pan/crop: slow-in, slow-out

Serena wrote on 11/9/2006, 2:14 AM
Despite the various options for keyframe interpolation I find it difficult to create a move or zoom that ramps smoothly up/down to/from speed. So a zoom, for example, comes to a fairly hard stop rather than decelerating to a stop. In FCP this facility appears to be called slow-in and slow-out (because a producer has just asked me to do that and provided an example that does look a lot better than I've been able to achieve). I've tried adding adjacent keyframes but first efforts haven't been encouraging.
Any ideas on this?

Comments

busterkeaton wrote on 11/9/2006, 2:21 AM
Pan Crop, Then velocity envelope?
Grazie wrote on 11/9/2006, 2:30 AM
BK's got it. I did this recently. Works like a charm.


Here's the tip Serena: Have the sequence on a loop and Preview AS you adjust the 2 top level parameters - here TM and Pan/crop. You can now sit and control the two until you get the "story" you want. It's great fun to watch and do.

Hey! Now there's another tool for Madison? Virtual Video Mixing levers!! "Cue Zoom 3, Easy . .. Cue zoom 2 .. easy" - I like this!!

farss wrote on 11/9/2006, 2:39 AM
I suspect part of the problem is the old issue of working with interlaced video. To keep resolution the frame needs to be de-interlaced prior to scaling but that's rather problematic. Very slow zooms would really highlight this problem with interlaced video, especially if there's lots of motion in the frame.
Shooting progressive would be s good start and I'd be checking if that's not what the 'opposition' had to start with.

Bob.
Grazie wrote on 11/9/2006, 2:48 AM
The other thing, Bob, we don't know yet if Serena has rendered it - see? I find post rendered stuff looks spectacular. When I get here I just trust to the Vegas-God to do the "thing".
farss wrote on 11/9/2006, 2:51 AM
And we've both forgotten about the type of keyframe interpolation and the Smoothness setting.
farss wrote on 11/9/2006, 3:46 AM
Well I just tried to repro her problem and I can see it very clearly.
With one or the ther settings of smoothness should produce a smooth acceleration but it makes not a tad of difference.

I tried using 4 keframes with one very close to each end of the zoom, still only a sharp change in acceleration, no smoothing at all. So then I tried making the nearby keyframe only slight zooms and still sudden changes in rate. Interpolation should turn the sawtooth 'curve' into a 'S' curve but is sure aint working from what I can see.
I seem to to work just fine for position but not magnification, wierd.
Another bug maybe?

Bob.
Tech Diver wrote on 11/9/2006, 5:59 AM
I brought up this issue a number times over the last year. A few folks have written some scripts to simulate smooth motion with a series of small linear moves for some specific cases of motion. However, the real solution is to have Sony implement BEZIER ENVELOPES. This is at the very top of my Vegas wish list.
TorS wrote on 11/9/2006, 6:33 AM
Maybe I am reading your question wrongly, but isn't this only a matter of changing the values for the keyframes (smooth, hard, slow etc)?
Tor
farss wrote on 11/9/2006, 10:55 AM
That's what I thought originally but there seems to be no such settings for event pan/crop.

Bob.
jetdv wrote on 11/9/2006, 11:07 AM
I just opened Pan/Crop, added a new point on the pan/crop timeline, and right-clicked that point. Several entries are available including: Linear, Fast, Slow, Smooth, Sharp, and Hold. Using "Slow" or "Smooth", combined with the "Smoothness" setting should allow it to slowly start/stop.
farss wrote on 11/9/2006, 11:27 AM
I think you're right. With Smooth the zoom eases in and out, one extra keyframe near the start and stop of the zoom with an intermediate value seems to make it a bit smoother but the interpolation does seem to work correctly with the keyframes at smooth.
Tattoo wrote on 11/9/2006, 1:07 PM
I can't tell if the "smoothness" setting does anything for the zoom/crop. It definitely affects pans (to visuallize set three keyframes. move straight right to the second keyframe, and then straight up to the third keyframe. With Smoothness of 0, it's a hard 90 degree change. With Smoothness of 100, the motion is a rounded blend to up). I don't see a difference between 0 & 100 when conducting a zoom test, but that a very quick, very unscientific test.

B
TorS wrote on 11/9/2006, 1:30 PM
The thing to look out for though, if you are using pan and zoom at the same time. The smothness will also try to smooth out the curve. Zoom out from somewhere not in the centre and pan to centre as you reach full frame. You may get a few frames where the image is a little off - creating a black line along one of the edges - before it settles nicely where it should.
Tor
Serena wrote on 11/9/2006, 3:50 PM
Right --- thanks for all the good inputs on this. I posted the query, left the internet and went back to struggling with other aspects of a somewhat tedious job. Now this morning the sun is shining and you've solved that particular problem. Probably I should have mentioned I was trying to do a pan and zoom to liven up an uninteresting still that the producer insists be included (because it shows a locally notable personality). So the velocity envelope doesn't do anything. Ah, yes I was doing RAM render preview to check the consequences of each fiddle. It was the "smoothness" adjustment that I was failing to use. Rather shame-faced to have to admit that after all this time I didn't know that was there. Didn't even notice when I clicked around keyframe interpolation thinking "there must be more to this". Should take my own advice: read the manual!
Thanks everyone.
Serena wrote on 11/9/2006, 4:12 PM
Having got the motion transients smooth for this zoom with pan, I realise that there isn't any independent control over how the zoom and pan is managed. In this particular case (starting with stationary frame, then pan/zoom to new framing, hold) that I see the zoom beginning, pan, finish zoom; I would prefer the zoom and pan to start and finish together. The difference is small and of little concern, but noticable. Is there something else I don't know but should? An unknown unknown? Or, since I've asked the question, maybe it's a known unknown.
farss wrote on 11/9/2006, 4:37 PM
If the same keyframes control both they should start and end togther. All I can think of is it's very easy to not see that you've got two keframes instead of one unless you zoom right in, been caught out by that a few times.
Serena wrote on 11/9/2006, 4:47 PM
Agree -- I'm forever generating superimposed keyframes. But not this time. It almost seems as if there is a zoom function and a traverse function, but I wouldn't expect this to be so (however it wouldn't be unreasonable to do the maths that way). The difference I'm mentioning is quite small.

EDIT: My description is probably confusing -- the zoom and pan do run simultaneously, but (taking them as separate functions) the zoom takes a little longer than the pan. The zoom starts before and ends after.
farss wrote on 11/9/2006, 5:02 PM
What you're seeing could be from how the best smooth fit interpolation is working. There's only one Smoothness type and value per keyframe. Possibly creating two very close keyframes and adjusting the values would solve the problem?