Stabilizer Sample Clips

Comments

bigrock wrote on 2/26/2010, 2:49 PM
Well I do usually use a tripod and have no need for stabilization then but there would no point in selecting that footage for a demo. I also shoot a lot while walking or from moving vehicles and that is where stabilization is quite useful.
BudWzr wrote on 2/26/2010, 3:33 PM
Yeah, sometimes the shot comes up when your tripod isn't set up, or your best camera isn't with you, or you're in a crowd of people, or on the freeway, etc., etc.

It's not like anyone WANTS to take shaky video.

BTW, This thread is so long, but what's the final consensus between products then? Not enough improvement between them to justify something paid vs. free?

I "get" the part about unrealistic expectations. Shooting from a galloping horse is out.
DGates wrote on 2/26/2010, 3:44 PM
AutoTune is a good comparison. It's like vocalists don't even try anymore because they can just modify it later.
DGates wrote on 2/26/2010, 3:45 PM
Yes, I realize you need shaky footage to compare the results. I was simply referring to camerawork in general.
johnmeyer wrote on 2/26/2010, 4:05 PM
Motion stabilization has been the subject of a large number of threads on this forum over the years. The length of these threads, combined with the number of times they get regenerated, indicates that stabilization meets a huge unmet need in the video marketplace.

As I have said in previous posts, those who assign amateur status to those who take shaky video are merely reflecting that they serve a different market. Those of us who are in the event videography business, or have to "run and gun," or who create videos consisting of material taken by amateurs, all have to deal with shaky footage.

And, sometimes even the pros get caught short and have no choice but to try to fix some bad footage.

As to unrealistic expectations, I think that expectations can and should be quite high. Especially if the video is shot with a relatively high shutter speed, there is no reason that technology cannot eventually provide near-perfect results. For example, I think someone posted the following clip in a previous thread about stabilization, showing a technology demonstration of work done at the university right next door to Sony in Madison, WI:



[edit]I just watched this all the way through for the first time in some months and am once again impressed with how well the Deshaker results compare to even this fantastic technology demo, especially when you consider that Deshaker is also doing the edge compensation trick, thus allowing you to maintain the original resolution of the shot.



JJKizak wrote on 2/26/2010, 4:14 PM
Now that's impressive.
JJK
PeterDuke wrote on 2/26/2010, 4:35 PM
Am I right in asserting that NewBlue Stabilizer does not have the option to fill in the missing border using info from adjacent frames, but deshaker can? If so Todor, can we have that option fitted to a subsequent release? Zooming in to remove the missing border is undesirable, particularly with SD video, and the other fill in methods never seem to work with my clips. Yes I know that there would be a difficulty since Vegas only passes one frame at a time to plugins.
DGates wrote on 2/26/2010, 9:31 PM
"The length of these threads, combined with the number of times they get regenerated, indicates that stabilization meets a huge unmet need in the video marketplace".

The need has not gone unmet. There have been plenty of software 'solutions' for many years. But what looks good in a tiny web demo doesn't quite translate into a realistic post tool.

metalalien wrote on 2/26/2010, 9:51 PM
Hey I uploaded my own test. The settings and side by sides are in the video.

Works good if you tone down the strength. My only problem was render time but I guess that can't be avoided.

http://www.vimeo.com/9720450
bigrock wrote on 2/26/2010, 11:47 PM
If you want to have NEwBlue take frames from the original to make the border I believe that is the replicate option Clip #4. If you watch closely in their own demo it appears they used it as well.
Jay Gladwell wrote on 2/27/2010, 4:53 AM

John, you do know, by any chance, what has become of the stabliization application those fellas developed? Is it available today?

I noticed a couple were from Abode (which I found a tad disheartening).


bigrock wrote on 2/27/2010, 12:42 PM
You said "But what looks good in a tiny web demo doesn't quite translate into a realistic post tool."

It is true that not all tools would be applicable to all situations. What you can get away with on the web is certainly much much more than what you can get away on the big screen or even a TV in the living room. Having said that I did upload every thing in 1080P so you can go full screen and give it more of an evaluation than just a "tiny web demo".

If you click twice on the video window it will take through to YouTube and you can select whatever resolution you want, all 4 options are encoded and available.
metalalien wrote on 2/27/2010, 12:53 PM
Video is still being converted but here is a 1080p test

http://www.vimeo.com/9786936
hackazoid wrote on 2/27/2010, 3:29 PM
Here's an amateur try as these threads got me to trial the New Blue product. Apologize for the sizing even at full screen, haven't done much PIP.

