Hi all,

A nearly up-to-date detailed manual is available here.

Aim is to make you understand the fundamental concepts: this can take a bit more time than applying a tutorial, but at least you know then precisely the reason behind a setting: So after its reading, you will have much much more clouds/smoke in your blender scene than in your head 😉

PS: This would be very cool if somebody could update the blenderwiki accordingly. Indeed, with a few more optimizations and additions, chances are really high that volumetrics make their way toward an official release: As Ton can not integrate a new feature if there is no decent documentation (have you ever seen an empty release log?), we have to make a full-feature wiki. Matt Ebb has told:

Anyone would be very welcome to help fleshing this out straight in the wiki, if there any mistakes I can clean them up afterwards – just getting stuck into it would be great because I don’t have a huge amount of time for it these days. There’s a lot of people here following this progress and offering thanks, which is great to see, but this is one way you can make a considerable concrete difference and I’d very much appreciate any help here.

He has encouraged to use his commit logs as support, so feel free to reuse Matt’s pictures and movies.

So if you want to support this project (better documentation -> more users -> more interest for the project -> more motivation for the coders -> more features/optimizations -> users go nuts), this is the right time to do it

Thank you in advance



GI coming to Blender

Hi all 🙂

Matt and I will  join again in another project: Photon map Global Illumination for blender.,since both of us started to work on photon mappers independently (Matt few weeks ago and me few days ago), a happy coincidence.

he’s said:

“It’s by pure co-incidence that Farsthary today said he wanted to look at doing the same, I guess we’re following similar trains of thought. Anyway, the work I’ve already done, although still in heavy dev, is quite functional so far – at least in a testing sense, I haven’t started putting it into production contexts yet. I don’t know if Farsthary has any code functional yet or not (I haven’t seen it) but I’d be very happy for him to work together with me on this again, it makes sense and two heads can be better than one!”

Currently, near every high end render supports photon mapping GI, this approach has several advantages over traditional unbiased renderers for calculating GI (as well as drawbacks) but represents a natural extension to raytracer renderers  and allows to add lots of new effects to raytracers.

Matt has previously implemented the photon map using the BVH tree, and I have used a Kd-tree ,  in order to learn and thinker (I especially like that 😉 ). As soon as he finishes to polish his code, we fusion forces. 🙂

Here are Matt´s tests.

Yesterday I posted that I have near all the photon mapper implementation but lack the blender related functions but thanks to matt code I could fill the gap and here are some test, some of them are really fast!

Render 17 secs
Render 17 secs
Rough GI very fast to calculate for stills
Rough GI very fast to calculate for stills
GI (ops! I forgot the render time 🙂
20 000 photons shutted Render time  14 secs
20 000 photons shutted Render time 14 secs
20 000 photons Render time 37 secs
20 000 photons Render time 37 secs

Hope you like it   🙂

GI coming to Blender

Addition of external shadows

Hi all,

Here is a new proposal for the sim_physics branch: like the MLS one, this means that this is successfully implemented but awaiting some code cleaning/UI modification from experienced blender dev like Matt Ebb: As he kindly helps me in his free time and because he has a lot to do, please understand that it can take more time than expected…

In the current time, there are still plenty of gifts from the generous Matt Ebb to test: do not hesitate to ask questions about the sim_physics on my blog: even if I am not always connected, a friend of mine has told me that he would help as much as possible those who are already lost…

Proposal for sim_physics: Adding external shadow support

Adding full shadows greatly enhance realism and directly allows spectacular VFX such as the alien Suzanator vessel darkening part of the cloudy sky 😉 (Fig. 2)

Matt Ebb had previously implemented shadowing derived from the very design of the raycaster engine:  objects inside a volumetrics cast the light. However objects outside the volumetrics did not occlude light sources, leading to unrealistic illumination. I have so corrected this issue: now let the pictures speak as usual 🙂


Fig. 1 – current sim_physics branch – render time: 9 s


Fig. 2 – external shadows support – 20 s

You can see on the 3 images just below that this leads to the expected illuminations.




Full volumetric shadows calculation can be a very time consuming process by increasing up to twice the render time. Consequently, since it is not always needed, the user should have activate/deactivate it in a panel (no other UI button is needed)


Fig. 3 a), b) – no external shadows


Fig. 4 a), b) – external shadows


Fig. 5 – external shadows with more lamp samples

The calculation also takes into account self-shadowing


External shadows are very important for increasing the relationship between the volumetrics and the rest of the scene through light interactions. High render time could easily be decreased in the future by storing the shadows in the Light Cache structure (work-in-progress)

Cheers to all, Raúl Fernández Hernández

Addition of external shadows

Ice test

Hi all,

Matt Ebb has kindly reviewed the MLS patch, so it should make its way soon to the sim_physics branch.

As another example of the usefulness of MLS, here is a full volumetric ice test with MLS: hope you like it 🙂


Fig. 1 – Basic material (no volumetric)


Fig. 2 – volumetric with single light scattering): the volumetric lightning is localized


Fig. 3 – volumetric with multiple light scattering (come soon..). See how the light diffuses up to the ice edges.


Fig. 4 – Surface (Fig. 1) and volume lightning (Fig. 3) combined through basic node set-up


Fig. 5 – result


Fig. 6 – result with compositing tweakings

Ice test

Some words about the integration of the smoke/fire simulator

Hi, this is a friend of Farsthary.

As you may know, integrating the simulator inside Blender is a tiedious task which requires deep understanding of many concepts. Just as an example, it has taken a whole summer to Nils  Thuerey to integrate his fluid simulator inside Blender (thank Google!). Daniel Genrich, father of the cloth modifier, has been improving the integration of this simulator and also its speed (OpenMP). He has also been working in unifying/improving all the physics tools,  becoming thus the responsible of the new sim_physics branch heavily updated by Matt Ebb and Farsthary.

Matt and Farsthary are currently pretty busy with work/2.5 developpement and study/researches on volumetrics respectively.  Daniel Genrich is also making a “coding pause” for researches, but he has previously stated that he is very interested in integrating the smoke simulator inside blender, and this would be really appreciated by Ton for realistic VFX in the next open movie project (Durian).

Remember that adding a whole new feature might exceed the time span allowed by an open movie project (the micropolygons integration project wished for Peach (QDune) was dropped because its integration was too time-consuming, and would have threaten the creation of vital features). Moreover, the addition of new features was often possible because they were under heavy developpement before the start of the project  (eg: the new particles system from Janne Karhu). Finally, a experienced blender developer is required to precisely follow blender coding conventions, and to think about the integration with the other physics tools (wind, obstacles…)

For all these reasons, one should support the integration of the simulator right now and Daniel Genrich is the guy for that, so if you are eagerly waiting for smoke/fire simulation in a near future blender version and want it heavily tested/optimized  in the production environment of the next open movie project (in less than 6 months?),  just tell Daniel how happy you will be if he implements this and if possible give him a donation (paypal) as a sign of consideration (remember how dedicated he is to the blender community, he has worked relentlessly for the cloth project )


Some words about the integration of the smoke/fire simulator