Developer Meetings/20071027

From K-3D

Jump to: navigation, search

To Attend

  • Meeting Time: 20071027 1700 UTC
  • Timezone Information: www.timeanddate.com
  • Meeting Protocol: XMPP (Jabber) Conference
  • Meeting Location: k-3d@conference.jabber.org

Agenda

  1. Action item followup
    • I have received a response from SFC, which you can all read at /SFC Response. Tshead 22:20, 25 Oct 2007 (MDT)
    • I've completed the scripted plugin capability, our Freeze and OGRE script-writers should start bugging me with questions. Tshead 22:20, 25 Oct 2007 (MDT)
    • I've posted K-3D 0.7 release files at http://www.k-3d.com/~tshead/k3d. They will be formally released once I have some feedback from experienced users. Tshead 22:20, 25 Oct 2007 (MDT)
    • Has everyone updated their personal wiki page? Tshead 11:58, 20 Oct 2007 (MDT)
    • Everyone should try testing Joe's demo movies. Tshead 11:58, 20 Oct 2007 (MDT)
  2. Animation: Where to go from the current state? Curve editor or other priorities?
    • Based on our first meeting I think the consensus was drop it for now. Tshead 11:58, 20 Oct 2007 (MDT)
  3. Proposal: create an "announce" mailing list.
  4. Proposal: Switch to trac (or something else) for bug-tracking. The sf.net system has a lot of visual noise and is slow for surfing. Besides it lacks of some features. (like global rss)
  5. Proposal: Develop slide presentations that can be used at local LUG meetings, conferences, etc.
  6. Proposal: Post releases to a concrete list of venues.
  7. Proposal: Improve website.
  8. Proposal: Ensure everyone has t-shirts / caps / stuff.
    • This might be funded with our SoC money. Tshead 12:00, 20 Oct 2007 (MDT)
  9. Proposal: Attend open-source conferences.
    • This might be funded with our SoC money. Tshead 12:00, 20 Oct 2007 (MDT)
  10. Proposal: Attend SIGGRAPH.
    • This might be funded with our SoC money. Tshead 12:00, 20 Oct 2007 (MDT)
  11. Proposal: Developer sprints.
  12. Proposal: Bug-days.

Action Items

  • Contact Debian about SPI (Bart)
  • Solve Joaquìns reset mesh question
  • Drop win9* support?
  • Test 0.7 on Vista (Bart)
  • Test win32 test-release (Joaquìn)
  • Work on the screencast transcoding (Joe, Rene and Bart)
  • Sort ModelingRoadmap according to priority (Joe and Joaquìn)

