Hi there,

I'm looking for a software version of a genlock. Thing is, I want to mix a still image (32 bit, alpha included) composed of hints, guidelines and markers over live footage coming from my dv cam. I'm trying to add stuff to existing footage and match camera angles etc.

example > users.pandora.be/kristof.indeherberghe/Genlock/I0_Screenshot_Vs001.jpg

Hope someone can help me out!

Cheers,
Xtof

I don't think there is something that will mix two images live other than a mixer. Only a few software-based systems will do that and they are fairly high cost. Something on the order of a video toaster.

A still image and a video image can be mixed in a real-time editing system and the result will be available on the digital and analog inputs (assuming it is a real real-time system).

Looking at your example, I could do that in my editing system and either create a new video file or output it to some type of video display or recorder.

I'm not certain exactly what you are trying to accomplish that is out of the ordinary. If you are trying to create the overlay in real time, that is, as far as I know, not available on desktop hardware.

First of all, thanks for your solid reply! Cool to have all this knowledge in the forum :]

What I'm trying to accomplish is this:

I have a series of sequences of different classic flicks I want to add extra elements to. Replace stuff and/or add 'actors'. Already did the tracking of those clips, now have a clue of the original camera's position, angle, movement and zoomfactor/lens. Did a breakdown in a 3D program, so have a relatively accurate floorplan of the original set as well. With this data, it should be easier to match both new and old footage because the perspective should match. Only thing left is to translate the original camera to my simple DV cam.

Now, if I have a cube of fixed dimensions both in real life and inside my 3D app, I can. Mix a live stream of the cube on my set together with the fixed image of my virtual cube in simple wireframe being viewed through a virtual cam with exact same settings as the tracked one (except for angle and position) and adjust my DV cam so it'll match (within reason). Next step is to port over the position and angle of my virtual, tracked scene to the DV cam (easy since it's in meters and degrees), and I'm set to go...

I think :)

I don't think it should be too hard, assuming it can be done with DirectX? Microsoft WDM preview coming from the DV cam, mixed with a fixed 32-bit image?

Or again, am I being too optimistic?

Taa,



Xtof

I think I understand what you are trying to do.

Implicit in your posting and reply may be the answer to another question: Does this need to be accomplised in real-time?

Quite easy (to merge the images, not generate them) in an editor or compositor.

Question 2. While it is easy to overlay video, somewhere in there, it must also have some transparency or you don't see the underlying layers. 3D programs do this all the time. Live shots don't except with a keyer.

Do I understand that your layers will be:

3D program image - OK with an alpha channel
Live DV image - OK with a keyed image, otherwise overall opaque
Classic film image - Background layer, opaque OK

I'm not certain an inexpensive DV camera will be able to deliver good enough images for the live part. I'd consider feeding the video output to your mixing system from another type of camera. Best would be an old broadcast camera with 4:4:4 video from the camera head.

You will also need a real-time keyer in the setup I described. Ultimatte would work.

DV is just as good as betaSP if you are using a good keyer (like Ultimatte).