Re: [Nuke-users] BBox after stabilisation

Frank Rueter|OHUfx Sat, 30 Apr 2016 21:59:24 -0700

I guess the trick is also to find the optimal centre for the scaling tokeep the actual required scale factor to a minimum.


On 30/04/16 10:31 pm, Justin GD wrote:

Sorry its was Firday... Forgot to treat absolute values for horizontalcase....I also, have troubles when the image is scaled before. I'll have torewrite the algorithm properly at some point. Or if anyone as timebefore.
Basically my idea was :

  * Get bboxes in dict for the frame range.
  * Compare abs values for x and y, regarding the distance where
    pixels are black from bbox to format  (don't forget y*FormatRatio
    to compare properly with x values)
  * Take the maximum value.
  * Retrieve bbox frame where max value is.
  * Get bbox (width or height) ; calculate bbox (width or height) plus
    max value.
  * Determine Scale factor
  * Feed Transform with scale factor.
As I said this (corrected) worked for my case, but need to test withany cases. It's was really quickly coded but maybe it will give youguys ideas...
set cut_paste_input [stack 0]

version 9.0 v8

push $cut_paste_input

Group {

 name JFX_OptBbox

 tile_color 0xa373b5ff

 selected true

 xpos -231

 ypos 70

 addUserKnob {20 User}
addUserKnob {52 embedded_callbacks l INVISIBLE -STARTLINE +INVISIBLET "#This hidden knob is only here to keep the indented callback partsfor further dev#\n#Though these lines are converted in a single oneand inserted directly in the .gizmofile#\n\n#onKnobChanged#\n\n'''\n\n\n#OnCreate#\nthisNode =nuke.thisNode()\nwith thisNode:\nthisNode\['startFrame'].setValue(nuke.Root().firstFrame())\nthisNode\['endFrame'].setValue(nuke.Root().lastFrame())\n\n\n'''"}
 addUserKnob {3 startFrame l "Frame Range"}