Discussion

(19:09:09) Bart: OK, as far as I'm concerned, we can start
(19:09:13) joe: do we have, an "agenda"
(19:09:20) joe: yes, is it that time!
(19:09:45) Bart: http://k-3d.org/wiki/Developer_Meetings/20071027 is the agenda, let's tackle it top-to-bottom
(19:09:45) Anders Dahnielson: http://www.k-3d.org/wiki/Developer_Meetings/20071027 ('http://www.k-3d.org/wiki/Developer_Meetings/20071027')
(19:10:20) joe: ah yes, actually, I just opened it. Thanks though.
(19:10:38) Bart: Any comments on the SFC response?
(19:11:18) Anders Dahnielson: It looks like they're a bit overwhelmed with projects showing intereset
(19:11:53) Anders Dahnielson: So I guess it depends on how quickly we want to be members of such an organization
(19:12:02) Anders Dahnielson: compared to the other alternative
(19:12:14) joe: Its likely that we can wait a bit.
(19:12:30) Bart: That depends on Google. Tim has asked them to hold on to our money, so if they can wait, we can wait :)
(19:12:36) joe: yeah
(19:13:05) Anders Dahnielson: we'll, they have hold my AdSense money for a year and a half ;-)
(19:13:13) Bart: :)
(19:13:37) Bart: OK, second point is the scripting. 
(19:14:08) joaduo: just one thing
(19:14:15) joaduo: http://www.spi-inc.org/projects ('http://www.spi-inc.org/projects')
(19:14:30) joaduo: does anyone knows people on those projects
(19:14:32) joaduo: ?
(19:14:48) joaduo: we still dont have any information about spi
(19:15:13) Bart: ah, ok
(19:15:45) joaduo: any idea?
(19:15:49) Bart: I got a mail from a Belgian Debian person before the mentor summit, I could try contacting him
(19:16:05) Anders Dahnielson: This was posted to the previous meeting agenda: http://www.spi-inc.org/treasurer/associated-project-howto.html ('http://www.spi-inc.org/treasurer/associated-project-howto.html')
(19:16:17) joaduo: ok, i will go on asking on forums and mailing list
(19:16:39) Anders Dahnielson: in regards with the application process
(19:16:56) Anders Dahnielson: information that we had email SFC about
(19:17:24) joaduo: ok, so, lets move to the next topic
(19:17:35) Bart: OK. Scripting.
(19:17:59) Bart: There are a lot of new and interesting ways to load scripts now
(19:18:23) Bart: I'll ask Tim about how I should use that in the OGRE exporter
(19:19:07) Bart: Anyone interested in scripting should definitely check out http://k-3d.org/wiki/Scripting
(19:19:45) joaduo: i am working on the freeze script
(19:20:13) joaduo: so i ask
 in_mesh = node.__getattr__("output_mesh")
 mesh = k3d.dynamic_cast(frozen_mesh, "imesh_storage").reset_mesh()
 mesh.reset(in_mesh)
(19:20:19) joaduo: that doesnt work
(19:20:27) joaduo: how do i reset a mesh?
(19:21:00) Bart: ok, I can't answer that without research :)
(19:21:12) joaduo: ok, so dont worry
(19:21:18) joaduo: i will research
(19:21:21) Bart: I'll add it to the action items
(19:21:26) Anders Dahnielson: post it to the ML
(19:21:40) Anders Dahnielson: better than this meeting for such things :-)
(19:21:42) joaduo: yes
(19:21:55) joaduo: anything more to say about scripts?
(19:22:06) Bart: not from me :)
(19:22:24) joaduo: next?
(19:22:33) joe: I wanted to ask something
(19:22:57) joe: Is is possibly to make a change so that the script writer can determine which menu the scripts show up in?
(19:23:11) joe: It seems very unintuitive to put scripts in their own menu.
(19:23:29) joe: For example, a freeze script should end up under the edit menu
(19:23:53) joaduo: i would suggest that that shouldnt be on the script
(19:24:06) joaduo: but on config file along with the script
(19:24:18) joaduo: as i suggested for normal plugins
(19:24:23) Bart: Scripts are configured using comment fields now
