Brainstorming
You can use this page to propose your ideas about my current working projects, suggestions and feature requests like a brainstorming, with the single condition that you must explain yourself in a clear way or provide links to resources. This is not a guarantee that this will be implemented, but I do promise that I will read everyone:
This will help me to keep in mind the user requests, new approaches, and who knows 
?
[...] Brainstorming [...]
I just got a crazy idea for kind of a special effect…
Never seen exactly this somewhere, however a 2D-equivalent…
This certainly is low priority
What I’m talking about is morphing.
Though, not mesh morphing but full 3D voxel morphing.
I experimented a bit with 2D-morphing, trying to morph for instance a human into a fox and vice versa…
And got an idea for kind of a story by this: The untracked backgrounds on the two photos kinda twisted and wobbled… as if they where boiling or something, which is clear. But so did zones, which where hidden on either photo. For instance, where you’d place some trackingpoints for the chin of the human, this is kinda hidden by the fox’ mouth.
I liked that boiling effect for the surroundings but I’d like to have “good” morphing for the faces.
My idea was, that they might switch between two worlds, in one being animals and in the other one being human. As always. Right now, there is no actual story-idea behind that, so ’till I get an actual thing to do, it’ll probably take time anyway. But my idea is, to kinda track similar points on two (or more) characters, turning into each other, the whole scene gets voxelified and during the render it looks just like the two characters being turned into each other smoothly (no matter how different the mesh might be) while all the surroundings distort seemingly random. ^^
Ok, enough of this
Keep working on voxel sculpting and all the other amazing stuff. Great work!
i think to be honest, and this sounds boring…
there are so many pending projects for blender that i don’t want to suggest new features.
i’d LOVE to see all the pending / wip features get finished.
all the summer of code stuff, bmesh, global illumination etc etc
the list goes on
kram1032’s idea sounds really cool… 3d objects morphing by using voxels to create the transition effect (+1)
but seeing as you asked
a fracturing solution would be nice… tied in with bullet physics
http://forums.odforce.net/index.php?/topic/9119-voronoi-dynamic-location-based-fracture-wip/
I’m still really interested in GI in blender.
I cam across this great tool to bake lightmaps. Something like this would be great to have in blender.
http://www.purelighttech.com/
How about a request for the full feature set of 3dcoat? :p
Looking good! Ideas? Lets see:
1. Conversion voxel->poly->voxel with customizable accuracy (maybe I am talking dumbness);
2. Voxel probably will not support all of the current modifiers, but some shaping related ones would come in very handy. For example, basic distortions and, maybe, layer-able displacement map.
3. Joining voxel objects (so you could sculpt parts in separated layers and repeat details on a main voxel object). Maybe it is too complex to be priority at this point. Artists would easily do that anyway using the item 1 of my list. I can say the same about item 2. ;]
4. Increase level of detail of the region being tweaked based on: zoom, brush, and maybe an adjustable limiter slider in the tool panel.
So.. Conversion and dynamic resolution are my main cents.
Thank you, man!
+1 for 3D texture-based fracturing! Could be dynamically integrated with Bullet after the closer integration of Bullet for Durian. In fact, I bet this is something they’d love to use _in_ Durian!
regarding fractures – those can allready be done with fractureme in similar manner as shown.
Morphing of non-same objects can be done quite easily with shrinkwrap(even with good control)
regarding voxelscupting ideas –
it’s quite sure the voxel structure should be adaptive, store ’smallest’ voxels only on the surface – something like octree.inner and outer voxels can then greatly save memory.
good mirroring options could save 1/2 half of performance?
ideasman_42 once made a very good script for sculpting mesh which added dynamically topology. maybe some similar system could be used instead of voxels – voxels are not the only way to represent volume, mesh represents volume too.
hello,
I just post to send a link to the Ron Fedkiw home page:
http://physbam.stanford.edu/~fedkiw/
It contains a lot of papers concerning physical simulation and level set. He also published a book on level set which I think can be useful to you concerning some voxelsurface conversions.