This is a really bumpy 25 sec clip from my wife's friend who rode in a bike carriage in NYC. Figured this would be a good test so did 3 different settings + a 4th which compared #1 to #3.



Seems to make a difference but there are some issues such as edges. BUT... I'm thinking for my non-professional use on golf trips, vacations, and parties, I can live with some of that.

1. Appreciate any advice/suggestions on settings for SD or HD clips, I'm still in R&D mode.

2. To my eye #3 seemed best---used the 'Replicate' on edges

3. Any rules of thumb on extra rendering time? Does it add 20%-30%?

regards, Hack
hackazoid wrote on 2/27/2010, 3:31 PM
My link didn't go live.

Usually if I paste and hit return it does. How do you do that here?
gpsmikey wrote on 2/27/2010, 4:44 PM
See the markup sticky at the top of the forum - basically you need to specify a link using their format:

http://www.sonycreativesoftware.com/forums/ShowMessage.asp?MessageID=521496

Your link is


mikey
DGates wrote on 2/27/2010, 5:00 PM
As we can see by this and other clips, these software programs can only do so much with excessive shake. But for minor adjustments, they'll usually do adequately.
metalalien wrote on 2/27/2010, 5:23 PM
I think the trick is to to balance the amount of smooth with the amount of crop. You can adjust all those in real time without a re-analyze. Set the strength to somewhere around 40 or above and then play with smooth/crop/border fill until you get something you like.
hackazoid wrote on 2/27/2010, 7:28 PM
GPSMIKEY.... thanks, have to admit I never read that.

Appreciate doing the posting vs just telling me about it... great service!

regards, Hack
megabit wrote on 2/28/2010, 12:24 AM
John,

Sorry for dumb question (and my laziness, preventing me from doing some search for the answer) - but:

- is the Deshaker (as the script in Vegas, interacting with VirtualDub externally) essentially the same as the "Deshaker Interface" version, which acts as an FX all inside Vegas?

AMD TR 2990WX CPU | MSI X399 CARBON AC | 64GB RAM@XMP2933  | 2x RTX 2080Ti GPU | 4x 3TB WD Black RAID0 media drive | 3x 1TB NVMe RAID0 cache drive | SSD SATA system drive | AX1600i PSU | Decklink 12G Extreme | Samsung UHD reference monitor (calibrated)

johnmeyer wrote on 2/28/2010, 7:13 PM
- is the Deshaker (as the script in Vegas, interacting with VirtualDub externally) essentially the same as the "Deshaker Interface" version, which acts as an FX all inside Vegas?

There are two really important differences between using a script (Deshaker) to apply motion stabilization, and doing the thing via a similar technology (like the NewBlue plugin) via an fX.

1. With a script, the resulting stabilized result is added to each event as a take. This lets you try several different settings and then, once the new stabilized video has been created with each setting, you can quickly do A/B/C/D/etc. comparisons simply by pressing "T" while playing back the clips.

2. With the script, you can select multiple clips -- potentially every event on the timeline -- and have them all motion stabilized while you go get a cup of coffee. Now I may be incorrect on this, since I gave up early on the NewBlue program for reasons that perhaps were not fair to Newblue, but I don't think you can batch-stabilize multiple events. Instead -- and someone needs to correct me if I'm wrong -- I think you need to "Analyze" each clip, and then when that is done, actually apply the stabilization. I am not aware of any way to tell it to batch "analyze" and then batch "apply" to all the events at once. This really slows down the workflow if you have a lot of clips to stabilize. If you only have one or two, then it makes no big difference.
bigrock wrote on 2/28/2010, 8:43 PM
FYI I am currently rendering and uploading 5 new clips that show the different products on a piece of video shot from a moving car. I will add them to the first post in this thread. I got some kinda of bizzare results. This was video that was shot properly with a mount that minimized shake so I wanted to see what improvements that could make for stuble road bumps. Check back tomorrow for that.
bigrock wrote on 3/1/2010, 1:37 AM
As promised 5 bonus clips at the top of the thread. Some interesting results.
vtxrocketeer wrote on 3/1/2010, 5:36 AM
BigRock, just viewed about 30s of your original simultaneously with the NB-stabilized. Surprisingly, the latter didn't look as good to my eyes, and a strange shake (in the stabilized footage) certainly was noticeable when the camera passed through 'mini-canyons' where the roadway had been carved through small hills. It gave me the impression that a small tremor had occurred each time you passed through these points. Weird. Maybe a tweak to the default settings would change that.

$0.02,
Steve