Development Update (odds and ends)
on January 6th, 2010, by ideasman42Since my last post and over the holidays I have mostly been working on smaller things which I didn’t even think it worth posting about, some of these were requested for specific tasks, others were just for fun. Recorded 5 videos, read on for the other 3.
And before you ask, all this is in the public subversion builds.
Shape Key Transfer
Grease Pencil Depth
Camera Switching
AutoDepth (back from 2.4x)
Baking (back from 2.4x)
Other developments
- Solidify Modifier (much requested, after the editmode tool)
- Sequencer copy Paste, needed for comparing 2 versions of the animated storyboards.
- Presets, Render/Cloth/SSS
-
API Reference, now generate using the sphinx documentation system (used by Python)
Access by Right Click “View Docs”, Can be searched unlike old API docs. - Fast ray cast, function that can be called from python.
- Passing on maintenance of Rigify to Nathan, development continues in smaller fixes and refinements.
– Campbell
January 6th, 2010 at 8:43 pm
Awesome stuff.
You code-heads are working overtime, Campbell even worked CHRISTMAS every day… O__O
Maniacs…but I love you all!
January 6th, 2010 at 8:45 pm
Nice!
I wonder, could you perhaps add a user preference option for that interactive baking, to maybe allow us to reload the viewport texture on intervals? I understand it’ll unpractical to do it realtime, but maybe reload every second/2 seconds would be doable and useful.
January 6th, 2010 at 8:49 pm
That grease pencil is magical.
Thanks for your work.
January 6th, 2010 at 8:50 pm
Awesome stuff! -I love the sketch into depth thing -now perspective will be a breeze to teach 😀 well.. maybe you could add an inter-object exception which omits the line inbetween two objects so it casts like a shadow instead 😀 -then add a button for it 😀
LOVING YOUR WORK HERE! :0
January 6th, 2010 at 8:51 pm
Blendersome! jajajaja
With the Camera Switching, can we make the transition from one camera to another (like an animation)?
Greetings and good work!!
January 6th, 2010 at 8:56 pm
It’s nice to see and watch the gradual evolution and improvement of Blender. Any specific plans to improve the render? This is and was, in my opinion, still one of the biggest weaknesses of Blender. What use is able Node editor without proper output from render… Anyway Hats off and thanks.
January 6th, 2010 at 8:59 pm
Does the shapekey transfer work only and always on meshes of the same identical topology?
January 6th, 2010 at 9:01 pm
Can’t thank you enough Campbell!
January 6th, 2010 at 9:09 pm
Really awesome job! Woot!
January 6th, 2010 at 9:17 pm
Awesome work. Keep up the great work Blender team.
January 6th, 2010 at 9:19 pm
Not worth posting it?! C’mon man, you are a genius Campbell, please post everything! We are hungry of your news! 😛
January 6th, 2010 at 9:21 pm
@Divine Rage, yep this is probably not so hard viewport-bake-update will look into it at some point.
@erik90mx, nope, for this I think you’d end up wanting ease in/out and for this its better to use animation tools, though smooth view between cameras could look neat but is a gimmick in this case I think.
@radek, I dont work on the rendering engine but lots of render development is planned, don’t worry.
@OriginalBBB, They must have the same topology yes, Id like to write a version that can do some geometry translation but this isnt needed for durian.
However I did try using barycentric transformations for this, which is interesting in that the 2 meshes can be a totally different scale and shape. You could for eg. copy the shapes from an Ogre onto a fairy. even if the scale is 100x different and the fairy happens to be upside down. The options there in the tool if you want to try it but I found it more technically interesting and not actually that useful for us especially since the results are not as good as a no-brainier offset.
January 6th, 2010 at 9:24 pm
Awesome, great to see new developments coming quickly out of the open movie projects.
January 6th, 2010 at 9:25 pm
Durian updates are the best updates.
January 6th, 2010 at 9:42 pm
Yay great new features 😀
Very nice stuff!
January 6th, 2010 at 9:47 pm
That’s the way to show features… many and straight to the point… 1 minute ~10 features.
All very useful! particularily like the shape key transfer. But most of the other are great, very manageable grease pencil on surfaces, that new pan/rot/zoom behaviour, camera switching…
What is fast ray cast?
Thanks for all these “smaller” features!
January 6th, 2010 at 9:48 pm
Awesome updates!
