Hello: Below is my process of how I completed Paraty House render by Marcio Kogan.
Softwares used : 3dsmax : Zbrush : Photoshop : illustrator | Render : Vray | Plugins :: Onyx : Multiscatter
Textures : CGtextures: My own art work ( textiles. ) | System spec : Amd 6 core 16gig ram
In order to try and achieve a good result I had to obtain the relevant information of the real building and http://www.archdaily.com was a good source. The images were taken to Illustrator where I created plan drawings which were then exported as .dwg files for 3dsmax. Once inside max they were assembled to be used as reference while modelling. (All Units were in CM) You can change this my going into preferences > unit setup.
I mainly used poly modelling , spline’s and slowly built up the scene. Modifiers such as extrude / symmetry / turbo smooth where stacked and added where needed. For details I cut out grooves and used different amount of chamfers to give a nice smooth edge. (Probably pointless in some cases ). I prefer to model everything as it’s good practice. Other components such as kitchen and furniture pieces were created in a new max file as I found it easier to work this way then merged into the main file. Again using the above methods, building the models with good geometry and making sure backup copies were created. ( max is a crash addict!)
Furniture is by a diverse range of designers , one them being George Nakashima and they were all modelled to the dimensions found accept a few. Coming from a product background I was itching to create my own furniture and i did! as it can get boring copying all the time! But a worthy session as creating models from sketches can make or break a design.
I love Zbrush but I despise all the technical stuff in any software!!. I just want to jump in model and jump out again. This is how I did the cushions and rocks. Create a low poly model and unwrap. Export as obj to Zbrush. Create details and export back either as a high res model or low poly and add details in max using normal maps etc. This way you can add a texture map while preserving details. Thanks to BBB3 for clarifying it out.
Displacement or model?? Hmzz. Well we have FloorGenerator from http://www.cg-source.com and that did the trick.
Only issue is that it works on a “flat” surface. So I had to break up the buildings into sections and apply the FG settings. Would be good if FG had a chamfer and extrude option too. Don’t ask why, I just feel it should. 🙂
Onyx was used for creating high poly trees and various bushes. Some i-tree models were also used, which saved me a lot of time and learned from too. Numerous smaller plants were modelled, some as basic planes and others such as the banana tree as a shaped model. The grass method is by http://www.peterguthrie.net/ or alternatively check out http://www.jamesshaw.co.nz/blog/?p=400 or for a free sample http://www.3dmentor.ru/free-models.html.
The vegetation was created in a different file then merged into this one, then converted as a proxy within the main file. The land area was broken up for various camera views and Multiscatter was applied to the larger sections. This enables a greater control to enable or disable for the views. For smaller sections or anything that was going to be close to the camera it was done manually apart from the grass!.
Tora Bora? 🙂 mentioned that it was a good idea to create high res textures which can then be downsized when needed. This can save a lot of time if you are saving finished textures for future use. All my textures were at 4096 but a few large areas I rendered the uvmaps at 6144. I saved HDD space by saving the textures as .jpg and around the 10 mark in ps. All my materials are very basic vraymtl. Depending on the surface, textures were inserted into the reflect and glossiness menus. Unwrap UVW + UVW mapping used. The wall texture below. After experimenting I found this version worked and went with it. Might be wrong in the dirt parameters. do let me know!
For the plants there are many methods. It would be better to see it from the original sources. One from http://www.peterguthrie.net/blog/category/tutorial/. I noticed Xfrog ( free trees ) and i-trees use extra settings in the vray sub menus. I did not use a back mtl as there were way too many trees. Not sure if this has an effect on render times.
Sand Material was by Javior Pinto. Planes were cut into patches for view dependencies. The noise modifier gave an uneven ground and worked great with the sand mtl. For distant shots I turned off the displacement in the sand mtl
For lighting I used Vraysun plus a Dome light with a bitmap/hdri attached in the texture box. Settings ( red ) for both Sun and dome light were changed according to scene configuration. ( This also included sun direction)The kitchen scene had basic Vray plane lights and photometric ies lights for the ceiling lamps. Free ies from http://www.lithonia.com/.
Each scene consisted of a Vray Physical camera however the settings in red were changed according to light Conditions. Check http://www.spot3d.com/vray/help/200R1/examples_vrayphysicalcamera.htm + http://en.wikipedia.org/wiki/Exposure_value to learn more about Exposure Values. This was taken from http://www.thecgschool.com/free-resources/
For rendering I used LWF ( Linear Work Flow ) , Irradiance map + Light cache. HOWEVER for each scene I used different settings ( red areas ). As mentioned before this was a learning project. But I would suggest going to the following site and reading on Image sampling and various other information on Arch Viz.
I saved my render files as 16 bit Tiff color with alpha on and with many other render element just to experiment with. The tools used were Adjustments > curves > color balances > exposure settings and even paint overs where needed. Layers methods vary as it’s dependent on the end result. The aim was not to do that much post work as everything was 3d but sometimes post is needed for improving images , computers specs creating limitations and speed. But remember if post is needed then do it but if the end result is an animation then make sure things are done correctly. Lesson learned! 🙂
The wirecolor layer is very important if you need to separate certain section for segregation. I forgot to turn my top trees layer on and instead of doing a new render section I just copied and pasted trees into new areas to create thicker foliage. It seemed a palm tree in the scene was too far away to cast a shadow and made the right area odd. Excluding the palm from the left and using the transform options I created a false shadow just to fill the gap. This method is great if the tree line fails.
This section was causing excessive render times ( left – 10min for 320×240 and without 2min ) and crashes. It was narrowed down to a palm tree there. I removed it and added it in post. Again the wirecolor element comes into action!. You can either spend 10 minutes to fix a problem or add new render section. Another good trick ( apart from existing tree photos ) is to render high resolution single or group tree files with alpha, which can later be used for post production or use existing foliage from other scenes. I am going to try that next!
Well I learned a tremendous amount from doing this project and now it is time to start another one! More stress and swear words! 🙂
My Flickr has been recycled so I’ve added the images here.