Anti-shake video filter

gigi10 · Post by **gigi10** » Fri Aug 18, 2006 3:20 pm

In my opinion, anti-shake does not work !
for example, with level 10 and enlarge size = 0, I must see now and then a black border at 1 of the 4 sides of the picture by playing the video clip.
Nothing at all.
I don't believe I am wrong.

[b]Will Ulead improve it ?[/b]

I'm testing Depan video filter, it works, but not very easy with version 8
(no possible with effect manager ...)

GeorgeW · Post by **GeorgeW** » Fri Aug 18, 2006 3:26 pm

Are you saying that it does nothing at all, or that it doesn't do enough

How much shake are you trying to shake out

Devil · Post by **Devil** » Fri Aug 18, 2006 3:43 pm

If you go to http://www.bnellis.com/msp8.1/shaketest3.wmv you can see a short clip of a handheld shot with no filtering, the Ulead, Depan, SteadyHand and Deshake. These were all adjusted for the best antishake I could obtain with them. The best was SteadyHand (with all its disadvantages) and the worst was DePan. The Ulead wasn't bad but not quite as good as SH.

gigi10 · Post by **gigi10** » Fri Aug 18, 2006 3:51 pm

It does nothing at all.

I've made several tests with it : nothing or worse (blur in the picture).
(to compare, with the same video clips, Depan gives results).

For information, the filter is not present in "Production Library" like the others (but present in the list of video filters by clicking over the clip ...).
Does that mean it is perhaps damaged or badly installed ?

rwernyei · Post by **rwernyei** » Sat Aug 19, 2006 2:46 am

Brian,
Were you eating while you were filming that?

Although, SH seemed to work the best in this test, I don't think any of them nearly achieved desirable results. With software stabilization there are too many variables: softening, overcompensation, cropping, etc... Hardware mounted stabilizers, IMO is the only way to go.

Post by **Ron P.** » Sat Aug 19, 2006 3:04 am

rwernyei wrote:Brian,
Were you eating while you were filming that? Although, SH seemed to work the best in this test, I don't think any of them nearly achieved desirable results. With software stabilization there are too many variables: softening, overcompensation, cropping, etc... Hardware mounted stabilizers, IMO is the only way to go.

I agree, either a tripod, monopod or steadi-cam is sure to get you a stable video. I use either a tripod or I have made my own steadi-cam. I learned the hard way. I've tried several software, including Depan's, Arcsoft and of course Ulead's. I found Ulead's to be as good as Depan's and better then Arcsoft, (which cost me $75 US).

Ron P.

sjj1805 · Post by **sjj1805** » Sat Aug 19, 2006 7:38 am

And if a tripod is unavailable try and make yourself into a tripod by leaning against something such as a wall, use both hands when holding the camcorder and squeeze your elbows into the sides of your body. Spread your legs slightly to get a firm stance.

If you have to zoom in/out then do so slowly. Use the camcorders eyepiece and not the LCD display.

dat7719 · Post by **dat7719** » Sat Aug 19, 2006 8:43 pm

If you are in a moving vehicle, leaning against it will make the shake worse. Not enough room for a large steady cam unless you want reflections on windows. I do a lot of taping along long drives thru the western US of the scenery along the way.

Opening the windows at 75MPH is not a good option and sometimes the road is not that smooth.

Devil · Post by **Devil** » Sun Aug 20, 2006 9:06 am

rwernyei wrote:Brian,
Were you eating while you were filming that? Although, SH seemed to work the best in this test, I don't think any of them nearly achieved desirable results. With software stabilization there are too many variables: softening, overcompensation, cropping, etc... Hardware mounted stabilizers, IMO is the only way to go.

I deliberately chose an easily visible but bad case of the shakes for the test. Explanation for the shakes: the shot was taken in the basement of the reception area of the LDS Church in SLC and maybe the spirits there were rendering me nervous, even if the Mormons are not related to the Shakers

I use a tripod whenever I can, but lugging one round is not always great fun and it's totally useless for wild animal work, cos the buggers don't stay still long enough.

pezastic · Post by **pezastic** » Sun Aug 20, 2006 10:18 am

Devil wrote:The best was SteadyHand (with all its disadvantages) and the worst was DePan. The Ulead wasn't bad but not quite as good as SH.

I agree with you on the worst, but I'm actually thinking that the Ulead filter was the best. The reason I state this is that the Ulead filter provided the best stabilization among those compared without considerable cropping. From what I can tell, the SteadyHand one zooms the picture in, working in-out, and the others are working out-in. While they're both ideal to have in a video shot, I would tend to choose control of framing over stabilization.

Devil · Post by **Devil** » Sun Aug 20, 2006 10:49 am

You have 3 modes of framing control in SH. I chose the one which zooms to fill the frame. You can choose no frame control or a limited one which replicates pixels around the edges.

sj1805 · Post by **sj1805** » Thu Sep 21, 2006 8:48 pm

On my recent Holiday my youngest lad decided to take my tripod along and mount the camcorder onto a table whilst the night time entertainment was taking place in the clubhouse.

Now on playback obviously the image was rock solid (except when my wife started clapping and knocking the table in the process - Grrr).

I have just been editing the video and found that the use of the pan and zoom filter really brings the video to life. Start off with a zoom factor of 120% and simply bung in plenty of keyframes where you centre on the action taking place.

Adjust the zoom factor every now and again dependant upon whats on stage.
Result is a nice rock steady video with gentle movement with the action.
As a bonus didn't miss the live performance either with my eye stuck in the camcorder eyepiece.

sj1805 · Post by **sj1805** » Thu Sep 21, 2006 10:32 pm

INTERESTING UPDATE

After watching the above video clip a few times I noticed that people walking about had a jerkiness to them rather like the effect experienced when you select the wrong field order.

Not to be put off I decided to experiment by using the Crop Filter instead. My 4 min 43 second clip contained 22 keyframes.
It didn't take long to create the keyframes in the same positions using the crop filter.

I then used a spreadsheet to calculate the crop % to be the same as the pan and zoom filters zoom %. A simple OXO style note
1 2 3
4 5 6
7 8 9

Enabled me to plot the positioning of the frame location.

This time there was no jerkiness - video was nice and crisp.
Another thing that I was not expecting surprised me. The render time using pan and zoom took some 45 minutes (I even took my dogs out for a walk whilst it was working.)

Using the crop filter the render time took 11 minutes 40 seconds.

So in the Crop -v- Pan & Zoom war, Crop won.