January 6th, 2010 at 10:02 pm
Awesome work! Thanks!
This is inspiring stuff… sure makes me wish (yet again) I was a programmer/developer beyond ‘Hello, World!’-status.
Cheers!
January 6th, 2010 at 10:07 pm
Oh man… I can’t thank you enough for the autodepth. Probably one of the most annoying things to work around. This makes things SO much easier.
January 6th, 2010 at 10:14 pm
I’m simply amazed. I have waited for camera switching for a long time and while it seems a bit inconsistently hackish for the UI (considering the consistency work done for 2.5), it works.
Autodepth was something I didn’t even know to hope for.
Very cool.
January 6th, 2010 at 10:22 pm
Sweet awesome grease pencil features!!!!!! Grease pencil is one of my favorite things to mess around with when doodling out something or I’m just plain bored.
January 6th, 2010 at 10:38 pm
Ingenious and really helpful stuff here.
Luv the grease pencil and auto depth 🙂
January 6th, 2010 at 11:08 pm
Very very nice indeed! 🙂
January 6th, 2010 at 11:37 pm
It is the little things that make blender so great.
January 6th, 2010 at 11:47 pm
The new API documentation is great, just I think this is a bit noisy:
arg foo: …
type foo: …
I wrote a greasemonkey script that converts it to:
foo: …
…
See:
http://twoday.tuwien.ac.at/pub/files/make_blender_api_doc_mor.user
January 6th, 2010 at 11:48 pm
Do you still have to hold down a key on the keyboard when using Grease Pencil or has that been fixed? in <2.5 I used it quite a lot for sketching up models before starting on the actual modelling, but in 2.5 that is just not possible. Holding down a key WHILE drawing? no, that just doesn’t work IMHO. Maybe for plotting primitive things like shown in the video, but definitely not for actual model sketching with a wacom.
January 6th, 2010 at 11:53 pm
Amazing! I have to say, it’s about time with the camera switching, but now it will be a lot easier rendering a series of shots in one go. About the Shape Key Transfer… is that only for identical geometry or does it transfer regions to similar geometry?
January 6th, 2010 at 11:59 pm
Why don’t you just have a field that says ‘active camera’ so you can put keyframes on that field?Or do animations only work on things that can be expressed as numbers?
January 7th, 2010 at 12:14 am
For Autodepth, I have an idea.
Could you use a low-pass filter on the z-depth?
Could you use PCA on the z-depth?
Both of these could have user-definable sensitivity, and both are intended to look for outliers in z-depth. For instance, if you’re drawing on the ground and the pen accidentally crosses a sign post or something, these methods would trace the ground near-perfectly but ignore the signpost.
Just a thought,
MTracer.
January 7th, 2010 at 12:16 am
That auto depth function seems to be a solution for quite some frustrating moments I had when navigating trough my scenes. Thanks very much for the updates!
January 7th, 2010 at 12:30 am
Impressive developments. Thank you very much all of you devs. 🙂
January 7th, 2010 at 1:12 am
Hmm… at about 1:01 on the autodepth, your mouse cursor looped around the screen. Will this be in blender, or is this something you have?
Also, I found out that the feature was already in <2.5 blender, I just didn’t see it/know what it was.
January 7th, 2010 at 1:40 am
@panzi, agree with your that type & arg everywhere is ugly. however this is the method shown in the Sphinx manual so not sure whats up with that.
@Tomm, Not sure if Holding DKey is going to be kept. Did you notice the ‘Draw’ button is now in the toolbar?
@jAkl, ray cast means you can say – drop objects into terrain from python very very fast without having to look into face by face intersections. Im working on a tool for scattering junk in the cities so its coming in handy there.
Its using the same internal function as shrink wrap.
@me, animating the active camera directly is no so simple
Im assuming you mean animating with a curve? – Cameras have no set order (appending a new camera may mess up the index). So if you used a curve renaming or removing a camera would break.
One way could be to have SceneCameraSlots – like material slots. Then a curve could be used to animate the active index. Nicer since it uses the animation system but still seems a bit dodgy to me.
@MTracer, what you suggest would not be so hard to add. There is an array of float depths so high frequency jitters could be filtered out. Its already using this array to interpolate infinite depths.
@Alexsani, Mouse warping is now default in 2.5, experimented with this a bit, tested warping the window, view, hiding-mouse-cursor and bringing back where it was clicked. Found view warping best overall since can can always see the mouse and it stays in the same context (you dont end up over a button in another area).