 addUserKnob {3 endFrame l "" -STARTLINE}

 endFrame 100
addUserKnob {22 AutoScale l "<b>Auto - scale</b>" T "from __future__import with_statement\nimport sys\n\n\n\nthisNode =nuke.thisNode()\nwith thisNode :\n\n thisNodeInput =thisNode.input(0)\n cv = nuke.toNode('CurveToolGenericOptBbox')\ntf = nuke.toNode('TransformGenericOptBbox')\n\n safePixels = 3 # 3pixels from the balck outside of the cornerpin that grow the bbox + 1safe pixel\n first\t\t= thisNode\['startFrame'].value()\nlast\t\t= thisNode\['endFrame'].value()\n total \t\t= int(last) -int(first)+1\n maxNum = 0\n max_x = 0\n max_y = 0\nbbox_x_list = \[]\n bbox_y_list = \[]\n bbox_r_list = \[]\nbbox_t_list = \[]\n bbox_dict = \{\}\n f_x =thisNodeInput.format().x()\n f_y = thisNodeInput.format().y()\nf_r = thisNodeInput.format().r()\n f_t =thisNodeInput.format().t()\n formatRatio = (f_r - f_x)/(f_t -f_y)\n tf\['center'].setValue((f_r - f_x)/2,0)\ntf\['center'].setValue((f_t - f_y)/2,1)\n optScaleFactor = 1\nprogress \t= 0\n progressBar \t= nuke.ProgressTask(\"Gettingbbox...\")\n\n \n for frame in range(int(first),int(last)+1):\n\n # Executing a CurveTool node is just to forcethe viewer to update so we can analyse on a range of frames.\nnuke.execute(cv.name <http://cv.name>(),start=frame,end=frame)\n\nx = int(thisNodeInput.bbox().x())\n y =int(thisNodeInput.bbox().y())\n r =int((thisNodeInput.bbox().w() + thisNodeInput.bbox().x()) )\n t= int((thisNodeInput.bbox().h() + thisNodeInput.bbox().y()) )\nw = int(thisNodeInput.bbox().w())\n h =int(thisNodeInput.bbox().h())\n\n\n bbox_dict\[frame] = \[x, y,r, t, w, h]\n\n # ProgressBar\n ifprogressBar.isCancelled():\n del progressBar\nsys.exit()\n\n\n progress += 1\nprogressBar.setProgress(int((float(progress) / float(total)) *100))\nif (progress) >= total:\n del progressBar\n\n \nfor frame, val_list in bbox_dict.iteritems():\nbbox_x_list.append(val_list\[0]-f_x)\nbbox_y_list.append(val_list\[1]-f_y)\nbbox_x_list.append(val_list\[2]-f_r)\nbbox_y_list.append(val_list\[3]-f_t)\n\n \n\n if max(abs(i) fori in bbox_x_list) >= max(abs(i) for i in bbox_y_list)*formatRatio: \nmaxNum = max(abs(i) for i in bbox_x_list)\n for frame,val_list in bbox_dict.iteritems():\n if (maxNum ==(val_list\[0]-f_x) or \n maxNum == -(val_list\[0]-f_x)or\n maxNum == (val_list\[2]-f_r) or\nmaxNum == -(val_list\[2]-f_r)):\n #optScaleFactor =((val_list\[4]+ val_list\[2]+ val_list\[0])/(f_r - f_x))\n\n \n optScaleFactor = ((abs(val_list\[4])+abs(val_list\[0])*2)/abs(val_list\[4]))\n \ntf\['center'].setValue((val_list\[2]+val_list\[0])/2,0)\ntf\['center'].setValue((val_list\[3]+val_list\[1])/2,1)\nprint 'x'\n print frame\n print(val_list\[0]-f_x, val_list\[1]-f_y, val_list\[2]-f_r,val_list\[3]-f_t, val_list\[4], val_list\[5])\n print(val_list\[0], val_list\[1], val_list\[2], val_list\[3], val_list\[4],val_list\[5])\n print optScaleFactor\n else:\nmaxNum = max(abs(i) for i in bbox_y_list) \n for frame,val_list in bbox_dict.iteritems():\n if (maxNum ==(val_list\[1]-f_y) or \n maxNum == -(val_list\[1]-f_y)or\n maxNum == (val_list\[3]-f_t) or\nmaxNum == -(val_list\[3]-f_t)):\n #optScaleFactor =((val_list\[5]+val_list\[3]+val_list\[1]) /(f_t - f_y))\n\n optScaleFactor =((abs(val_list\[5])+abs((val_list\[1]*2))) /abs(val_list\[5]))\n\ntf\['center'].setValue((val_list\[2]+val_list\[0])/2,0)\ntf\['center'].setValue((val_list\[3]+val_list\[1])/2,1)\nprint 'y'\n print frame\n print(val_list\[0]-f_x, val_list\[1]-f_y, val_list\[2]-f_r,val_list\[3]-f_t, val_list\[4], val_list\[5])\n print(val_list\[0], val_list\[1], val_list\[2], val_list\[3], val_list\[4],val_list\[5])\n print optScaleFactor\n\n\n\ntf\['scale'].setValue(optScaleFactor)\n\n\n\n\n\n\n" +STARTLINE}
 addUserKnob {26 ""}
addUserKnob {26 help_1 l "" +STARTLINE T "<divstyle='color:#ffd200;'>Scale the image to get theoptimal<br/>\nbounding box for the indicated frame range.</div>"}
}

 Input {

  inputs 0

  name Input1

  xpos 0

 }

 Transform {

  scale 1.198751301

  center {770 540}

  shutteroffset centred

  name TransformGenericOptBbox

  xpos 0

  ypos 139

 }

 Output {

  name Output1

  xpos 0

  ypos 300

 }

 CurveTool {

  inputs 0

  ROI {480 270 1440 810}

  autocropdata {480 270 1440 810}

  intensitydata {{curve} {curve} {curve} 0.3067211802}

  name CurveToolGenericOptBbox

  xpos -179

  ypos 122

 }

end_group
2016-04-30 4:17 GMT+01:00 Frank Rueter|OHUfx <fr...@ohufx.com<mailto:fr...@ohufx.com>>:
    I just tried this with a animated Transform node and it ended up
    scaling things down rather than up:

