2014 Installment baby, first release of the year!
Really nice piece for Caddi, in my opinion, the technical execution was very nice and clever. Heavy render effects like depth of field and motion blur were done 100% in Fusion. The depth of field is OpenGl based and quite fast to render (If you have a Quadro of course) ;)
Will be adding a little breakdown soon but don’t expect fancy stuff here, lately I barely have time for anything.
This thingy works like this, the render features a Blinn material with an animated texture, HDRI reflections, soft shadows, real camera depth of field and a nice animated displacement. The scene is quite simple just a sphere.
By the time I made this scene like a year ago I was submitting a bug to Eyeon (which unfortunately is still there) so there’s actually no HDRI but an image I made in fusion copying the shape of the HDRI with some masks and background tools. Plus, I didn’t need to attach the file when submitted and rendered faster.
So here’s a little screen grab where you can see the whole process. The network is quite simple, it’s just a texture (the blue and yellow/orangish one in the lower left corner) which was animated and colour corrected. then that was mapped into the blinn’s diffuse and the alpha of it, colour corrected and used as a bump. The alpha, again was used for the displacement of the sphere via a spherical mapping tool. You can see the “fake” HDRI used for the reflections there.
…and now you are thinking, “…but wait, this guy said he used soft shadows and he rendered it in OpenGL… which doesn’t support soft shadows!” So here’s the trick…
If you thought of that, you were right, Fusion’s OpenGL render only supports hard shadows. So the trick is to render the shadows in the software renderer and since we have the whole scene already set up there, project the shadow render as if it was a pass over the textured model. In the screenshot you can see from the side, the texture swimming across the 3D model, of course, from camera view, looks fine. Another option can be to actually use it later render out all your passes and comp them together BUT here’s the thing, you wouldn’t get incredibly nice camera depth of field. If you have Lenscare you might get a really nice and decent result, still not as nice and accurate as this one.
Although, all this comes at a price. Sometimes quite high… geometry, for a smooth displacement, in any 3D package you need to subdivide your meshes. Whether it’s mental ray micropolygon displacement in Maya or a simple and basic mesh smooth using your old-schoolish and obscure 3d tools of choice (which we all know you like ;).
In my scene, I’ve used a Fusion primitive which was subdivided like hell, you can see the mesh before and after displacement, in both cases mapped and in wireframe mode. It does look way darker than the render, because the 3d framebuffer was not assigned a LUT and the texture is linear, so you are looking gamma 1.0 straight.
That’s about it folks, I was thinking in making this comp public since there’s nothing special about it or making another of this balls made out of glass. Maybe in a future post, because I have a rocky shader I might like to post also, if there’s a chance with a different or more detailed walk though.
Zain Help is a tv show or more likely a short section sponsored by Zain Telecom for which this two pieces were created. The first idea was to develop an animation using negative space, but due to Zain’s brand guidelines constraints, I was unable to create. As a brand. Zain, is well designed, and quite strong. It has that feeling of modern branding, simple and straight to the point. Sometimes the application of this “straightforwardness” draws the graphics extremely simple hence it starts kinda defeating the purpose of the branding manual itself.
Since all the actions in the “piece” have to be surrounded by the “active pattern” I thought that maybe the pattern itself could be the centerpiece of the animation. I took the illustrator curves for the pattern and took them to Maya, where I built the geometry in Nurbs for the thin colour lines.
And it started like that, first the illustrator curves for the active pattern were taken to maya and extruded along a path. I needed them to be Nurbs because they are lighter on the graphics card, and i wanted to tweak the curves later to get a nice looking shape.