January 7th, 2010 at 1:55 am
great new features, very pro and useful!
January 7th, 2010 at 2:13 am
GREAT WORK!!!
January 7th, 2010 at 2:51 am
“You could use it to draw a person… okay it’s not a really good person…”
lol
-Really cool developments all around!
The cityscape and Guardian heads looks quite awesome so far too. Keep up the great work!
January 7th, 2010 at 3:30 am
Wow. Those updates are great! I love the pencil thingy. You can almost model stuff with it.
Not sure if that information might be helpful, but in Cinema 4D the camera issue is solved with a “scene object” that has a keyable camera property. So if you have a setup where you want to switch between cameras, you create such a scene object, then at the certain frames you drag your camera into the scene object’s camera field and set a key. This is the nicest solution for the issue that I have seen so far (then again there are a lot of programs that I don’t know).
I guess this is kind of similar to your slot idea. But it seems a bit more “tidy” to have the cameras packed away in such a specific object.
MD
January 7th, 2010 at 4:31 am
AMAZING!
If this is material you consider barely post-able, then I can’t wait to see the stuff you think is news!
I really need to get better with blender … so far my uses for it seem so lame in comparison to the work that is being done with durian … I almost feel ashamed. Keep up the good work!
January 7th, 2010 at 7:15 am
I am from china! I can’t see the flash movie!
please give the url of media files!
I can download the files and see see!
January 7th, 2010 at 9:07 am
You. Are. Amazing, Campbell.
January 7th, 2010 at 11:52 am
2.5 alpha is already wonderful to wok with, but the full release is gonna be awesometacular!
January 7th, 2010 at 11:53 am
Hallucinate great features! Happy New Year everyone! 🙂
January 7th, 2010 at 11:42 am
Great, incredible, amazing job !!! I especially love the Grease Pencil development.
GM
January 7th, 2010 at 12:48 pm
(re. animating the camera)
> I’m assuming you mean animating with a curve?
Well I don’t see what the use would be of a curve since it’s not a continuous value. There would probably have to be special case code because of the problems you note, like if you store the active camera as a number, what if you add or remove a camera. But even that doesn’t work, because every object is a potential camera.
If you want to do this completely right, you’d have to have a separate architecture for managing the active camera. Some top level object, like the scene, would have to store a link to all objects that act as a camera in one of more keyframes or in the scene as it is now. The list could then be used to translate from an object to a number and from a number to an object. The number could be used to store the active camera. This list would have to be saved and on load could be checked for items that are no longer used in any keyframe, so unused items can be removed from the list and the numbers in the keyframes can be kept up-to-date at the same time. Kind of how the single-user, multiple-user and fake user reference counting works on textures, meshes and other datablocks.
It sounds complicated but most of it would be invisible to the user, who could just skip to a keyframe, change the active camera using the normal user interface and then insert a keyframe on the active camera setting. That seems more natural than the “drivers with camera’s attached” thing in the demonstration.
January 7th, 2010 at 2:26 pm
I love everything. Like the way the grease pencil works. I’m happy with the camera changes, but wouldn’t it be wonderfull if we could choose wich camera we wanted to use in the sequence editor? Then it would be like editing a video!
Many thanks again!
January 7th, 2010 at 4:33 pm
It looks like you’ve been working through christmas just to make everyone else’s January 2010 soooo much better than christmas!
Heh, I’d thought my last 2.5 build had most of the cool stuff I needed… but no.
That grease pencil is frickin awesome, and I like how you nonchalantly debuted the model of the street scene >.>
Oh aye, and I’ve been using Blender since ~Feb 2007, and on the second day I realised I (and about 75% of the forum community), was crying out for easy camera switching. It’s so indescribably important that’s now fully in Blender for easy use by nontechnical users, so thank you!
(From the tech end tho, how come it’s so hard? i understand it’s more than just niggly referencing; that fundamentally keyframes apply to continuous ranges… and you don’t like to bodge camera indices into that kind of thing, but what *would* be your ‘ideal’ way? does there eventually need to be a way to keyframe wonky discrete set values?)
And! … mouse wrapping! Dude! Wanted! Per-viewport does sound good. Is mouse wrap an option in the test builds? Can I access it somehow? If not, what bribes do you accept?