    CheckerBoard2 {
     inputs 0
     name CheckerBoard1
     selected true
     xpos 155
     ypos -236
    }
    BlackOutside {
     name BlackOutside1
     selected true
     xpos 155
     ypos -164
    }
    Transform {
     translate {{curve x2 -70 x11 -5.555648804 x21 -20 0 x36 -190} 0}
     center {1440 1080}
     name Transform1
     selected true
     xpos 155
     ypos -138

    }




    On 04/30/2016 05:22 AM, Justin GD wrote:
    Awesome !

    I had a play today as well; Here is a beta group node that worked
    for the plates I had. Didn't test it properly yet.
    I'll have a look at your code thanks Brian !

    Cheers,
    Justin

    set cut_paste_input [stack 0]
    version 9.0 v7
    push $cut_paste_input
    Group {
    name JFX_OptBbox
    tile_color 0xa373b5ff
    selected true
    xpos 1980
    ypos 497
    addUserKnob {20 User}
    addUserKnob {52 embedded_callbacks l INVISIBLE -STARTLINE
    +INVISIBLE T "#This hidden knob is only here to keep the indented
    callback parts for further dev#\n#Though these lines are
    converted in a single one and inserted directly in the .gizmo
    file#\n\n#onKnobChanged#\n\n'''\n\n\n#OnCreate#\nthisNode =
    nuke.thisNode()\nwith thisNode:\n
     thisNode\['startFrame'].setValue(nuke.Root().firstFrame())\n
     thisNode\['endFrame'].setValue(nuke.Root().lastFrame())\n\n\n'''"}
    addUserKnob {3 startFrame l "Frame Range"}
    addUserKnob {3 endFrame l "" -STARTLINE}
    endFrame 100
    addUserKnob {22 AutoScale l "<b>Auto - scale</b>" T "from
    __future__ import with_statement\nimport sys\n\n\n\nthisNode =
    nuke.thisNode()\nwith thisNode :\n\n  thisNodeInput =
    thisNode.input(0)\n    cv =
    nuke.toNode('CurveToolGenericOptBbox')\n    tf =
    nuke.toNode('TransformGenericOptBbox')\n\n  safePixels = 3 # 3
    pixels from the balck outside of the cornerpin that grow the bbox
    + 1 safe pixel\n    first\t\t= thisNode\['startFrame'].value()\n
       last\t\t= thisNode\['endFrame'].value()\n    total \t\t=
    int(last) - int(first)+1\n    maxNum = 0\n  max_x = 0\n    max_y
= 0\n bbox_x_list = \[]\n bbox_y_list = \[]\nbbox_r_list = \[]\n bbox_t_list = \[]\n bbox_dict = \{\}\n
       f_x = thisNodeInput.format().x()\n  f_y =
    thisNodeInput.format().y()\n    f_r =
    thisNodeInput.format().r()\n    f_t =
    thisNodeInput.format().t()\n    formatRatio = (f_r - f_x)/(f_t -
    f_y)\n  tf\['center'].setValue((f_r - f_x)/2,0)\n
     tf\['center'].setValue((f_t - f_y)/2,1)\n  optScaleFactor = 1\n
       progress \t= 0\n  progressBar \t= nuke.ProgressTask(\"Getting
    bbox...\")\n\n    \n    for frame in range(int(first),
    int(last)+1):\n\n        # Executing a CurveTool node is just to
    force the viewer to update so we can analyse on a range of
    frames.\n        nuke.execute(cv.name
