5 things which should be improved in the Warp Stabilizer

1. Performance.
When stabilizing any footage I always ask myself: "What the heck is he doing there?!" not even a single core is really working hard. It looks like an idle background process  The GPU is idle as well, as the analysis pass is only run on the CPU.
This maybe okay if you still need much work to do in Premiere. But often, the next steps DEPEND on the stabilization and you have to WAIT and WAIT and WAIT...
It's just painful to imagine that, running a machine with 8 cores, more than 8 times the amount of video material COULD be analyzed.
The free Deshaker for Virtualdub actually uses more cores (even though not at 100%) but analyzes at about 20-21fps compared to the Warp Stabilizer at about 4-8fps.
Admitted, the Deshakers results aren't nearly as good (especially with rolling shutter, which can be configured though, but you don't always know the right percentage for the cam you're using).
This also applies to opening projects with much stabilization data. It opens... and opens... rarely any disk activity and rarely any CPU or GPU activity. Again: WHAT IS HE DOING THERE? Just waiting for the clock itself?! Or is he sending it all to the NSA? That at least would explain the unneccesary delay:P (just kidding!)
2. Stabilization data.
For me, I wouldn't have any problem with storing it just inside a project folder or a separate file. Just add a checkbox into the plugin settings:
[x] Store stabilisation data in separate file/folder.
This may be then called [projectName].stabilization
Inside that folder, there will be maybe one file for every effect used in the project.
...or you can use the old way, if you don't have so much stabilization work to do.
3. Small bugs.
When using "stabilize only", the resulting frame, at least in Premiere Pro, has a HUGE "DC offset" sometimes. This is a term from music producing, but exactly fits in here. You can clearly see it when there are still images that shouldn't receive much of stabilization. They're somtimes shifted far out of the viewing frame, leaving big black borders, so you have to manually adjust the frame position to fit back into a "neutral" position. This could be resolved to let the whole correcting curve undergo something like a "low pass filter" which will try to keep the resulting frame SOMEHOW centered.
In this image above, from a longer clip, there isn't much motion at all. But you can see how far the offset from the actual video frame is. I have to do this for almost every video, also after I change the "smoothness" setting.
4. More control
I would LOVE to be able to disable "zoom detection", because it gives me lots of "Vertigo Effects" in many cases.
I heard this has been alredy addressed for CC.
5. Improvements / new features
One thing I always do with every stabilized video:
Instead of synthesize edges, which takes AGES to render, I just do the following:
I place the same clip, unstabilized, behind the stabilized clip (stabilized, borders only) and give the stabilized clip softened borders with "rough edges". This even still renders fine on the GPU.
This would make a great option for the stabilizer. Just call it "overlay over original with blurred edges" or something like that.
This has worked extremely well for any stabilized footage so far and isn't nearly as disturbing as the synthesized edges.
I'm considering to join the Creative Cloud, because unless I do this I don't expect to see any improvements in my old and out-dated Premiere PRO CS6...

https://www.adobe.com/cfusion/mmform/index.cfm?name=wishform to file a feature request... insert a reference to this message link

Similar Messages

  • HAS THE WARP STABILIZER CHANGED OR IMPROVED IN ANY MANNER IN After Effects 6.0?

    It was quite an improvement over point tracking alone.
    Have there been substantial further advances in the Warp Stabilizer in AE 6?
    Also, has Mocha AE changed in any manner?
    Thanks!
    Matt Dubuque

    Have there been substantial further advances in the Warp Stabilizer in AE 6?
    What would be "substantial"? No, apart from a few minor workflow and performance enhacements there have been no changes.
    Also, has Mocha AE changed in any manner?
    You still get mocha 2.x, just a slightly newer version, not mocha 3.
    Mylenium

  • I have two airports, one quite old and one brand new. Which should be connected to the modem in an extended network?

    I just got a new airport express to extend my wireless network.  It's 802.11(n).  I have an older airport-- I don't know what iteration of the 802.11 standard.
    If I connect them with Ethernet, which should be connected to the modem?  Does that answer change if I have to connect the two airports wirelessly?

