final

This didn't work either. I've run out of year and i still haven't managed to get anything to track perfectly. Very, very frustrating and i  will have to sort it out before 2014. The issue must either be frame rate or just doing the tracking wrong. Test objects work fine in boujou and nuke but then screw up from maya. my time management on this term has been bad, and im finding being a director alot of extra work. thats all i have to say. im a failure at tracking.

My Final Attempt.

If this doesn't track. I don't know what will and I will give up.

Hopefully this is high contrast enough to work. 

I filmed it using a 28mm lens, which may be a bit distorted but it has an USM motor in it to smooth out the jitter. The difference between it and handheld is staggering already. But I am also smoothing it in AE aswell just to be on the safe side. 

update

As you can see again without any tracking markers my little golden oscar moves around a little bit.

Museum Track update

Boujou seemed to track the scene OK. I masked off the legs of some people and put a little golden statue in. I made a cheap image sphere out of a tiff and rendered it out with a colour grade, grain and a little blur. The shadow needs to be corrected. 

Boujou VS NukeX

I've decided to focus with Boujou for the rest of the time I have left. I like NukeX for splitting apart EXR files and sorting out renders, but for this kind of work, getting out image sequences and exporting to maya, The Boujou and AE workflow is a lot faster and more friendly.

I filmed this footage inside the British Museum Aztec section. Doing this made me realise a few things.
- A steady-cam is a must.
- Focus pulls are a pain in the arse on the smaller lenses that I love for photography.
- tracking markers are a must.
- People are a nuisance and make filming anything difficult.







Im not sure what I was aiming for filming this. It wobbles all over the place and I had no idea what I was actually going to put in the scene. However this gave me a wonderful new job of "IMAGE STABILISATION"

Now. Im a patient man. But the idea of 2D tracking footage before 3D tracking it made me sick to my stomach. But apparently the new creative cloud is good for something.

The Warp Stabilizer VFX button, smoothed it all out with a single click. 

As good as this will get.

This is as good I think i can get this without using markers. I don't think its too bad. the model gets close to the camera which is often said to be a bad thing to do. But it was a good effort all the same.

Revelation

Well today it hit me. The issue with trying to position things and being unable to work out were to put them. DUH!.....tracking markers.....this is why we place tracking markers.....what an absolute idiot. I have some more footage to test, but without tracking markers I will have the same difficulty with placement. I hate myself for being so dense.

More tracking testing

Here is a more successful track. All the frame rate is sorted out and it looks better. A small drifting does occur.

Importing these into maya is a pain because of the lack of a ground plane. I think the drift occurs due to positioning. 

Im bored of dinosaurs in ravensbourne now so need some different footage. 

Museum Updates and tests

Went to the British Museum today to survey it, it was a darker day which was good to see the inside. After learning about the motion tracking. I'm doubting if the Aztec room is the best place to try and film. It would be ok if we are able to bring in lights of our own but I doubt this would be acceptable to the museum. There are other locations in the museum which would be high contrast enough for us to track or bringing in blue tacked tracking markers might be another idea. The lighting in the room and masks is so low it was hard to get decent photos with a 50mm lens at 1.8 on a 5D with an iso of 1000. I got some footage to try and track which i will do later.

The room was too busy to take a HDRI which I had planned to do. But I at least got these photos of one of the masks. 

Return of the dinosaur

After retracking, exporting and seeing the cylinder work better. I put the dinosaur back in the shot. He's looking quite a bit better now the position has been changed.


Here is another example with the dinosaur slightly lower. There is still drifting problems. But I will try with the other clip and export that from Boujou to maya aswell now, to see if its the footage causing the problem.

Walk before you can run

Because of the prehistoric failure of the last test. I decided to do something simpler and attempt a cylinder.
I think the reason the objects are moving all over the place is to do with the placement. I have other theories as well. Here are my 3 ideas of why they suck:

1. The frame rate may be off between the programmes. Maya should just render the frames with no issues, so perhaps playback in Nuke is off.

2.  The objects are not positioned in the scene correctly. This would cause parallax issues.

3. The camera settings are wrong in Maya. The focal length says its 14mm when it should be 28mm. this cylinder made and rendered through maya tracks ok-ish. But the footage is out of focus in parts. I wonder if this has affected it.

Maya Issue

Tried to add a dinosaur into the footage in Maya. As you can see this hasnt worked out. The camera seemed to export fine. and worked when I positioned, but reintegrating seemed to throw some problems.

Boujou test

I then tried the same thing in Boujou. I like Boujou as it is simpler to export scenes to Maya, but its a pain because it likes image sequences.

I tried to track one of the 2 bits of footage I had filmed. 

the scene seemed to track ok. 

and rotating the scene to the correct angle was much easier than in nukex

I cant create a video from within Boujou but here is a screen recording showing a box that is tracking quite well.
So I've been more successful at tracking within both programs. Boujou feels more intuitive, but it involves another programme in the pipeline and it would be nice to stick to Nuke. I will now experiment exporting those cameras and scenes into Maya.

Drift Issues Continued.

I watched a tutorial on digital tutors to try and work out the causes of the drift. I found out a few things about the track I did.

1. My track error was above 1. This would mean an inaccurate track. 
2. Because you can't select individual points in the point cloud in Nukex it makes placement quite difficult. This would explain why people like to select their own points rather than auto tracking. 
3. My detection thresh hold was too high, I was tracking 150 features inaccurately

To sort these issues I;

1. used the refine tool to refine tracks and reduce the error to below 1. 

2. Picked obvious points that referred to things i knew in the footage. For example, the few floor tracks. 

3. I upped the amount of points and increased their accuracy. I ended up with the same points but they were better.

This is the video changing these settings. 

As you can see the track is far better. But I know this is more down to it being quite easy to do. The ground plane in this video is still off but because this is a sphere it doesn't notice so badly. I am going to attempt a similar experiment in Boujou and see how the process compares. 

Drift Issues

I found out the drifting was due to the ground plane not being set correctly.

First NukeX test

Today I have had my first attempt at tracking in NukeX, using some footage I got from filming in uni. 

exporting the camera into Maya created a weird point cloud that I didn't understand. I spoke to Nick about this and he says the points can export as weird sizes sometimes and to double check that. 

This is the first test video I have created using the 3D geometry available in Nuke

As you can see, the box drifts around the floor. I'm not sure why this is.

Laptop Back

2 weeks later I have my laptop back and working.

laptop disaster

Before the weekend my laptop died. This means I am having trouble updating my blog and getting computer work done. It is being repaired at the moment and I hope to have it back soon, in the meantime I am trying to use the computers in college.

Lens Choices

After looking at the different lenses. I think I will need a distortion free wide angle lens (maybe a 35m) for filming inside the museum. Aswell as something like an 70-85mm for close ups. Professional cinematographer and lighting engineer Barry Gross recommended a very wide angle lens which has minimal distortion available to rent from;

http://www.vintagelenses.com/

Lenses etc 2

28mm

135mm

In this next experiment I was looking at the way focal length distorts at different distances. As you can see 28mm is quite distorted and fish-eye at this close distance where as there is very little distortion from 135mm.