Tracking tutorial for Visionlab.
Posted: Mon, 31st Aug 2009, 6:09am
Post 1 of 15
I've been wanting to do this tut for a while now but have been procrastinating.
Anyway, I hope you guys find this helpful even though I didn't do anything really fancy with it. If you want a more in depth tut... let me know and I'll make a part two.
The video seems to lag just a tad bit behind the audio so I apologize for that, I must've done something wrong.
So here it is on Vimeo.http://www.vimeo.com/6352007
Posted: Mon, 31st Aug 2009, 2:30pm
Post 2 of 15
That is seriously cool, spydurhank, thanks for putting that together. Separating the origin from the quad like that is quite brilliant.
Posted: Mon, 31st Aug 2009, 4:56pm
Post 3 of 15
Thanks Axeman, that means alot dude.
Posted: Sun, 24th Jan 2010, 12:01am
Post 4 of 15
Spydurhank, I've only now been directed to this tutorial by Koradin and I have one question. Would this procedure be possible in, say, CompositeLab Pro? I'm guessing it would, since it seems at first glance to be simply a matter of using layers.
If so, then this is a clever way to approach the labourious job of adding elements in a shot which stay in place in the scene environment when the camera moves.
In any case, it'll certainly do until someone writes a program that automates what you did there.
Thanks very much for your efforts on our behalf.
Posted: Sun, 24th Jan 2010, 4:06am
Post 5 of 15
Yes you should be able to follow the same process in all of the Lab products. It's important to plan ahead and place tracking markers in your scene beforehand. Plus it only works with imported video and not the effects engines though, you could alternately render an effect into a video with an alpha channel and then import it and copy the quad data onto it. You can also do the same thing to stabilize shaky or footage that has rotation although this does take a bit longer.
I tracked this shot twice.
The girl had 2 tracking markers on her forehead for the wounds which if you look close... I forgot to scale down and up to simulate depth when she leans her head back, and I placed 2 more on the washing machine for the blood splatter.http://www.youtube.com/watch?v=PvmhMaLgIRA
It'd be cool to have this as an automated option in the Lab products but for now... this is the best I could think of. Glad it helped though.
Posted: Sun, 24th Jan 2010, 4:38am
Post 6 of 15
The girl is obviously very resilient, Frank!
I think your "faux tracking" works quite well, enough to be effective.
Just a minute ago, before receiving this post, I PMd you about the procedure as you outlined in your tutorial. I'm still a bit hazy on some details, like copying quad values, etc. This is because I'm very new to multi-layering Apps and not familiar with the workflows involved. Some of the things you did were over my head, I admit. I wondered if you were planning to make a second tutorial some time down the track?
Alternatively, is there any chance I can get a dumbed-down guide to the procedure, maybe as text or .pdf?
Posted: Sun, 24th Jan 2010, 7:58am
Post 7 of 15
Yeah I can make another tut if you want... just let me know the specifics of what you actually need or want to do in your video clip and I'll work on a tut for you but I'll be more specific on what I'm actually doing. That was one of my 1st tuts' and I know it came across very sketchy so sorry about that. I'm better with video tuts' rather than written 1's so give me some time but I'll get it done.
Let me know.
Posted: Mon, 25th Jan 2010, 12:44am
Post 8 of 15
See my PM, spydurhank.
Posted: Mon, 25th Jan 2010, 12:46am
Post 9 of 15
Posted: Tue, 26th Jan 2010, 3:05am
Post 10 of 15
what's the best way to download the tut?
Posted: Tue, 26th Jan 2010, 7:03am
Post 11 of 15
Don't have to download anything... just click on the link a few posts above this one.
Posted: Thu, 28th Jan 2010, 4:41am
Post 12 of 15
Managed to figure it out, spydurhank. Another viewing of your Tut and a bit of cogitating and it fell into place.
Couple of points. Do you think there's really any need for the marker to be very small? If there's, say, a large geometric shape in the frame, eg, a window, you could draw a solid-coloured rectangular marquee in Gimp and as long as one edge and one corner of the marquee perfectly matched the window, it would be a lot easier to position it for each frame.
Second thing, the process would be perfect if there was a way to nudge the "background" layer a pixel at a time, using the arrow keys. I tried to find a combination of keys/arrow keys that would do that but can't find a way.
Is there any way to do it, do you think?
Posted: Thu, 28th Jan 2010, 5:38am
Post 13 of 15
Yes angles or any shape will work, I would have done this except the detail in my footage was crappy. I'm not sure what you mean about nudging the background though.
Posted: Thu, 28th Jan 2010, 8:40am
Post 14 of 15
When I imported the transparent layer with the marker, (which I named background), that I drew in Gimp, and placed it over the footage, I used the mouse to click on it and manually move it so that it lined up with the same spot on each frame as I clicked through the footage, (I'm assuming this was your procedure).
This was extremely difficult to get right every time, even with the frame magnified quite a bit. I thought that if one could move just that marker layer by hitting one or the other of the arrow keys, as in Photoshop or Corel, it would make positioning of the marker much more accurate, and quicker as well. In Photoshop, you can nudge the upper layer by one pixel at a time, in any direction N,S,E,or W.
I tried several keys, and combinations of keys, but couldn't find a quick way to move the layer by a fixed amount each time. The result was that the marker wasn't always perfectly aligned for the length of the clip, and there was some jitter when I played the footage.
Posted: Thu, 28th Jan 2010, 4:57pm
Post 15 of 15
Ah okay I understand now. I think that you can manually enter cordinates in the quad, this might be the answer you're looking for but I'm not sure right now. Will try it out in a bit.