    Unless I use the WAN port on the new APE, right?
    Correct. But, if you do this, the question about which device connects where is then already answered....since the "new" AirPort Express must connect to your modem. So, your best wireless performance will occur in the area near the modem. Hopefully, that is where you need the best performance.
    Side note-- the tech specs linked below for the 1st gen, draft-n APE say that it, too, is dual band, although it doesn't actually seem to be.)
    It's confusing since a device can be dual band, but not simultaneous dual band.
    1st Gen AirPort Express (AX) is dual band because it can provide a 2.4 GHz or 5 GHz network. But, it cannot provide both at the same time. It is one or the other.
    The 2nd Gen AirPort Express is simultaneous dual band because it provides both a 2.4 GHz and 5 Ghz signal at the same time.
    1)  Cable Modem --->  Switch ---> Two APE (all ethernet)
    2)  Cable Modem ---> 2nd Gen APE ---> 1st Gen APE  (all ethernet)
    3)  Cable Modem ---> one APE ~~~> other APE (wireless between APE)
    1) Will not work if you have a simple cable modem....one Ethernet port only.
    If your cable "modem" is a modem/router with multiple Ethernet ports, then you could connect both AirPort Express devices back to the modem/router....and you would not need a switch at all unless you need more Ethernet ports for other devices.
    2) would be correct if you have a simple cable modem
    3) If you want to make use of the simultaneous dual band capability of the Express, then that device must be connected directly to the modem. The 1st Gen Express will "extend" one band only.....probably the 2.4 GHz band, since it is stronger.
    If you use the 1st Gen Express connected to the modem, the "new" Express will not extend two bands. It will only extend the band that it "sees" from the 1st Gen Express, so you lose the simultaneous dual band capability of the "new" Express in this type of setup.
    If you "extend"using wireless only, there is typically a bandwidth loss of about 50%, so it is always a compromise in this type of setup.

  • How to make the warp stabilizer analyze interlaced (upper + lower field) instead of progressive?

    I am having a little problem since I am new to Premiere pro.
    Until now I always used the warp stabilizer in After Effects for single shots in need of that. But this time, I was handed a project that requires a whole lot of stabilization. My plan is to drop the whole timeline (FCP 7/ProRes422 HQ Interlaced) via XML into Premiere Pro and apply the effect there to multiple clips.
    PROBLEM: When I do this, the warp stabilizer always (no matter how I interpret the footage or set the sequence) analyzes the footage progressively. At least that is what the effect preferences tab says during analysis.
    In After Effects I always see which field is beeing analyzed (it is indicated by a L and an U behind the frame count). Not so in Premiere Pro. Here it always seems to analyze full frames.
    I made a test and compared the results from After Effects and Premiere Pro: The file generated in Premiere Pro is much softer then the one out of After Effects even though I export both with the same output settings.
    Any help would be greatly appreciated !
    Regards, franse

    I just verified and the Description for Products (and possibly other objects too) are stored in table COMM_PRSHTEXT.
    The relevant Data element is COMT_PRSHTEXTX which has the domain PRSHTEXT40.
    If you look at this domain you have a parameter "lowercase" which is available as a flag option.
    <u><b>The explanation about this Flag is the following:</b></u>
    Lowercase letters allowed/not allowed
    If this flag is set, upper case and lower case are distinguished when you enter values with screen masks.
    Otherwise all the letters entered will be converted to upper case when you enter values with a screen mask.
    This mechansim is in effect for all the fields referring to this domain.
    <b>Solution would be</b> to verify if for this data-element the same Domain is used.
    If yes, you can flag this option in the domain.
    this should solve your problem.
    kind regards
    davy pelssers
    please reward if helpful.

  • Where is the Warp Stabilizer plugin located in After Effects CS6 program files?

    I need to delete the plugin temporarily to gain access to my project because it constantly crashes on start up due to the Warp Stabilizer.

    NOt sure if you can actually remove it,but it should be in the plugins/extensions sub-folder inside the AE install folder. Anyway, try the usual troubleshooting first before taking such a radical step that may destroy AE and require a reinstall. Flush the caches, open the project with capslock and remove the effect.
    MYlenium

  • How  do you get the Warp Stabilizer VFX to work in After Effects CC?

    Upon launching the Warp Stabilizer VFX an error message appears:  "After Effects error. Couild not launch Adobe Analysis Server. (515) (25 101)"  Am running MAC OSX 10.9.2  An error message also appears on launching After Effects CC:  "Quicktime is not installed on the system. Some components will not work correctly." This OS  is supposed to have Quicktime installed. Has anyone had this same problem with the warp stabilizer and been able to solve it?  I am restoring old movies and this function is essential.
    Thanks,  Sufferincats

    You need to check your user permissions and firewall network stuff. Something is simply blocking the connections.
    Mylenium

  • I am THRILLED with the Warp Stabilizer

    I recently came back from a ten-day filming trip across the deep south. It was my first field experience using DSLR (two Canon 7Ds). I shot everything handheld with no support, with the Warp Stabilizer in mind. The results have been AMAZING beyond my expectations. Some of these shots rival the best steadi-cam, dolly and jib shots I have seen. At times, the camera seems to be floating or hovering. This is just awesome.

