Here’s my random pick from last week’s ‘weekly folder’.
Here’s one of the shots we’re currently working on; tracking, keying, cleaning, masking, adding 3d environment & props, add guns, light bolts, robots, light design, composite and final grade! Hopefully that’s something we can show next week.
-Ton-
My first blogpost as a mango artist; I really like to show everyone the development of the making process.
I started the mango project as a production assistant helping Ton with some production stuff for the shooting days. After that I started modeling. I have been doing this for a long time for myself and now I really like being part of the mango project as an artist.
Down here you can see the development of a couple of parts I’m working on. It’s not finished yet, but I’m happy to share it with you. I’m only doing the modeling and texturing as my specialty is environment development. Today I started working on the city skyline; I wonder if it is possible to make it look photorealistic.
It’s texturing time for the dome , except for the scientist boat and a few props still needing more polies (the silos and boxes in a corner), I’m now setting up textures : a first pass tells me if the object is detailed enough and whether it needs more mesh detail or more refined textures.
An example is the bridge , right now it’s quite low poly for being in the center of the set , but textures are going a long way , and right now it doesn’t seem to require crazy high-poly sculpting , but rather a solid material and some greeble models as filler (cables, lamps, and misc. tech probably)
So here’s some individual renders dome props , you can see some of the nodes used : they are just groups (made using trunk cycles, no custom or secret code :D ) .
I’m really excited about the current texturing/shading possibilities: thanks to batchbake, vertex color channels, blended box mapping, nice tileable textures and the flexibility of Cycles nodes … you can get really nice materials, really quick.
Hopefully this will work nicely (as in technically stable and manageable) and make lighters/shaders happy (using groups heavily is now making these materials way more readable !)
The modelling for the dome environment is proceding steadily, here’s a test render (from last week, actually) :
Light design and compositing are totally temp, the shaders will need lots more testing and you can easily spot some meshes that are still at previz-level-of-detail.
This week I was more focused on modelling details so that for the shooting (or right after it) shots can be framed and setup with a good idea of the final shapes and silhouettes (see renders at the end of post).
For the upcoming Blender release 2.63 the Mango team worked hard to produce a splashscreen. We tested various concepts, as well as different techniques. It was a good exercise to take the tests to the next level and produce a “final-ish” image, that would also look good in the movie.
Both scenes still heavily rely on Blender Internal, in particular for the volumetric effects. The main renders come right out of Cycles, but halos and smoke could only be achieved by linking the scene to Blender internal, in a new scene, with it’s own setup of render layers and passes. As you can imagine that makes everything a bit complicated, since you work with different scenes, juggling around layers, passes and baked smoke files. The interactive preview capabilities of Cycles have been of great help and the speed and quality of the engine is improving almost on a daily basis! We are very excited to work with such great tools.
Grab the latest release of Blender here: http://www.blender.org/download/get-blender/
Here are the node setups for both scenes:
Textures .. there’s never enough of them (especially tileable , equalized , easily reusable ones)
While modelling stuff for the dome i’ve been doing some tileable textures kits, still lots to model, but some assets already need to be tested with materials to define the pipeline (.. or simply to check if modelled detail is enough :D)
This weekly was even streamed live via Google Plus! We have already like 700 followers there. Will do that next time again!
Further; just enjoy the pictures! A full animatic (minus last minute) should have been online already too.
This week was spent on a lot of modeling and testing efficient ways to recreate photo-realistic parts of Amsterdam. How much to do with photo/footage projection? How much is real modeling here? A lot depends on the final shots of course, but it’s crucial to master the art of recreating realism in Blender asap, before actual filming. The tests I’ve copied above are just first trials, it’s research in progress. As soon as good results or conclusions can be shown it’ll be here on the blog within a day. :)
We’re also very happy to have the help from production designer Romke Faber for the sets! Ian produced an extensive briefing. Together we should visit the studio asap to inspect the possibilities there.
For next week the main question to answer is “Are we going to film the bridge breakup scene in Amsterdam on location, or in a studio with greenscreen?”. Lots of factors to weight here, including financial ones!
Oh: anyone knows where to get the best quality chrome balls to photograph reflection maps?
-Ton-
The dome environment is currently at the end of the design/previz stage and at the beginning of the detailing / actual modelling.
Just a few minor props left to block out , but everything still needing many, many polies to look anything like a final…
Here are some shots of the dome and church sets : (still animatic level , materials and lights are totally temp ! )
Rendered in Cycles with the Tesla, 3-4 mins per frame is still too much for an animatic – and indeed we’ll probably use OpenGL render for it- unless we want to design/previz lighting already.
Then some detailing for the bridge and roads : i’m trying different combinations of shrinkwraps , solidify , particle scattering to organize these for close-up (bridge) and mid-ground (rest of the dome) views.
After a few tests , it seems promising : the risk is to rely too much on ‘heavy’ meshes for these tiny greeble details, but seems to be flexible enough for different situations.
