This is an order I got for a building set in the middle of nature.
As I said in earlier posts, I wanted to make it full 3D. However, didn't know Luxrender didn't support instances in Hybrid mod and crashed because of RAM saturation issues all the time. Since the deadline was approaching, I just ended up post processing some landscape with three bad quality low res photos I was given by my client...Most difficult thing in this was to make a bird's eye view while photos where taken from the height of human eye. The image you'll see here are mixing pure 3D and post processed images.

Hi there. I've been working on a lot of projects lately and I experienced lots of things but never had time to post here.
These days I'm working on some visuals I've been ordered, and there will be bird's eye views and exteriors...

I made a first visual with 2D plants and grass added in post processing. Took me the whole day to get a good result on a single image. But I got more than 10 images to produce..so yeah I just don't feel like doing that again and again.

So I decided to really get into 3D vegetation. This way it will be done once and for all.

So here a grass sample I'm Satisfied with.

I wonder how many time did I crashed my computer because of RAM saturation...However I'm happy with the result. I just would have loved it to be much lighter.

This is some strands I modeled/mapped coupled with a particle system in blender. Then it was rendered with Luxrender wich became my default rendering engine.

One more update with luxrender.


Yet another Luxrender image shot from Blender.
I'm really happy with the result.
Obviously I used the blender sea modificator, a real delight (however I would have loved if we could manipulate the proportions of the piece of sea we get...).
Yet something bothers me a lot : the lowpoly-like reflexion on the water...However I guess it's easy to fix in an image editor...

The most bothersome point is I needed 130 hours to get to this level of smoothness which is far from perfect.

When I say 130 hours it's not on a dorito, it's on a I7 2600k...

This is just too slow for production under heavy preasure, which is the case in 99% of the situations in Architecture.

Lately I've been experiencing a very annoying bug with Lux. So I'll just be exploring another renderer while I find a solution.

I like to be able to rely on a lot of stuff (that's what saved me when I moved this scene to blender and switched from Indigo to Luxrender). Lots of people will tell you "you need to master only one program, and go deep into it".
I believe this is one of the most looking right but totally wrong advice you can get.
Once you understood the way a type of program works, you can use any clone within a tiny adaptation time. This will allow you to have many options when facing a problem...discover new things permanently, and stay fresh in your head enough to keep on the edge of technology, rather than getting affraid of improvement and inovation.

So to get my solution, rather than getting me a render farm, I'm just gonna try it again with Cycles.
I've been very frustrated with it since the beginning because I invested lots of money in my ATI graphic solution...and well everybody knows at the time it only works correctly with NVIDIA.

There are also issues with Cycles of course, but I'll talk about it later when I get going with it and produce some renders.

Here two visuals made with Blender + Luxrender.




There are still lots of shots to come from this scene with Luxrender...
However I'll soon be trying Cycles also..even if it's not OpenCL working...(it doesn't work even with the most basic scene I do....a simple diffuse cube). But all this is still to come...the OpenCL development of Cycles seems difficult and still far from ready for work...

So here a new shot made with Luxrender. Tonemapping the image made me realize how flexible and featured the luxGUI is.

So let's be architectural ! This is a shot I made in the playroom as you may have recognized.

I wanted to show what kind of ambiance you can have in the library space I made...I chose that ball chair because once you are inside you can easily sink in your own world since you are much more closed to the environment. However, you still can look at the water situated...uppon your head.

The pebbles are not put into water, they are dry, however you walk on the big stones to arrive here, just as if the pebbles were a lake you are crossing...

I wanted this image to show how wide that part of the playroom's space's immersion potential was. I'll call it a success.

Organic modeling via Blender.



When I first begun learning Blender it was for total modeling. However it quickly revealed not suited for engineers work, especially since the bmesh wasn't implemented and push pull functions aren't as simple as in sketchup.

So I left it, until I feel the need to go deeper in modeling and wish for organic modeling to make..crazy shapes first, but also organic models to enhance my visuals...trees, stones, water, grass, ivy etc...

So the first idea was to make those in blender and import them in Sketchup once they were textured. But it revealed really not to be a good idea, since the models become muuuuch fatter and slower once imported in sketchup.

So since I intend to make animation in the end I decided that once I reach the rendering process (which means conception, structure and volume modeling are done) all will migrate to blender and continue there to be worked on for visual purpose.

This allows me to have finer and reversible control on models while the weight less than 1 MB in Blender. (the tree alone in sketchup would be over 250 MB and the stone would be 110 MB).

Two shots from my luxrender latest tries.


