Camera mapping… (sigh)

[insert flipped desk.jpg here]

I am still struggling with camera tracking, it’s getting to be quite infuriating as I know the theory of how it works an IUnderstand it, however I can’t seem to make it work. Perhaps I am overlooking something…

I just can’t seem to work out what I am doing wrong; however I have taken what I have learnt from Blender and applied it to a new image to see what the actual problem is, I have decided that the final shot for this will be either a horizontal tracking shot or a vertical one, as I want to demonstrate the skill without trying to show off. Or to think more technically, is to replicate the type of camera movement that I would do, as I don’t know how to use a steady cam set up replicating one would seem almost pointless, though I am sure it allows for steady movement through a location I also need to remember that I am working with a 2D image to start with.

[update]

I think I have worked out what I am doing wrong, though I’m not sure how to fix it… Yet.

When setting up a scene I have been trying to move the cube to match the image with varying degrees of success, but I think this has shown me the problem. I have been assuming that Blender would work like photoshop, well the materials to be precise. However if I am correct the materials work like masks. When I project an image from the camera viewport I see almost what I would expect but with one issue, there is a lot of black in the image. If I’m right this is where there is no data, like how a mask would work white = on, black = off. So the chances are that I need to go back into photoshop and fix the layers to suit the shape.

Working out the project

Now that I have gotten motion tracking/matchmove out of the way and understand how it works I have spent the past few days looking at various showreels from Escapees (the graduates from Escape studios) although they are masters level students this is a viable research option as it allows me to work out what should be included in a showreel and how long it should be. Something I found interesting was that the overall time differs from 1 minute to 2 minutes with 1:30 being approximately the average time for a reel. These usually involve a shot that lasts a short amount of time before a breakdown of the shots to show how it was constructed.

 

Each video seems to follow a similar pattern, each starts with a title card containing the artist’s contact details and what role they are training/showing skills for.

Something that is also important is that the music throughout the reels can be muted without detracting from the viewing as none of the footage relies on audio.

Each reel shows the final product first, then there is the breakdown, starting with the original plates, matte layers and first composite, followed by a series of corrections and finally the final composite.

The reels have more than one scene, from what I can gather it is to show that the artist can work with a variety of scenes rather than just the one location, it also allows for seeing whether or not the environment dictates how the artist would work.

And finally to end we have another contact card, sometimes with additional information or where it was made, in this case being Escape Studios.

 

For the past two weeks…

For the past two weeks I have been looking at other capabilities that Blender has other than “just being a 3D modelling program” like I had originally thought. Turns out that Blender has a lot more to offer such as Motion Tracking which I have been looking at for the past week along with attempting 3D camera mapping… And it’s back to the annoying part of understanding the concept but not being able to apply it (more so with Camera Mapping… And I didn’t say “How hard can it be”) but both of these skills are necessary for my work and showreel.

HOWEVER it’s not all doom and gloom; I have managed to grasp how motion tracking works within Blender, in fact my first camera track that I attempted had a very low solve error, meaning that the tracking points I selected matched up correctly and in theory this also means that what I put in the background should be matched to the camera movements. The main issue that I have is that rendering with my current (default) settings is pretty slow (so far it’s taken at least 3, nearly 4 hours; though it could be because it’s rendering through cycles rather than blender render)even on the beast in the digilab (Don’t really want to mess with the settings, since it isn’t my computer; however I did change the RAM allocation via Blender’s user preferences to 1024*2 meaning 2gb of RAM is dedicated to Blender, though this may just be for the one project.)

https://www.facebook.com/zak.atkin/videos/847629168710087/

Now I need to finished the basis of my mood board and continue to update it, hopefully I can record my screen as I am working on my work to document what I do soon as it may prove useful when trying to reflect on my work.

Over the weekend I aim to have the tasks Jon had set in the tutorial finished and start on creating a matte painting, luck for me there is also a masterclass hosted by Jon next wednesday; this will prove useful for refreshing my skills from 2nd year project.