(19:24:37) joe: that would not be too bad, however, having the registration stuff in the script is actually very advantageous.
(19:24:37) Bart: Possibly the menu they show up in could be configured there
(19:24:42) joe: It simplifies things a lot. 
(19:24:55) joe: For example, a single script can register a lot of menu commands
(19:25:05) centipede901 entered the room.
(19:25:12) joaduo: hi centipede901
(19:25:14) joe: and installation of scripts is far easier if there is just one file
(19:25:25) centipede901: hi. sorry i'm late
(19:25:26) joe: If you look at Maya, 3dsmax, and XSI
(19:25:52) joe: they all have a feature that allows a scripts in the user interface
(19:25:56) joaduo: yes, i was thinking from a not so expert user point of view
(19:26:15) joe: however, one of the best things about Maya is that its scripts can directly generate UI elements
(19:26:17) joaduo: so that he/she can edit the "look and feel" layar
(19:26:21) joe: in particular, menu entries
(19:26:24) joaduo: layer
(19:26:30) joe: yes
(19:26:53) Bart: Joe, I'll have to look over the current documentation in detail to see how it works now. Maybe we should keep the question intil Tim is back
(19:27:01) joe: yes
(19:27:05) joe: we will need tim
(19:27:11) joaduo: so i could be optional... although sometimes i strongly believe that information from different layers should be keep in different files
(19:27:43) joaduo: so it* could be optional.
(19:27:47) Bart: OK, go to the 0.7 release point?
(19:27:50) joaduo: ok
(19:28:03) joe: I suppose the python script could always make the text configuration file itself.
(19:28:07) joe: yes, on to that
(19:28:12) joaduo: question: should we drop support for win9* me?
(19:28:19) Bart: I've had severe problems on windows (64 bit)
(19:28:19) joe: yes
(19:28:42) joe: if really old versions of windows are problematic, I don't think they are important enough to support
(19:28:46) joe: a waste of developers time
(19:28:58) joe: no one serious would be running that OS anyway.
(19:29:09) Bart: I agree, drop it if it's extra work
(19:29:25) joaduo: i agree... so we will have to publish on the wiki somehow (action item)
(19:29:52) Bart: Did anyone else try 0.7 on windows?
(19:30:00) joaduo: i couldnt
(19:30:04) joaduo: i will on virtualbox
(19:30:15) joaduo: did anyone tested on vista?
(19:30:44) joaduo: i plan to test on winxp and winnt
(19:30:55) joaduo: maybe on win2k
(19:30:58) Bart: I can test on vista
(19:31:07) joaduo: winxp(32) and winnt
(19:31:17) joaduo: i have no 64 machine
(19:31:40) joaduo: can anyone on winxp(64)?
(19:32:00) Bart: yep, I did, and it crashed when changing the number of rows in a cube
(19:32:50) joaduo: i suppose the action items is to test them and submit the bugs to the bugs tracker?
(19:33:13) Bart: Yep, adding that. any further comments on the test-release?
(19:33:18) joaduo: So... bart: vista and vista-64?
(19:33:30) joaduo: me winxp 32 and winnt 32?
(19:33:32) Bart: just vista32
(19:33:37) joaduo: ok....
(19:33:37) Bart: I have no vista64
(19:33:47) joaduo: vista64 inst out?
(19:33:47) Bart: ok. Next point?
(19:33:51) joaduo: yep
(19:34:38) Bart: Personal page. Please remember to create/update it (just did mine today :)
(19:35:14) Bart: Next are the movies
(19:35:52) Bart: I couldn't find them, maybe I missed something?
(19:35:58) Anders Dahnielson: Joe
(19:36:01) Anders Dahnielson: ?
(19:36:08) joe: yes
(19:36:10) joe: I have one
(19:36:16) joe: basically, when you log in
(19:36:23) joe: on the upper right you can click your name
(19:36:28) joe: then you can make a page there
(19:36:33) joe: I *just* edited mine!