Then, it looked kinda like this…
What you can see getting thick blue is … yes, what u think it is… wire frame. I told you guys that it was geometrically Nurbs which is lighter on the graphics card for many reasons which i’m not planning to elaborate now, but the main thing about Nurbs is tesellation. Believe me, I’m a nurbs junkie, it was the first geometry I’ve learned to model onto in Maya 4 and I still find it extremely versatile and cool to work with… clearly not in this case. This thing was un-renderable. Latest 12 core Xeon, 32 gig ram could not render this… Maya 2008, 2010, 2011 could not render this…
The solution I’ve found was to convert it yo polys and that’s what’s on the screenshot. What did we end up with… a 2 gig file of “good old polys boy”. After tweaking and setting the render properly, it worked flawlessly with final gather and everything. Although, all this was just for a kind of animatic/mood-board, and after converting our friends to polygons you can forget about animating the extrude. I wanted to do a flow path for the animation, but as you can see there are way too many streaks to animate. Ok since this was done in a span of 3 days (weekend included sadly enough) there was no time to build a new rig or anything or start from scratch. Working with 2 gig poly files is something I do not enjoy particularly. Many crashes, and like 5 mins to save the file at least praying for Maya not to crash while translating the scene.
The answer: Animated shaders. Not the nicest way to go, I know it. But during the time frame till the deadline, animating the textures in Maya for all those objects and rendering with the same sampling (which was quite high) was the way to go. Full on. Rendering surface shaders with animated black and white ramps with no raytracing in mental ray is extremely fast even with all this geometry. I could re render the animation if i didn’t like it in less than 20/30 mins.
There you got, a few surface shaders and the animation curves. If you are wondering about the steps in the curves, is because of the different shots. I have like 4 to 6 animated cameras, and creating curves for all of the to be continuous can be very time-consuming you in this case is easy to fake continuity.
What else is there… lots of utility nodes, various mia_exposure_simple to preview my renders in sRGB and output linearly and the ones which are taking more space, gamma nodes. The gamma nodes are mainly (in my workflow at least) used to linearize my textures. in this case there are only few textures if there’s any but as I said before, Zain has a very strict branding guide, in which of course they have a nice spare chapter just for brand colours and how they should be used. They give you RGB and CMYK. Hence as any other thing you see in the monitor, if it looks good on-screen it means it has a colour profile hanging around there. So I created a ramp node with just one value, no ramping, just the colours from the manual in RGB and plugged those into the gamma nodes for them to look the same or closer to the actual brand colour.
After all this set up, render time!
The animation needed to have the pattern floating around there, as it’s name implies “active pattern” so it follows the actions. My proposal was to use Fusion’s 3D particles. I exported the camera in FBX and took it to Fusion where I animated the emiter in 3dspace to match the scene accordingly. Since is the same camera and animation the render was very simple to set. Also, for the background I used a passive pattern. Which the Zain boys were happy to provide. Although in my version it’s animated with some perlin noise in grey scale. That mapped into a Fusion sphere primitive and since I had the camera there, it was rendered in the comp. You can see this in the upper left corner with it’s render on the side. Then the final render and the render I got straight out of Maya. Below the particle right with lights and shadows and some of the textures used for the pattern. The “depth of field” was achieved with fusion’s depth tool. Motion blurred with reals smart motion blur using La maison lm_2DMVshader out of maya. Kinda old school but production proof, that’s how I roll haha.
And that’s about it, there’s a reflection pass added later (like two months later) due to an agency change but they loved it and didn’t ask for any change, which was a relief.
Hope you like the walk thought.
And if you are still thinking “is this guy still using Maya 2008?!” Die antwoord is YES! Still rocking it. I find the new ones extremely clunky and slow to work onto. And to be honest I was planning into moving to renderman and the only version I have for renderman server has plugins till Maya 2008 =)
Although not the final online or audio, the edit remains the same, just my version lacks of branding =)
I will post a little before and after as in the Qatar post later on, meanwhile you can check the crystal clear video. Lots of grading and clean up, also, the dark shots required an incredible amount of de-grain, details to be posted later on with the images.
HD version available on vimeo click here!
Simple and nice project. No fancy stuff here. Some stabilization, cleanup and grading. This project was shot in the “Pre Epic-M days” with a MarkII and lovely Zeiss lenses. After the edit, all the footage was exported in Tiff. Then I’ve treated in nuke to reduce the rolling shutter and optical flow. Afterwards had to rebuild the edit in Generation and start the grading and Noise reduction in Fusion.
I’m quite satisfied with the colour, hence here it is for your own eye pleasure!
And here you have a before and after for a few shots, you can also check the compression artefacs and how the treatment affected them.