    We're glad you like it.

  • How to store the data, which should accessed by all the users

    I want to store some data into some variables, which has to be accessed by all the users on the webserver. i think it has to be stored in the application memory but if any body gives the code it will be clear to me .
    Thank Q
    K.Srinivas

    if it is for the same webapplication you can keep it in static attributes in a class which can be loaded at the time of server startup

  • Client, which should always listen to the server

    Hello,
    I have question regarding socket communication using standard java means. Since I would call myself a beginner in java please be patient and generous. Every hint would be appreciated.
    In my scenario I have a server that sends messages via TCP/IP from time to time to its client. The server is up and running and I cannot change it. My part is the implementation of a client that should receive the message of the server. The client will handle this message and eventually send a response or follow-up to the server.
    So far so good. The scenario is implemented and working. But my problem is the way I convinced the client to listen all the time. I use an endless while-loop (while (true) {...}) and check whether the input stream of the socket contains any data. As soon as there is any data I fetch and process this data. Afterwards the loop goes on. I am wondering whether this consumes a lot of CPU workload. Is there a better way to always listen to the server? I assume the while-loop is not very clever. Any hints would be greatly appreciated.
    Thanks in advance for your help and comments
    Slarti

    Hello pkwooster,
    thanks a lot for your quick reply.
    It was my fault that I have not been precise enough. Actually I already have one thread that listens to the server and another one that will send answers. Both are created and started in main. Nevertheless I still have an endless loop in the thread, which listens all the time.
    The application itself does not have any UI since it should run in background. So the main method creates and starts the threads. As soon as the listening thread receives data it eventually passes it to a third thread, which processes the data. Afterwards the sending thread might be called to respond to the server.
    I am sorry but I did not understand how the thread can wait for data to arrive without using an endless loop.
    Best regards, slarti

  • Question about a bug which should be fixed in the 10.2.0.4 patchset

    Hi,
    We have a 10.2.0.4 single instance database on Sun Solaris SPARC (64-bit) and have the same symptoms as described in Bug 4643723 (error codes and the stack trace are the same). That bug should be already fixed in 10.2.0.4 patchset.
    Does anyone have experience on how to proceed in such cases - I mean, is making a SR on metalink an obvious action?
    Is it possible that the patchset wasn't applied correctly although "opatch lsinventory" outputs that the "10g Release 2 Patchset 3 (10.2.0.4)" is installed and SYS.REGISTRY$HISTORY also contains a record of the upgrade. The upgrade from 10.2.0.1 to 10.2.0.4 was performed 6 months ago and the database was continuously up and runing till then.
    Thanks for any answers

    many bugs can have the same errors. Raise an SR.
    Oracle Support will ask for an opatch lsinventory, and may ask you verify library versions, or relink.

  • Old packages which should be removed with the new python package?