(19:36:37) joe: It was real easy.
(19:36:55) joaduo: ha....
(19:37:20) joe: http://www.k-3d.org/wiki/User:Joetainment ('http://www.k-3d.org/wiki/User:Joetainment')
(19:37:23) joaduo: did you add your page to people category?
(19:37:31) joe: no, how do I do that
(19:37:49) Anders Dahnielson: take a look at Tim's page
(19:37:55) Anders Dahnielson: click "edit" on it
(19:37:57) Anders Dahnielson: to se the code
(19:38:00) joe: ok, cool
(19:38:11) joaduo: [[Category:People]]
(19:38:12) Anders Dahnielson: it can serve as a template
(19:38:35) Anders Dahnielson: but, we're was actually talking about the demo movies :-)
(19:38:45) joaduo: :)
(19:38:45) Anders Dahnielson: how's it going on that?
(19:39:30) Anders Dahnielson: s/we're/we were/g
(19:39:44) joe: oh
(19:39:45) joaduo: i really believe we can try to model more stuff
(19:39:45) joe: ha
(19:40:06) joaduo: that will lead into new workflow models
(19:40:16) joe: ok, demo movies going well, but I haven't been able to create free software versions
(19:40:28) Anders Dahnielson: ok
(19:40:31) joe: I have to learn ffmpeg
(19:40:36) joe: right now they are techsmith
(19:40:48) joe: apparantly, ffmpeg has a techsmith decoder
(19:40:58) joe: although I've never used it, I was just reading only
(19:41:00) joe: online
(19:41:10) joaduo: techsmith is the capturing software or the codec?
(19:41:16) joe: the codec
(19:41:24) joaduo: and wich soft do you use?
(19:41:27) joe: camtasia
(19:41:29) joaduo: which*
(19:41:35) joe: ok, here's a question:
(19:41:52) joe: does anyone know of a screen capture program for linux that will capture opengl/k-3d
(19:41:57) joe: and audio
(19:42:13) joaduo: i was going to suggest searching for one
(19:42:14) Bart: nope :(
(19:42:21) joaduo: i think i have seen on sf.net
(19:42:28) joe: right now I am using a remote windows machine with an x-server, connecting to my linux machine, and capturing on the windows box
(19:42:31) joe: it sucks
(19:42:33) joaduo: but haven't tested any
(19:42:35) joe: but it does work
(19:42:49) joe: unfortunately the codec is not common
(19:43:04) joaduo: ok, action item -> search capturing software
(19:43:05) joe: I want to make something really common, like h264
(19:43:35) joe: its ok to use camtasia
(19:43:38) joe: its working well
(19:43:47) joe: I just need to learn ffmpeg to reencode the video
(19:43:49) Anders Dahnielson: but, if ffmpeg has implemented a codec (at least reads it) it ewould be easy to convert it suing ffmpeg or mencoder
(19:44:15) joe: does anyone here know how to use that program? (I could read about it, but a short example would go a long ways.)
(19:44:35) joe: yes, ffmpeg supposedly has a codec
(19:44:52) Anders Dahnielson: and mencoder uses ffmpeg for lot of stuff
(19:45:00) joaduo: seems no one knows...?
(19:45:04) Anders Dahnielson: and mencoder has a nicer command line interface
(19:45:09) Bart: I've used ffmpeg in the past to encode a series of stills
(19:45:27) joe: ok cool, however, I still haven't used either app.
(19:45:39) joe: I'm sure I can learn
(19:45:46) joe: What is the best format?
(19:45:47) Anders Dahnielson: it can be tricky sometimes, lots of options ;-)
(19:45:55) joe: .mp4?
(19:46:04) joaduo: ok... should we post on the ML about this?
(19:46:11) joe: I'm thinking of "what is the most cross platform?"
(19:46:20) joe: I was thinking .h264
(19:46:23) Bart: Joe, if you want I can experiment with ffmpeg a little, if you upload the file that has to be converted
(19:46:28) joe: because its a true standard
(19:46:39) joe: yes, I will do that if you can help that way!
(19:46:39) centipede901: mp4 would be best. Smallest, very widespread
