Hi!
I’m developing a noded base compositing app.
It’s in a very early stage of devepment, but I think it could be
a good complement to blender’s compositor.
It’s tested on linux (ubuntu 8.10), but it should be portable with minor work.
You can get it here: http://ramenhdr.sourceforge.net/
I’d like any feedback on things you like / dislike, future ideas and
possible ways of inter-operation between my app and blender.
Of course, contributions are mostly welcome.
Thank you.
Est.
No Windows version? 
Lol. But seriously, it looks like a great app so far! And about complementing Blender’s compositor, I agree. I glad to see a good looking open source video app being developed.
Ok your expecting this question so I’ll ask, tongue in cheek half joking, not a flame 
Why not just advance blenders node compositing rather than create another app with another set of dependencies and another code base to maintain?. Learning blenders code first? Lack of blender documentation for developers? No access to developers for questions? Better design?
How many of the nodes that you have implemented so far have a blender equivalent? How many are unique to compliment blenders nodes?
I notice it’s KDE based, I personally don’t want to bloat my Gnome based Ubuntu Karmic with KDE libraries for one app.
And how on earth will it compliment blenders compositor, I mean really? Isn’t that said to deflect the crit I just made.
Criticism out of the way, I admire coders like yourself and I bet there is nothing more rewarding than creating your own app and for Linux as well.
I genuinely wish you every success.
est77_, looking good. The open source community can really benefit from this. Good roto and wire removal tools would make you a hero. They will come in handy for project Mango, the open vfx movie after Durian. I’m glad you’re setting your sights high, like Shake and Nuke. Will it have an image viewer, similar to this? http://www.tweakfilms.com/news/rv-release
Check out what this guy is doing. I’m not sure if he’ll releases it open source.
Good luck, stay encouraged, we’re rooting for you.
The main purpose of the open source blender films is to develop blender and prove it as a production tool, so if roto and wire removal are needed for Mango as well as anything else, even if ramenhdr might offer it, one might expect the blender devs to code what is needed into blender in this case?
Sure there is other open source software used in the process like Gimp but separate compositing tools outside of blender’s own rather than code the additions?
Thanks for the efforts! But, I have to agree with “yellow” in that I think it would be better to focus on developing Blender’s existing nodes system. Unless you hope to develop an app that will work with other apps as well, etc.
When the core developers start focusing more on pushing Blender’s internal compositing nodes they will likely catch up to wherever you may have gotten with your independent efforts. That could potentially mean a lot of wasted time for you.
Also, focusing on developing Blender’s nodes will bring lots of expert help from more experienced Blender devs. That would save a tremendous amount of time. A great example of this process would be Farsthary’s Volumetric work: http://farsthary.wordpress.com/
Check out his blog from the very beginning to see how things unfolded and the amount of support he got from the community and the developers.
Good luck!
It’s possible that they can copy that part of the code, and other parts, from the software. Ton said that’s how they got DPX code, they copied it from Cinepaint.
http://www.bigbuckbunny.org/index.php/our-first-film-print/
http://www.cinepaint.org/
Ugh. What do you need the ASL dependency for? That thing is a pain to build. Is there really any advantage to using ASL instead of boost?
Same goes for TBB. Would it be possible to achieve the same results with normal multi-threading code or something like OpenMP?
The app looks very nice, but it has a LOT of dependencies, many of which are not going to be commonly found on any of the mainline distros.
You might find this recent discussion about threading in node graphs on the blender development list interesting: http://lists.blender.org/pipermail/bf-committers/2009-August/024130.html
I completely agree about the roto and wire removal tools. Masking in Blender is a pain right now. Perhaps it’s because I’m not as experienced as many of the other users. Either way, I agree that decent masking tools are a must.
Based on the dependencies, its not KDE based, it just uses the QT libraries. When things are actually KDE based (like KDEnlive), they force lots of background programs to run which then stay running after the application is closed… QT is a reasonable requirement…
It’s not just you, vfxjunkie. Remember that thread you started, June 08, about Hanger No. 5?
http://blenderartists.org/forum/showthread.php?t=126740
Nathan Matsuda said he used Shake and the main thing that Blender’s compositor is lacking are good roto and paint tools.
Lol, I certainly do remember that thread. And one thing I’ve started to notice about many people that use Blender for their visual effects oftentimes use a separate app for compositing. Sometimes they’ll use Shake and sometimes they’ll use AE. BTW, what is the most popular compositing program right now? I guess you could say industry standard. I mean from what I’ve heard Shake is dead and I think Fusion just released a new version. Last time I checked, Shake was industry standard.
When Apple stopped developing Shake, a lot of houses went looking for a replacement. Even though some will still use Shake, the one used by a lot of houses, node based, is Nuke. Digital Domain( of course ), ILM just got a site license and Weta. I think Tippett Studios, and Rhythm and Hues, also. I hear SPI has it’s own in house software, Socrato, but I wouldn’t be surprised if they have some Nuke seats too. Many smaller shops, also.