    I recently pacman -Syu'ed and got all the new updates Arch has to offer. In doing this, quodlibet wouldn't work anymore. Doing my own investigation, it apparently looks for libs in /usr/lib/python-2.4 before /usr/lib/python-2.5 come to find out a bunch of old packages still use /usr/lib/python-2.4 and therefore the folder sticks around and quodlibet complains it cant find what it needs in /usr/lib/python-2.4 when it is in fact in /usr/lib/python-2.5
    I then proceeded to pacman -Qo /usr/lib/python-2.4 and remove packages until i could successfully delete the folder with no packages owning it. I would then assume, that the packages I had to remove need to be rebuilt with the new python. If someone can confirm my logic here, I'll post a bug report and  hopefully this clears up some of the python migration issues that seem to be cropping up.
    [will@Ghost32 python2.5]$ sudo pacman -R pyxdg
    removing pyxdg... done.
    [will@Ghost32 python2.5]$ sudo pacman -R python-m2crypto
    removing python-m2crypto... done.
    [will@Ghost32 python2.5]$ sudo pacman -R setuptools
    removing setuptools... done.
    [will@Ghost32 python2.5]$ sudo pacman -R python-elementtreee
    error: could not find python-elementtreee in database
    [will@Ghost32 python2.5]$ sudo pacman -R python-elementtree
    removing python-elementtree... done.
    [will@Ghost32 python2.5]$ sudo pacman -R python-musicbrainz
    removing python-musicbrainz... done.
    [will@Ghost32 python2.5]$ sudo pacman -R picard
    removing picard... done.
    [will@Ghost32 python2.5]$ sudo pacman -R libtunepimp
    removing libtunepimp... done.
    [will@Ghost32 python2.5]$ sudo pacman -R python-pyicu
    removing python-pyicu... done.
    [will@Ghost32 python2.5]$ sudo pacman -R python-pylucene
    removing python-pylucene... done.
    [will@Ghost32 python2.5]$ sudo pacman -R python-dateutil
    removing python-dateutil... done.
    [will@Ghost32 python2.5]$ sudo pacman -R libnewt
    error: this will break the following dependencies:
    libnewt: is required by partimage
    [will@Ghost32 python2.5]$ sudo pacman -R partimage
    removing partimage...
    warning: /etc/partimaged/partimagedusers saved as /etc/partimaged/partimagedusers.pacsave
    Deleting user partimag
    done.
    [will@Ghost32 python2.5]$ sudo pacman -R libnewt
    removing libnewt... done.
    [will@Ghost32 python2.5]$ sudo pacman -R winki
    removing winki... done.
    [will@Ghost32 python2.5]$ sudo pacman -R boost
    error: this will break the following dependencies:
    boost: is required by wengophone-ng
    [will@Ghost32 python2.5]$ sudo pacman -R wengophone-ng
    removing wengophone-ng... done.
    [will@Ghost32 python2.5]$ sudo pacman -R boost
    removing boost... done.
    [will@Ghost32 python2.5]$ sudo pacman -R pacmon-svn
    removing pacmon-svn... done.
    [will@Ghost32 python2.5]$ sudo pacman -R python-notify
    removing python-notify... done.
    [will@Ghost32 python2.5]$ sudo pacman -R python-eyed3
    error: this will break the following dependencies:
    python-eyed3: is required by libgpod
    [will@Ghost32 python2.5]$ sudo pacman -R libgpod
    error: this will break the following dependencies:
    libgpod: is required by gtkpod
    [will@Ghost32 python2.5]$ sudo pacman -R gtkpod libgpod python-eyed3
    removing gtkpod... done.
    removing libgpod... done.
    removing python-eyed3... done.
    [will@Ghost32 python2.5]$
    and in the other terminal I had..
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by pyxdg 0.15-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by python-m2crypto 0.16-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by setuptools 0.6c3-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by python-elementtree 1.2.6-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by python-musicbrainz 2.1.2-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by picard 0.7.0-2
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by libtunepimp 0.5.0-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by python-pyicu 0.5-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by python-pylucene 2.0.0-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by python-dateutil 1.1-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by libnewt 0.52.2-3
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by winki 0.4.1-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by boost 1.33.1-5
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by pacmon-svn 14-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by python-notify 0.1.1-2
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    /usr/lib/python2.4/ is owned by python-eyed3 0.6.11-1
    [will@Ghost32 ~]$ sudo pacman -Qo /usr/lib/python2.4/
    No package owns /usr/lib/python2.4/
    [will@Ghost32 ~]$
    Edit: I just realized that a good deal of these packages could be ones that I built from the AUR, but hopefully theres a few in there from current and extra which will help the devs, and hopefully if nothing else it'll help some other people out there.
    Last edited by The_Nerd (2007-03-21 22:47:39)

    Hi,
    You can a cretae role by copying( 1st icon in application tollbar) from other one in t.code: PFCG & then need to assign role to user in  t.code: SU01.
    For reference check link:
    http://e-mory.blogspot.com/2007/12/sap-pfcg-create-role.html
    Regards,
    Biju K

  • Two things which should revert to "normal"

    Here's some reiteration of my post under "Photoshop Mac."
    I have various workstations running Mac OS 10.5.8 three run PS CS3, and one runs CS4.
    In CS4 files always open on the monitor I have allocated for my palattes.  Before CS4, Photoshop would open files on the monitor that my last filed was opened on.  This is highly annoying!!! Please bring this back in CS5!
    (Note: My menu bar is on the monitor I've designated for my palettes menu and not the monitor I do all my color work on.)
    Also, any hopes that we will be able to have a preview button for adjustment layers again?  If I'm able to do so when adjusting directly onto a layer, why don't I at least have the option to have a preview button when having an adjustment layer????  And sure, I understand I can go back in my history, but it seems unnecessary to create a whole new history state for each little adjustment you make.. then need to look at your history palette and make sure you're selecting the correct state.

    https://www.adobe.com/cfusion/mmform/index.cfm?name=wishform to file a feature request... insert a reference to this message link

  • I am not able to view the default view of FF 4.0, It is showing menu bar with the windows border. When i turn off menu bar, the strip which should be present above the tabs goes missing.