(19:46:44) joe: I'll send you an email Bart.
(19:46:49) Anders Dahnielson: MPEG-4 is good
(19:46:50) Bart: ok
(19:47:09) joe: so mp4 is the container format
(19:47:37) joaduo: ok... next item?
(19:47:43) joe: the most standardized video inside that format would be what? h264? or another kind? and is there a standard audio format that goes with it?
(19:47:47) Bart: OK, we'll experiment, and get back to this next week
(19:47:52) joe: ok
(19:48:07) centipede901: Bart, I do ffmpeg regularly. Write if you are stuck
(19:48:08) Anders Dahnielson: mp3 is usually "standard"
(19:48:25) joe: I will write you Bart, and send a file, because I am a bit lost.
(19:48:36) Bart: centipede901: ok, I will!
(19:48:51) joe: oh, cool centipede901
(19:48:57) centipede901: video-codec: Both ASP and H264 is video-podcast compatible.
(19:49:02) joe: I'll send it to you too
(19:49:08) centipede901: But ASP is a lot faster to both encode and decode
(19:49:17) centipede901: ASP = xvid/divx
(19:49:36) joaduo: i think google videos are a good example
(19:50:07) centipede901: Make a 320x240 version that Ipod users can watch in the bus :)
(19:50:49) joe: they won't be able to see anything
(19:50:52) joe: but sure\
(19:50:59) centipede901: I'm joking
(19:51:01) joe: my version now is 1024x768
(19:51:06) joe: ha, ok
(19:51:20) joe: and the techsmith codec is very efficient at keeping it small
(19:51:31) joe: but hopefully another format will do a good job too.
(19:51:52) joaduo: we also can think of a gif...
(19:51:53) joe: what's your email, centipede?
(19:52:01) Anders Dahnielson: I guess, since a lot of the interace is more or less static it would compress really good in any temporal compressionscheme
(19:52:02) joaduo: but should be kept simple
(19:52:02) centipede901: centipede@takhis.net ('mailto:centipede@takhis.net')
(19:52:03) joe: we need audio though
(19:52:08) Bart: OK, so Joe, Rene and I will work on the transcoding, and try to have a transcoded movie next week.
(19:52:21) joe: ok, sounds like a plan
(19:52:25) Bart: I suggest we move that to the action items, and move on
(19:52:34) joe: ok
(19:52:37) Anders Dahnielson: sure
(19:53:22) Bart: OK, Animation: So far we have linear interpolation and a new timeline to drag around keyframes that are stored in a track
(19:53:40) Bart: As the wiki says, I suggest we leave it there until the more urgent matters are handled
(19:53:58) joe: yes
(19:54:39) joaduo: i think there will be a point where we will need to animate when we are ready with modeling
(19:54:55) joaduo: and animation probably will come with IK
(19:55:07) joaduo: the need :D
(19:55:14) joe: yes, and a lot of other rigging stuff
(19:55:21) joe: animation is pretty complicated
(19:55:32) Anders Dahnielson: yes, there's a truckload of other stuff that comes with animation
(19:55:54) Anders Dahnielson: better to concentrate on modelling and not get distracted
(19:56:10) Bart: OK. So I'll start working my way down the modeling roadmap, while Tim works on the viewing mode design
(19:56:28) joe: to me these are the most pressing: framing zooming to the model, and framing component selections
(19:56:43) joe: right now its hard to even look at models, because you can only frame the center of them
(19:57:18) Bart: OK.
(19:57:19) joe: I think even before the modeling stuff, (which actually works reasonably well) some of this basic view nav stuff is still a bigger problem.
(19:57:58) Bart: joe, OK, the framing sounds like a self-contained problem that won't need much design work, so I could get started on that
(19:58:15) joaduo: talking about that....
(19:58:35) joe: Cool Bart.

