuesday is a gui interface for modifying the properties on the nodes, and playback of the composite (as it stands now, clips can only be stepped through). And For Wednesday, probably a bit of polish on the gui, and creating releases. Thursday night (late central time) will see a update of the webpage, CVS commit, and upload of the offical Win32 build.

48

PassiveSmok,

Probably because it’s far easier to criticise than to create. Good luck with your project.

Well the critiques I have seen have been mostly by developers. I think most of it is due to the massive hype on the website (which apparently tbc++ didn’t do?) but a lack of original code. Another issue has been that critiques of Blender have been made which, on the surface don’t seem merited.

From tbc++ comments it sounds like he has some serious development that he is prepared to ‘take the wraps off of real soon now’. I’m sure once that happens, that there will be far less naysayers.

I, as antont (another Blender dev), wish him luck and look forward to playing with the code.

LetterRip

49

Because despite some benefits to a certain degree of integration, the apps still remain separate. If the industry wanted a mega app, you’d think demand would have spurred something by now. Even if only rumors.

Apparently we were having two separate debates, then. My arguments were about the current state of the industry. And currently 3D, compositing, and NLE are separate tasks. The software has some overlap–as I acknowledged in both of my posts–but full integration doesn’t seem to be popular. The industry appears to prefer that the software largely remains separate. I suggested some possible reasons for that.

I seriously think we’re just having two separate arguments…

So… I’ll switch over to your argument, for the sake of debate. :stuck_out_tongue:

----computer beeps switching debate… please wait… computer beeps—

The basic issue with such large-scale integration of all three parts of the process would be conflicts due to limited UI real-estate. Both screen space and hot keys are limited. You simply cannot optimize the interface enough for all three tasks (maybe even just two). Unless, that is, you make things so heavily context sensitive that they might as well be separate apps anyway, in which case there isn’t much point.

I suppose part of it may be an issue of attitude. I tend to look at things as, “If there’s no appreciable reason for them to be together, separate them.” I imagine there are those who say, “If there’s no appreciable reason for them to be apart, join them.”

I prefer firefox over mozilla, for instance, because I like my web browser, mail client, and web development applications separate. I also prefer separate printers and scanners, rather than a printer/scanner.

I like modularity, basically. And I think there are quite a few benefits that come from modularity.
For me the primary one is choice-without-bloat. If every web browser had integrated everything, then I would probably have three of them downloaded (all of them comparatively large) just to use a small portion of each. I just dislike the thought of that sort of thing. And when it comes to apps that cost money, it’s even worse (jumping back to the previous discussion for a moment: perhaps this is really what makes the industry prefer separate apps?).
Another one is that, for whatever reasons, specialized apps just tend to be better.
And lastly, as a developer I simply like the idea of several smaller code bases instead of one huge one. It would be easier to maintain and contribute to–less unwieldy.

Now, having said all that, I want to emphasize that I’m not arguing for the sequencer to be ripped out of Blender. I think that’s pointless. It’s already there, so we might as well leave it in. It would just take pointless effort to separate it.

What I’m arguing for is for there to be a separate (totally unrelated to Blender) app for NLE, that is designed specifically around that task.

50

This raises an interesting point.

In the past the compositing and NLE apps were seperate Premeire and AE for example. FCP and Shake/Motion. However, more recently there has been a melding of those two. Simply because a NLE user has the ability comprehend how a composite should be and vice versa.

But that is not the case with NLE and 3d apps. Personaly I can make some fairly nice videos/movies if needed, and I did so for a job for several years. However, I cannot do 3d modeling worth beans. I’ve tried, but my skill is limited to simple objects. My use of Blender has been limited to making logos for the videos I create.

That’s where I think apps like Nuke are headed. Sure they can do 3d, but you won’t find skinning, advanced NURBS, caustics, bones, and the list goes on. What Nuke’s 3d is basicly for is creating simple logos and animating them.

That being said these next few releases of Positron are going to be mostly centered around a compositing view (think combustion). Why? Because in essence a NLE is a compositor where the placement of the nodes on the y accsess determines the linking. A compositor with auto-linking.

51

As long as Positron can work as a vanilla normal NLE app, I’ll be happy. IMO, I’ve seen too many apps out there that have all kinds of crazy effects work but are just impractical to use to simply edit some video footage to music with fades and cuts and titles and stuff. So, I get slightly paranoid with the talk of nodes and compositing… :wink:

As an aside, I agree with the people that mentioned that having too many apps in one can cause problems as far as hotkeys and simplicity. I was playing around with Animanium a little while ago, and it is an app specifically geared just for animation (like how Wings3D is just for modeling), and it was interesting to see just how easy it was to get started and just animate.

Then again, perhaps the “extreme context sensitive” wouldn’t have to imply seperate apps per se. I think some amount of increase consolidation of hotkeys and consistancy of interface would help. Like why does LMB select in trees instead of RMB? Why doesn’t GKey let you reorder? Inconsistancies like that actually makes different parts of Blender seem more like separate apps. Perhaps eventually Blender will become more modularized where you could choose not to install/load certain parts of it, but perhaps there is still more room to make things simpler without being quite as specialized as seperate apps like Wings or Animanium…

Anyway… looking forward to Thursday to see new Positron ver. :slight_smile:
Shawn

52

CVS is up.
Win32 Release is uploaded.

Before checking into any of these, please listen to this:

Post your questions and feedback!

53

congrats on getting your release out,

ah python, goody :slight_smile:

LetterRip

54

Looking forward to trying it out…
Podcast didnt have audio?
Downloaded positron how do I start it? When I open positron it just crashes?
Nvidia 5200 fx
Pentium 4 2.8
1 gig ram
python 2.4 installed

55

I’m using quicktime 7 and it plays fine. Make sure you are using Quicktime or VLC to play it.

As a quick rundown (this is explained in more detail in the podcast). Download and install MySQL Server from www.mysql.com. And modify the settings in positron.cfg to reflect your setup (password and server etc). The rootdir should point to a folder where Positron will keep it’s media data.

But like I said, the podcast explains all that.

56

podcast is working now hmmm?? thanks for the myslq tip checking it out

57

Just got done researching what we can use to optimize the inner loop, and I ran across PyInline. Exactly what I was looking for. Now we can directly insert C code into the python code right where we need it. None of this module stuff. PyInline should do it all.

58

BTW, if you can goto Start->Run

and type cmd

And the

cd directory you extracted python to

and type:

python.exe

The info it spits out might be nice to see.

Thanks

59

yeah this is a big news, i’m going to download and test

60

Got it to work. :smiley:

So far I’ve only succeded in importing mpg2 files with the jpeg compression.

BTW hitting the cancel button in the import window will close the window - and I’ll have to restart positron to attempt to import another file.

I guess that the node view is for compositiong? I’m getting quite curious on how you edit window will be - with a conventional timeline - or something like the node view?

Anyway, this is really great work!! :smiley:

Tin2tin