January 7th, 2010 at 5:18 pm
this enhanced grease pencil is awesome!!!:D
thanks!!
January 7th, 2010 at 6:00 pm
This is so awesome 😀 I like the smiles, they’re cute 😛
January 7th, 2010 at 6:01 pm
Wow! awesome! Shape Key Transfer!!! Jack looks weird with that smile though…lol
and Grease Pencil 😀 and the new Zooming, and Camera Switching, wow!!!
really awesome 😀
GO DURIAN TEAM!!! :-DD
January 7th, 2010 at 6:28 pm
City looks very cool. 😀
And so’re the little features, Campbell. 😛
January 7th, 2010 at 6:29 pm
@waipy–
http://www.youtube.com/user/ideasman42
(his channel, all the videos are on there)
January 7th, 2010 at 6:29 pm
am spellbound !! greasepensil is just awesome.. thanks a lot campbell 🙂 . dont know wot u term as “worth posting”.. waiting to see tat 😛 .
January 7th, 2010 at 6:58 pm
Campbell:
Take care on the streets.
Most Blender users are male and maybe someday you just can’t avoid a crowd, rushing for kisses.
January 7th, 2010 at 9:58 pm
Although, I still think the NLA is better for managing camera angles.
http://watchmike.ca/blog/page/8/
January 7th, 2010 at 10:43 pm
Wouldn’t it be kind of neat to be able to change cameras from within the sequencer? To be able to have one or more strips for each camera view, and be able to edit the sequence using the opengl display, before even rendering! You could then just press render, and have a finished product. The different transitions should of course also work (you’d just have to render double the number of frames during the transitions).
I think it’d be nice to have something that close to regular movie editing, anyway 🙂
January 7th, 2010 at 11:37 pm
Great great guys! Just wanna ask something: how long takes a build that you’re current using to be compiled from graphicall? Maybe is a really stupid question 🙂
Thanks for your all magic work
January 8th, 2010 at 10:04 am
I think what Cris meant to ask is: when can we expect to see a stable release of this? Do you work on a schedule, is it possible to give us a release date? Based on what you’ve done, can you take a wild guess in, say, months?
January 8th, 2010 at 5:35 pm
@me: regarding camera switching. I think we’re on the same page mostly.
It could work like this…
– Camera slots are managed internally, users can just key cameras from a scene camera dropdown (as a number of people suggested).
– Blender would update on deleting objects and adjust the animation curve too when slots are added/removed.
– Use animated curves, otherwise we need to have some different animation system/interface which IMHO is too much just for this feature. The camera names could be displayed on the left of the fcurve view, curves draw with sharp changes to show its switching from 1 camera to another.
– Users don’t need to be aware of these camera slots unless they use the curve view.
– This way we get nice camera keyframing that fits into the existing animation system.
This isn’t that big of a project so could be done before durian ends if I get some time.
@acro, the problem with camera switching is that it needs to be implemented in a nice way, dont think its especially hard. Above I mention a way that I thinks nice enough.
Glad you like the mouse warping. This will be in any recent graphicall build. I added this mainly because I remember having angry moments when transforming some objects and having the mouse hit the screen limits… grr. Its also much better for a vertical layout when you have buttons that drag horizontally.
@TobiDn & Mike, a track view could be nice for switching cameras however the sequencer and NLA both are dealing with different kinds of data so it doesn’t not make sense to put camera switching in these, and defining a new kind of track view JUST for camera switching is overkill IMHO. Some kind of animation event system could be good but this is a much bigger project. – Almost like game logic in the animation system if it was done properly.
@Cris & Fabio, Artists have a desktop icon on their ubuntu systems which downloads and compiles an up-to-the-minute blender. So fixes can be used within a few minutes of developers committing them.
As for releasing – blender is a huge project and each of us only maintains a small part. Its very heard to say but more then 6months at least.
January 8th, 2010 at 8:28 pm
Wow! All I can say is “Wow!”
Camera switching question: When rendering out to file (images or video) is there a mechanism for file naming that would be used to differentiate which camera rendered which files?
January 8th, 2010 at 11:10 pm
Impressive work Campbell !!!
2.5 is going to be really impressive. Thanks for the videos.
January 9th, 2010 at 1:03 am
Dude, switch your viewport rotation to ‘turntable’ please, It’s killing me in your videos. That way the screen won’t go all tilted every time you rotate!
January 9th, 2010 at 1:04 am
I forgot to add, thanks for the Baking!