The next one I would suggest would be making the UI respond to output selections, so when a new modifier is inserted that outputs vertex selections, the UI switches to vertex mode.
(19:58:49) joaduo: i saw some changes requires UI designing and others doesn't
(19:59:42) joaduo: i think we could divide them.. in easy to implement and the one that needs designing?
(19:59:49) joe: Yes
(19:59:51) joe: good plan
(19:59:58) joe: how about this one:
(20:00:00) joe: When a modifier is used in one mode, such as subdivide edges, and outputs selections of another type, make sure that when switching to that other type, the selection is not cleared or converted. Right now after you subdivide an edge, if you want to select another vertex, and switch to vertex mode, you loose your selection. Output selection cleared on component mode change bug  ('http://k-3d.org/wiki/Output_selection_cleared_on_component_mode_change_bug')This is related to the next entry...
(20:00:30) joe: "the next entry" being things I commented on before.
(20:00:34) joaduo: i think that's a bug...
(20:00:35) joe: its just a quote from the wiki
(20:00:43) Bart: Yes, I saw that
(20:01:13) Bart: This one will require discussing with Tim
(20:01:18) joe: ah
(20:01:19) joe: ok
(20:01:54) joe: are any of the things on the list easy? I suppose identifying them as joaduo suggested would be key.
(20:02:02) Bart: I suppose this is only valid for the "top" of the pipeline?
(20:02:26) joaduo: i see some of them are easy, i can classify
(20:02:27) joaduo: them
(20:02:39) Bart: The second one is easy
(20:03:01) joe: how about the move manipulator bug? (3rd entry)
(20:03:07) joe: I think its an honest to god bug.
(20:03:10) Bart: Third one is a bug, but should be easy to fix once found :)
(20:03:23) joe: cool
(20:03:27) Bart: Framing sounds easy
(20:03:49) Bart: What is backfacing?
(20:03:49) joe: sweet
(20:03:54) joaduo: we should discuss next week how do we handle bugs, right now i think we have no method. should we?
(20:04:18) Bart: Well, OK
(20:04:37) joe: Backfacing is faces that don't face you, or even better, faces that aren't visible on the object.
(20:04:42) joe: Like the back of a cube
(20:04:49) Bart: I suggest the ModelingRoadmap gets ordered by priority, and I work my way down, contacting the list if there are larger issues
(20:05:25) joe: so if you look at a sphere from the top, and select a bunch of faces, it only selects the top faces
(20:05:28) joe: and not the bottom
(20:05:38) Bart: ah, ok
(20:05:44) joe: but if you turn on ignore backfacing, you select through the model
(20:06:10) joe: right now, vertex selection seems to go right through the model, but face selection seems to use ignore backfacing.
(20:06:15) Bart: Sounds like it could be a painter property
(20:06:21) joe: Its a bit confusing.
(20:06:23) Bart: Correct
(20:06:31) Bart: the painters paint it that way
(20:06:34) joe: I would strongly suggest making it a global property
(20:06:39) joe: even putting it on the toolbar
(20:06:54) joe: it will be used relentlessly on/off by artists
(20:06:55) Bart: Is there ever a need to make points and edges behave differently?
(20:06:58) joe: no
(20:07:03) joe: make them all behave the same
(20:07:11) joe: according to the global switch
(20:07:24) Bart: ok, then it will have to be set a the render_engine level, probably
(20:07:27) joe: xsi calls the ignore backfacing "raycast" mode
(20:07:36) joe: and actually implements it the best of anything I've seen
(20:07:53) Bart: ok, I have XSI mod tool installed, I'll have a look
(20:08:14) Bart: The good news is its easy, just a few OpenGL commands :)
(20:08:17) joe: sweet, XSI mod tool, btw, it a fantastic piece of software.
(20:08:19) joe: cool
(20:09:15) Bart: OK, it's been an hour now. We could still discuss one or two points, or stop now?
(20:09:19) joaduo: action times -> order by priority and design complexity and specify clearly high priority times?
(20:09:26) joe: lets finish
(20:09:31) joe: yes joaduo
(20:09:38) joe: although I can't help with complexity
(20:09:40) Bart: order by priority only
(20:09:44) joe: as I don't understand it all
(20:09:53) joe: I tried to put it in priority order
(20:09:54) joaduo: i help with that, will talk on the ml
(20:10:07) Bart: complexity is hard to judge, I'll ring the alarm bell if there are design issues that need Tims attention
(20:10:30) joe:  maybe, leave the order intact, but actual developers can flag them with complexity ratings, and judge whether or not Tim's help would be required
(20:10:38) Bart: ok
(20:10:41) joaduo: so, we finish, and move other topics to next week?
(20:11:05) Bart: OK, I'll save the action items list
Personal tools