I hope it helped a bit.
good luck and thank you
So here is quick thought. Jahka has stated he wants to rename particle mode because it does so much more now, why not include voxels with the same tool set. Use brush grab etc with the same user interface for ease of user training and continuity amongst tools, just make the mode smart enough that if you have a voxel object selected it edits them instead of particles… they are quite similar are they not?
I was surfing the web when I came on this. http://vimeo.com/groups/ice/videos/4697942
I know that your internet is limted and this a HD video site. So I will try and explain as best I can understand what it is.
It demostrates XSI Ice nodes for creating Tubing waves. It looked like they used vectors in a voxol grid to create a morphing mesh. I know isnt exactly sculpting. But was wondering if some of the particle effectors could influence the voxol grid and create vectors that could then become new animated meshes(using shape keys?) But maybe this is already possible using fluid control particle objects. But the possiblity of recording a morp over a number of frames would be nice. (I have done this with the existing sculpt + shape keys)
Anyhow just some ideas thanks for all the work
+1 for “textur based demolition with bullet physics”
+1 for “morph”
My ideas:
- Particle Morph (like thinking particles in c4d)
- Particle-Objects-RealBulletPhysics:
So if you have a particle system and use iE cubes to make particles visible, so this cubes should have a FULL collision-detection.
This would be amazing stuff for VFX!
You could make a sand-clock, growing snow hills, very good explode effects. (perhaps in combination with texture-collision.
So- this Particle-Objects-RealBulletPhysics should be a new feature of blenders particlesystem, the expolde-modifier needs a 3D-function (very long awaited feature!!!) and in the modifier we need the checkbox “texture” for explosion-control.
Yes, that is what i would like.
And then, if every thing has exploded- each broken part can morph to anything … lol i could think deeper and deeper.
Time to sleep!
Great idea- brainstorminsite!
Have a good time!
Greets,
Dennis F.
Hey Dennis,
I was trying this out in Blender the other day. Its already there and it might be better to post this to Jahka. on http://jahkaparticles.blogspot.com/ he is the one doing the particle development. So here is a super fast tut on how to.
Particle real physics
1st create your object. I used suzanne
1 Chose emitter and emit from grid set the rez. choose what object to emit I choose a cube. then resize the object emittance to be the right size(non overlaping)
2 Now make a particle instance of the 1st particle system. You need to set up the ridge body for the cube before making it real
3. Now apply and make the instanced objects.. Select all the objects and y or p to separate them from each other. And center new. They should now be ready to simulate
Hi Ibkanat!
Thank you for your reply.
I know that method, and tryed this a few times.
I also used a group of different ownformed objects to simulate demolition.
And i found a way, to use textures to control the demolition (there is a way to use a particlesystem controled by the density of bw-values of textures).
It works- but not in a way we like it.
Its just looking like improvisating- not like improving.
The particle thing can be realized with the “make particle objects real”.
But all this things have a problem: The Movement can not be controled.
What we need for HQ-VFX are particlesystem, with real collision detection. So you can make ie many many small balls that follow an empty (swarm-system) and collidate REAL with other objects.
Actual blender does not support real rotation-collision-detection for particle objects.
And it does not support absolut-collision you would need to make sandhills or snowhills…
The abolut dream for real good vfx:
You take an object: A cube.
with a random-value for different sizes of the broken parts.
The Cube has a defined “break-texture”- choosen in the demolition-modifier.
The checkbox “demolitionparts in 3D” is enabled
The Demoltion-modifier is controled by a particlesystem (this is the normal way in blender).
The cube falls down with physics.
On the floor it breaks to different parts, controles by a texture.
This parts are 3D and rolling over the ground.
Now, this particles become to small insects and fly away in a swarm.
Try to realize that!
Greets,
Dennis F.
all you have done, farsthary, is a brilliant job, i can’t wait to see it all working (volumetrics, GI etc.) in official relese. now – my idea:
i’ve always found blender lacking a good matchmover. Of course there are some alternatives like old versions of ICARUS or VooDoo Camera Tracker, but ICARUS has wide spread of limitations, and VooDoo in most cases is insufficient. So i’am dreaming silently about good matchmover inside blender, with all this shiny featuers like movie and image calibration, maybe even with footage-based scene recontruction.