Thanks for the comments!
I really need a more complete website and more docs about the program.
First, some history:
I work as a freelance compositor. As I know C++, over the years I made a
series of command line tools to complement my work. Some are based on
my code and other on external libs like Motion2D, the klt tracker, …
At some point it became inconvenient to use them together so I wrote a
minimal node editor based in FLTK.
Some months ago I started a rewrite using QT and had the idea of open sourcing it,
to get help for development, porting, documentation, … so it can become a complete
compositor.
The best things right now are the optimization to process only the needed part of each
image in the composite, the caching of result images in ram, and ease of writting new nodes.
I think the design is flexible and there is a posibility to override graph traversal, interface display and
user interaction at every node.
The app is portable. It’s only I don’t have a windows machine to build and test.
I’ll work on reducing the dependencies. I’m using ASL for the range based algorithms,
any_regular and poly. I look forward to use the property model and layout libraries.
I think I can copy these files inside the source code. Same for Loki, AGG, Vigra and
some other libs. Intel TBB is included in modern linux distros and I don’t see it as a problem.
Conditional compilation of nodes should also reduce the dependencies. FFTW is needed by
the tonemap node. I’ll also work on that.
No KDE libraries needed.
The roto tool is being rewritten. The previous version was a prototype. Expect bezier shapes with
inside/outside feather and motion blur.
About blender’s compositor, I used it for some jobs and I like it. But it’s limited in many areas.
I see it more as a tool to merge render passes and apply some post-processing than as a complete compositing solution. Also, for people only interested in compositing maybe learning blender is too much.
After effects is the standard for motion graphics work. Nuke is very good and also very expensive.
Shops here in Spain keep using Shake. In advertising, Discreet’s Flame, Smoke, … are most used.
Est.
How good is Fusion? It looks pretty advanced from what I can see (lol from an amateur standpoint).
QT well no problem then.
i understand the difference, thought it was kde based, you say it’s not, I understand the difference and how KDE apps or (Gnome for that matter) share common libraries across numerous applications and how pressure is put on for an developer to use those core libraries rather than add extra dependencies.
You mention kdenlive, Kino for example (Gnome) not using GStreamer, Diva doing so and not actually achieving very much. As GStreamer was inferior at the time. But whatever.
Man, that’s a bit specific isn’t it Not really the same as comparing two apps with potentially very similar purpose and share code? seriously? If the code was that similar why not just write the blender nodes needed.
in vfx/commercials the trend is definitely towards nuke. its the only commercial node based compositing app that runs on all 3 major platforms used in production (osx/linux/windows), offers an easy switch from existing shake users, and has a bunch of killer tools.
re esteban working on his own app vs assiting with blenders, I’m in 2 minds about it. yes it’d be nice if he helped with blender, but then again nicholas bishop worked on his own sculpting app before working on sculpt mode in blender. might be useful for esteban to work out some issues, methodologies, try out stuff with a clean slate first, then later maybe try his hand at blender dev. also it strikes me that with 2.5 being in such a state of flux, its probably not the best time for someone to dive in unless they already have a good working knowledge of blender, and where its headed.
choice is always a good thing, and great to see someone taking a stab at this on the oss/linux side. there’s a definite hole in the market here, jahshaka is little more than crashy vapourware and promises.
@pixel master
I work with the guy who’s developing vfxdesktop. and I doubt he will put it out opensource. for several reasons, but I could be wrong.
agree, btw do you work a hsoc?
@est_77
Im glad that you have rotoscoping in consider, because there’s no really good workflow in blender for rotoscoping and that should be one of the easiest thing to do in a vfx suite. So im currently looking into creating a own software where you can rotoscope and have vfxdesktop kind of management.
most important with rotoscope is that you can not only render alpha mattes, but exporting the splines for importing in flame and other suites.
Yes but Nicholas I don’t think had heard of blender at the time of Sharpsculpt and it was users such as myself who built Nicks app regularly to test it and posted on his forum at the time how great it would be to add sculpting to blender and had he considered it and it was fantastic news when he said he would, shelving his own app and concentrating on blender. I’d guess that is because he saw more future in blender. But I’m surmising and don’t really want to suggest what Nick’s reasons were.
We did the same with Stephane with Freestyle used to build it from svn and suggested Stephane approach Blender Dev’s, they were somewhat cool about the whole thing comparing Freestyle to some post pro filters, all went quite for about a year until it was announced as a GSOC project. Thanks to Maxime and other devs now. The choice to integrate Freestyle rather than recode the flaky app that Freestyle was as a stand alone.
But this is a different situation the original poster does know of blender, blender already has a compositor (of sorts and has a very long way to go to compare with Nuke) but creating a seperate app is opted for?
But good luck to est_77, no doubt he has considered the options and respect to his choice.
yep, tho currently freelancing at the mill.