These are images I made using Blender + Luxrender, which is IMHO integrated to Blender better than indigo.
In my case I needed that integration advantage Lux has, so I switched to it.
The image on the right is a WIP shot of the playroom, as you can see even if it's at an advanced stage, there are still things to set up.
However I'm particularily happy with the concrete material I could get back very precisely similar to the Indigo one I made. Alos the tonemapping abilities of Luxrender are really very featured.
By the way It is important to note that Luxrender is Open Source.

Hey everyone,

While working one those textiles and pillows, I got an idea for the playroom I'm modeling these days in my spare time.

The part that was desined to be a garden with mold, will rather be covered with black peddles.

I had several ways to do it :

-The straight way : A simple displacement map with photographic texture as a diffuse map.

-The catch your right ear with your left foot way : making a particles system with 40 000 stones replacing each particle...

Of course, I love to make my life complicated so I went for the second.

I successfully made my particles system and everything was fine, I had 40 000 round peddles for 700 KB which is wonderful.

However, problems begun when I tried to export those peddles out of Blender. I had a 65 MB file just for those and they were so much, that just selecting them to move them or so would make my computer go down on its knees.

The other problem that was quite annoying is that the particles were colliding wildly and I didn't find a way to avoid it. This is no problem for hair or grass, but for stones it's not very cool...

So first I thought "ok from now I'll migrate to blender for all the render work"

But then I realized I was not ready for it yet, since I don't have many advantages to do it while I still don't use Cycles, nor make animation...

So I postponed this for later and decided to go for the displacement.

I must admit in the end I'm very happy with the result and won't change even if I could.

A simple displacement map on a very simple 4 edges plane...you can't be lighter, the stones are all really different from each other, they have authentic peddles shapes and each one has its own character because of the diffuse map fitting them perfectly.

In one word this is from far a faster, more realistic and lighter way to do.

Here are my first physics simulations shots. On the left a cloth simulation, on the right a softbody simulation.


OK so...it's been something like 2 weeks I've put in my head I wanted to make beautiful textile and soft bodies (like pillows etc) so I can make my renders more realistic..especially indoors. And I got to admit...I'm just fascinated by the beauty and complexity of textile and soft bodies.

So of course, making it yourself by modeling it by editing a traditional mesh a little by little is possible, and I've seen people do it. You can get very nice pillows this ways, but for textile it's just horrible, I've seen very talented guys make horrible textiles and be proud of it.

So, since I love to make it complicated when it can be made simple...I decided I'd make it by physics simulation...and I'm lucky...blender has a cloth and soft body simulator.

However, all this was new to me, cloth, soft body...the difference between those to physical entities...etc...so I first had to know what are the ways to reach my goal...then to know how to use them.

I didn't struggle so much with cloth...I kinda quickly managed to get what I want except thickness....that point got me mad a few days. However there is still one side of it I don't master : how to make seams that give blankets a puffpuffy look...

However, Soft bodies were a big pain to me especially because I red everywhere fake information and also because there was a parameter I tweaked by mistake and that was just screwing all my simulations during a week.
Biggest problem to me was to find an advanced tutorial...which I never found. All I was finding was people having fun with some jello cubes...so that was miles away from pillows. And in addition to that pillows are more complicated than jelly toys, it has to be able to bend smoothly, without dancing, without getting flat, without...bugging etc etc..so I really had a hard time with it, especially since I wanted to reproduce precisely the physics of my lovely pillow I sleep on since I'm born !

The goal with those simulations was obviously to make stuff that can be thrown in the middle of a scene and get the good shape automatically. Bad surprise when I realized some soft body setups were efficient in a case and not in another. However, I ended up finding the configuration I aimed for ! Finally !

There is still things to solve, like the way I'll proceed to make garments, particular shapes etc...I wanted to use Marvelous designer for this...but It bugged with every obj I sent it from blender or Rhino...and in the end I got fed up losing half of my 30 trial days on nothing. Blender fits my wishes better...I'm not Jean Paul Gaultier after all !

Here a new update, the scene didn't progress, it's still empty and composed with dummies, however, I felt I caught good shots so here they are.

I won't hide I'm very proud of that one where you can see the scene through the reflexion of the bulb.

You can find the original light model here : 3Dwarehouse

However the bulb was originally a cylinder and the curve was sharp and totally dirty...so I rebuilt the light and made a globe rather than a cylinder.
I naturally made the curve with Rhino since this is typically the kind of tasks you use it for, however you can do it using lots of other ways...tsplines, blender etc....talking about blender, I used it to model the globe itself, artistic modelers are really funny, it's like playing with plasticine.
It took a while (since I'm still playing around with all those programs) to decide what I would do with which program, however the best way to get answers is obviously to try it :)