    I have recently downloaded FF 4.0 and using a lot of add-ons. I am not able to view the default view of FF 4.0 with no menu bar. Even after turning off the menu bar from customization, I am able to see only the tabs and FF button, but not the strip over the tabs.

    When you hide the menu bar to display the Firefox button, the titlebar (strip above the tabs) is only displayed if the Firefox window is not maximized. If you always want the Window titlebar, you can use the Firefox 4 UI Fixer extension - https://addons.mozilla.org/firefox/addon/firefox-4-ui-fixer

  • I have a problem with the Warp Stabilizer in Adobe Premiere Pro CC on mac

    If I want to apply the Verkrümmungsstabilisator on a clip, suddenly only a small neckline of the clip appears in the upper left corner when the effect has finished analyzing. The rest of the clip appears black. Why is that? Is this some wrong settings, I've met?

    Hi skwosnjak,
    Please try to switch the renderer to software only mode, So, go to File Menu>Project settings>General and change the Renderer option to Software only Mode.
    Regards,
    Vinay

  • FR: Big yet incredibly simple improvement to Warp Stabilizer

    I love the Warp Stabilizer effect in PP.  It's great!  But depending on the footage and the effect's settings, sometimes it can actually make things worse.  This means that any time this effect is used, users must review of the results after the effect finishes analyzing the clip.
    Here's the problem:
    Analysis takes a long time, often far too long for a user to stop editing and wait to review the results.  In my case, I keep editing, but often I forget where my stabilized clip for review is located.  In a multicam sequence, where every video clip occupies the same track, there's little to no visual cues to remind me where the Stabilized clips are.  Setting an in and/or out point as temporary markers can work if you have up to two Stabilized clips.  But what if you just stabilized 5 clips, and you want to keep editing while they're being analyzed?  How do you visually recognize which clips they are?  All current solutions add work for the editor, such as moving stabilized clips up a track and then back down after approval of the effect's results, etc.  If you have transitions between a Stabilized and an adjacent clip, or clips on a track above it, the situation gets worse.
    Here's the (incredibly simple) solution:
    Give us a visual color cue any time a clip has the Warp Stabilizer effect added to it!  This could be done through a new user selectable color category in the Preferences window's 'Label Defaults' panel.
    It's interesting to note that while 8 different colors can be set in the Label Colors tab, only 7 are currently used by PP in the Label Defaults tab, so setting a unique color to this new eighth item should be even easier to program.
    Adobe, this would make a great little addition to your next PP update!
    What do others think?

    Jim Simon wrote:
    Your request calls for easy visual recognition of that one effect, mine calls for easy visual recognition of any effect, which I feel would be far more useful to far more editors.
    Be as it may, even though I highly doubt that editors need an easier way to determine which clips have any effect applied in general, WS clips would still have to have a different color, otherwise editors wouldn't be able to differentiate WS effects from any other old effect that simply does not require the editor pause or return later to see the results.  I've explained this pretty clearly but it seems you're determined to not get it.
    Jim Simon wrote:
    Or, if you don't want to wait for the effect to finish, add it after editing, as most editors do with most effects anyway.  Then you'll have the time to wait to see the results.
    Adding effects is part of editing, or isn't it?  In case you don't know, editors work differently.  If I want to add effects as I do my cuts, what's wrong with that?  The software should be intelligent and flexible enough to accommodate a user's workflow.  And if it isn't, then it should be improved.  If you're an advocate against this kind of progress and think users should adapt to and accept a programs limitations, rather than speak up, which version of Premiere are you using?  4.1?
    Jim Simon wrote:
    Mark them, as you can do now. 
    The whole point of this thread and FR is for Premiere to remove steps from our workflow!  Still don't get it, do you?
    It seems that every time I expose a limitation in Premiere or an area for improvement, you suggest that I should change my workflow rather than have the software improve in a way that could benefit countless other editors.  At this point, I really have to ask... what's in it for you Jim?  What's your point?
    "Then you'll have time to wait to see the results"  Is that what you do, wait 10 minutes for a long clip to finish analyzing so you can see the WS results?  You're just sounding more and more ridiculous IMHO.  It would be nice if overall you made more intelligent contributions to my threads.
    Frankly, I'm getting tired of this nonsense Jim.  If you have something intelligent to say, please, by all means.  If you don't, please stop diluting the message of my threads with your nonsense arguments/comments.  If you want all clips with effects applied to them to be labeled with the same color, and thus make WS clips visually indistinguishable (which is the current problem I'm asking Adobe to fix), go ahead and start your own thread.

Maybe you are looking for