Brain cell network, Micron Stroke, Alzheimer's prevention, animation.

This 3d animation is part of my work on the prevention of alzheimer’s disease and dementia. A model of the crystalline network, an individual neuron (in blue) is supported by other neurons (also in blue), yellow shaded cells are support cells (oligo) and they wrap and protect the most important neuron. White cells (astrocytes) take nutrients from the blood capillary network (red), and feed and supply the neuron. Green cells are microglia, and they provide cleanup and other duties inside and around each crystalline network. Your brain has 100 billion of the cells in blue, each organized into a specific network, and each supported by a great number of cells.

I hope that this forum can give me some feedback and guidance. I already have a list of a few things that I am in process of changing. I did this with PNG at 300 samples. Then into the VSE with a H.264 at 24fps.

I am disturbed that most of the shaders look great in still images but when in animation there is a ‘noise’ that comes across the surface, almost pixilated.


(I already noted that my transparency was set at 8 causing black shadow error, adjusted it to 12. Also a few small modeling defects ill repair in the mesh)

  • thanks in advance for some ideas.

In the rendertab under sampling, there is a seed value (default is 0) Put this into that setting

#frame

This will make cycles use a different noise seed for each frame which will definitely help your cause.

It is nice to see Blender being used for a worthy cause like this.

that is very interesting. were the meshes generated from scan data?

I modeled all the meshes Modron. Started with simple geometry, and a few times i used the skin modifier and joined meshes. (over 1000 hrs into this project easy) - but I also had to go back and redo some as I learned Blender better. Also I had a lot of model adjustments as information from the physicians i’m working with in Spain guided my hands).

Harleynut97, I am going to give that a try. So do you animate the seed? (example: frame001 seed at 0, frame 4000 seed 1) or what would be a range you think is appropriate. And I also am very happy to be able to use Blender to get into a visual animation information that is inside the mind.

The nice thing about using the #frame is it acts like a script and automatically changes the noise seed value on each frame, you do not have to keyframe anything. Just type that in and render normally.

thanks harleynut97 - that is a great feature! I think the translucent shaders are the most affected

Translucent shaders tend to introduce LOTS of noise into a render with cycles. There are two ways you can deal with it, you can up the samples per frame until it’s no longer noticeable (the #frame trick is very useful but won’t eliminate noise, just make it less noticeable), or you can fake the translucency. I think I’d do it with with a separate render for your translucent objects (looks like its just the oligodendrocyte cells at the moment? Forgive me if thats the wrong name, not a neuroscientist :P) where they just have a diffuse shader and then another render with a holdout shader for all other objects in the scene, I’d then composite the translucent only render back in on-top of a render for your full scene (obviously omitting the separately rendered translucent objects) with the desired opacity, using the holdout shader render as a mask. Perhaps that’s a bit too convoluted though :confused: you might think it easier just to bump up the samples.
Is this a purely “diagrammatic” animation? If so I think you’re fine with the scene you have now. If not it would be great to add some noise to the camera move so it looks a bit more “floaty” as these microspace scientific animations often look, also the flat colours don’t fit so well, maybe some subsurface scattering or glassy shaders, would look a bit better.

Sorry for the long post, just exciting to see someone else doing scientific animations.

@Subdural : Congrats, it is a very nice modelling.
As other said, Cycles induces a lot of noise. Why don’t you use Blender Internal for this kind of renders ?
With this kind of lighting it would be great for your purpose.

Thanks Words and Crus-T. I was considering going into the compositing - I am still learning that compositing gets deep. I still have a lot of work to do: Labels for each cell type, flash/glow effect where they attach, some lattice movement animation and the astrocytes (white) sorta reaching in to their connection, when the camera pulls back I am thinking of adding force and then ‘blowing/windstorm’ the green microglia out to clear the model up a bit when the camera pulls back, some smoke effect or volumetrics to the model when camera is pulling back. mesh repairs.

Word - i struggled with the camera movement - it is on one long curve, with series of animated constraints. Did you think that adding like a noise in the graph editor? I would really want you to expand on your camera ‘feel’ and what you think looks better. I am building a renderfarm this week so I should be able to render and look at the results faster. you nailed the oligodendrocyte too - (Important cells that wrap like insulation around the neuron to protect and stabilize it)

Crus-T, i go with the cycles as I liquid cool my GTX 780 - and when there is a new one out…someone in the house gets a better game system! Other than what I think was ‘helping’ myself with GPU and render times (wont help with the renderfarm) - I guess when I read that they will discontinue the internal engine at some point - I sorta feel I want to avoid getting used to it - my current model will go on for additional animations that describe how the brain damage (micron stroke) and amyloid protein interact to affect memory and mind.