(sorry for language mistakes, if any – i’am not good at english)
Cheers
joorva
joorva I am looking forward to this as well.. Just to let you know there is good work being done on this at http://code.google.com/p/libmv/ Please let them know that you are looking forward to this.
OpenCL is released btw
http://developer.nvidia.com/object/get-opencl.html
there are some interesting proyects from 2005-2007 such as DSM (deep shadow map) or Lightcuts (lights simplification and a nice GI) that i think are important but those proyects have been cancelled or paused.
I have a request for some feature tweaking in the volume renderer. The scattering feature is fighting the density scale feature. If I try to make the object denser (harder to see through) the light will not scatter as far. If I decrease the scattering value then the object becomes more transparent. I was wondering if you could make it so that the scattering value determines how far light travels through the volume and density scale determines how far you can see into the volume, without affecting each other. Although this may not be physically correct it could be helpful when trying to make objects that are extremely dense but light scatters vary far into.
Feature request: How about an option to use both volume and solid materials at the same time? That way the object has a noticeable solid surface and an internal volume at the same time. I have tried this effect and gotten nice looking results:http://img59.imageshack.us/img59/4647/volumeshell.png
I can’t wait to see your GI solution implemented! congratulations on the external shadows being implemented!
I just saw ibkanats post here about the tubing waves and thought – yeah, that would be an awesome feature to have.
Matt Ebbs Ocean Simulator is quite interesting too and combined with the possibilty of creating wave breaking and shore effects would be really cool. I’d just love this feature. Thanks for your effort and hearing. There are many, many feature request concerning blender all over blender related sites and I think it’s important to have an appropriate place to post them. So, good idea to make a brainstorming place (especially since blenderstorm seems to be offline).
So I am apprehensive to even ask but would it ever be possible to have volumetric data show in edit mode? I work with medical imaging data and this would be very useful. I imagine this would require a ridiculous amount of computation to show this in real time but the usefulness is huge.
I would use this to be able to select out specific anatomy from scan data. Presently I use two different programs to output stls then load the results into blender to select parts of joint surfaces. If I could see the volume data in blender in real time the process would be simpler and more accurate.
I loaded up a dicom set with the recent build. Thanks putting that together I admire the dedication to the blender project.
Ciao Farsthary,
thanks for all you did for Blender! Your contribution is incredible and your current project seems to be spectacular!
Anyway, If I could ask you a feature, I would ask a “solidify modifier”.
I think that cloth simulation would be finally complete with this.
I hope you will code for Blender for the most time!
For better detail in the fluid sim, would integrating some of the smoke math in the fluid sim be possible?
More specific, Nils’ video on turbulence has a small render of water, and the detail is much better than anything i’ve yet to see from the Blender fluid sim. Overall fluid sim in Blender is too clumpy for anything bigger than a glass of water imo
Here’s the link to Nils’ video (water sim is at the end) :
http://graphics.ethz.ch/Downloads/Publications/PaperVideos/2009/Pfa09%20-%20SIGGRAPHASIA2009%20-%20Synthetic%20Turbulence.mov
I second those tube waves by the way. They look fantastic. Hopefully we will be able to do something like it in Blender someday.
I had an idea. I came across this while surfing the web:
https://launchpad.net/scandraid
This guy has launched an open source 3d scanner software and if you look on his blueprint page he plans a plugin for Blender. I have worked on a bunch of projects and the ability to sculpt would be really enhanced by the ability to scan in 3d objects into blender. I noticed another web site where a guy used a ruler and a light and the software created the 3d object from webcam video. Yet another software takes three or more pictures and creates the 3d object from those. After a bunch of looking I realized that the above solutions are mainly software. You could theoretically take three photos of an object from different angles and if you could use point cloud analysis or make a special modifier to an existing object based on the positioning of a photo with respect it to reconstruct the object. Obviously there is a sliding scale, the more effort you put in the setup (i.e. with lasers and turntables) the easier the software is to write. Anyway just a thought.
Sorry for the double post. My browser was not refreshing properly. :S could you remove one?