January 9th, 2010 at 1:55 am
Good job, nice bunch of features 🙂
January 9th, 2010 at 6:25 am
The grease pencil looks really cool and I would love to see it developed further.
Sure, it wouldn’t have priority for Durian, but here are some features that I think would be cool:
1. Let it be possible to set properties for them like width and color and allow them to be directly renderable.
2. Pressure sensitive to control line widths
3. Animatable (step or interpolated)
4. Ability to close shape and create fill
5. Ability to tweak lines after they’ve been drawn
Just imagine how quickly you could throw together an animatic with real blender camera moves with these features.
I haven’t used Blender much but intend to start when 2.5 gets more stable, so apologies if any of these don’t make sense. I know Python pretty well, so maybe I could write it myself.
January 9th, 2010 at 5:32 pm
(sorry Rudiger)
6. Ability to convert the grease pencil contour on the mesh surface in actual mesh geometry (a bit like the retopo tool…i’ve seen a similar feature in a video on youtube, i think the 3d package was zbrush)
January 9th, 2010 at 8:07 pm
@ Campbell: This might be small steps for you, but they are big steps for our imaginations! 😉
@ Rudiger: I thought about requesting your points as well, but had a look at the grease pencil to see how it already works within the current version that I use: 2.49b. – And – what can I say – your first three feature requests are already available – except allowing them to be rendered.
I would like to see that we can render the grease pencil, too. Maybe, simply convert them into particles which show up only in the given frames?
But, probably you only need to add a button to switch the visibility in the render options?
January 10th, 2010 at 5:16 am
@Saverio: Great idea!
@Achim Luebbeke: Animatable? Really? I must investigate!
January 10th, 2010 at 10:03 pm
@ideasman42:
That would appear to address my main concern.
January 10th, 2010 at 11:40 pm
Thanks Campbell!
The videos are very good to understand your new features and make it easy for us to understand the benefits.
Hope everybody is healthy. In my region of germany (Stuttgart) many people have been sick over the holidays.
Aussies, enjoy the snow 😉
———————–
2010: The year of Durian.
January 11th, 2010 at 12:56 am
Great features indeed… So I did test some of it already and I think that the autodepth rotation in trackball could use some more work. Now if I rotate around from something that’s in the center of the srceen it works great but when I rotate around surface that’s near the edge of the 3d viewport, it rolls the screen around like it would normally do when rotating from near the edge of the viewport. I’d prefer that the trackball center was under the mouse when rotating from the edge. I don’t know how this is going to work though but at the moment I can’t really work with this. With turntable it works fine but I prefer trackball. How to do rolling then would be another thing…
It would be nice to have autodepth separately for both rolling and panning as I’m really attached to the old trackball (but I’ll try this out more of course). I already found the new panning a lot more easier. Also new zoom with Ctrl is really something I could use but I would prefer mouse wheel roolling do the original zoom. Some more options would be really nice.
But anyway. You do great progress. Thank you all.
January 12th, 2010 at 12:43 pm
I really love the new additions!! Great job!
I don’t mind using the change current camera from the timeline, however I agree with @me that it would be much better to go to a frame, select from a list (Not a dropdown) the active camera and set the keyframe for it.
I would also love it if it if the markers were added to a bit; for instance change color if a camera change occurs on it, color coding the name to differentiate between types of markers, or even a layer, color or tag system for markers perhaps, to group things like dialog vs animation, or transitions etc.
Just a thought!
January 15th, 2010 at 2:29 am
Great work I’m loving the new features!
I have been using autodepth a lot lately, and i have to say it’s incredibly useful but sometimes in some situations it’s a bother so what would be best in my option would be to have additional option to have autodepth on only when some key is held. I don’t know how much work it is to make that but it would be megawesome 🙂
January 16th, 2010 at 5:37 am
Wonderful progress!
Grease pencil with depth -> can we use it to make a path?
Perhaps a rough mesh object?
Perhaps to (re)define bone paths in animation and select bone position for a given frame along the path? (wish wish)
January 17th, 2010 at 1:32 am
Funny, I thought mouse wrapping was a bug. And not a welcome one at that. But then, I don’t think the version I was using was very stable, it probably works better now…
February 3rd, 2010 at 2:25 pm
AMAZING it all is amazing there is nothing bad about your project just stunning outstanding and fascinating!!!!!