A while back I shot some footage in my living room to use as a backplate – with the eventual aim to integrate some 3D into the scene. I made an HDR of the room and tracked the footage.
Well, I played around with a couple of ideas – one of which was to integrate in a snare drum. I got as far as the initial comping, but found it ultimately wasnt working. It might have been let down by shaders, but anyway I abandoned this initial idea.
Drum Integration test
Recently, I found a really cool transforming robot model when browsing Digital Tutors – and thought it would be a good test to try and integrate into the backplate. (Transformers, original huh?). Here’s a tracking test I did to the footage taken last year:
Luckily I didnt take up a career as a camera operator, as apparently I have jelly arms – the footage is really shaky, however the track came out well. The pose is WIP, and I am hoping to have the model eventually animated.
I set up the scene – using a proper linear workflow, then let the HDR cast indirect light with final gather as the ambience in the room. A single key was then used for the direct lighting. The following render is a straight beauty render without any editing. On the whole it felt like a good starting point, I was quite happy with the lighting integration, with very little work it seemed to match fairly closely – it is possibly let down more by shaders and textures – and plausibility in the scale of a metallic robot sitting on a table, which in itself doesnt feel realistic.
I carried on with it anyway, hoping that at the comping stage a realistic integration would be possible with full control over the passes and grading the render. I sepnt a bit of time compositing, colour correcting and grading the renders to achieve as closely as possible a plausible integration. This is how it is looking at the moment:
With my limited compositing experience, this is what I have managed so far. It is getting quite close to what I wanted to achieve – the lighting is matching well and the grade is starting to get closer to a photo real integration. It is definitely still missing that little bit extra that really sells the image, it is still obviously a CG robot, I feel this is down to the scale of an industrial looking transforming robot placed on a table which doesn’t really feel correct. The shaders could probably do with more work, and I’m sure the composite can push the realism further than I have managed so far.
The next stage is to render out the full sequence, and see whether the integration feels more believable when given camera movement and some basic animation. Hopefully this will be done in the next couple of weeks!