<http://cv.name>(),start=frame,end=frame)\n#nuke.Root().setFrame( frame )\n #time.sleep( 0.1 )\n#nuke.updateUI\n x = int(thisNodeInput.bbox().x())\n y =
    int(thisNodeInput.bbox().y())\n        r =
int((thisNodeInput.bbox().w() + thisNodeInput.bbox().x()) )\nt = int((thisNodeInput.bbox().h() + thisNodeInput.bbox().y())
    )\n        w = int(thisNodeInput.bbox().w())\n        h =
    int(thisNodeInput.bbox().h())\n\n\n  bbox_dict\[frame] = \[x, y,
    r, t, w, h]\n\n    # ProgressBar\n        if
progressBar.isCancelled():\n del progressBar\nsys.exit()\n\n\n progress += 1\n
     progressBar.setProgress(int((float(progress) / float(total))
    *100))\n        if (progress) >= total:\n            del
    progressBar\n\n  \n    for frame, val_list in
    bbox_dict.iteritems():\n  bbox_x_list.append(val_list\[0]-f_x)\n
     bbox_y_list.append(val_list\[1]-f_y)\n
     bbox_x_list.append(val_list\[2]-f_r)\n
     bbox_y_list.append(val_list\[3]-f_t)\n\n  \n\n    if max(abs(i)
    for i in bbox_x_list) >= max(abs(i) for i in
    bbox_y_list)*formatRatio: \n        maxNum = max(abs(i) for i in
    bbox_x_list)\n        for frame, val_list in
    bbox_dict.iteritems():\n        if (maxNum == (val_list\[0]-f_x)
or \n maxNum == -(val_list\[0]-f_x) or\nmaxNum == (val_list\[2]-f_r) or\n maxNum ==
    -(val_list\[2]-f_r)):\n                #optScaleFactor =
((val_list\[4]+ val_list\[2]+ val_list\[0])/(f_r - f_x))\n\n \n optScaleFactor = ((val_list\[4]+
    val_list\[0]*2)/(val_list\[4]))\n  \n
     tf\['center'].setValue((val_list\[2]+val_list\[0])/2,0)\n
tf\['center'].setValue((val_list\[3]+val_list\[1])/2,1)\nprint 'x'\n print frame\nprint (val_list\[0]-f_x, val_list\[1]-f_y, val_list\[2]-f_r,val_list\[3]-f_t, val_list\[4], val_list\[5])\nprint (val_list\[0], val_list\[1], val_list\[2], val_list\[3],
    val_list\[4], val_list\[5])\n  print optScaleFactor\n    else:\n
           maxNum = max(abs(i) for i in bbox_y_list)        \n    for
    frame, val_list in bbox_dict.iteritems():\n            if (maxNum
== (val_list\[1]-f_y) or \n maxNum == -(val_list\[1]-f_y) or\nmaxNum == (val_list\[3]-f_t) or\n maxNum ==
    -(val_list\[3]-f_t)):\n    #optScaleFactor =
((val_list\[5]+val_list\[3]+val_list\[1]) /(f_t - f_y))\n\n optScaleFactor =
    ((abs(val_list\[5])+abs((val_list\[1]*2)))  /abs(val_list\[5]))\n
                   \n
     tf\['center'].setValue((val_list\[2]+val_list\[0])/2,0)\n
tf\['center'].setValue((val_list\[3]+val_list\[1])/2,1)\nprint 'y'\n print frame\nprint (val_list\[0]-f_x, val_list\[1]-f_y, val_list\[2]-f_r,val_list\[3]-f_t, val_list\[4], val_list\[5])\nprint (val_list\[0], val_list\[1], val_list\[2], val_list\[3],
    val_list\[4], val_list\[5])\n  print optScaleFactor\n\n\n\n
     tf\['scale'].setValue(optScaleFactor)\n\n\n\n\n\n\n" +STARTLINE}
    addUserKnob {26 ""}
    addUserKnob {26 help_1 l "" +STARTLINE T "<div
    style='color:#ffd200;'>Scale the image to get the
    optimal<br/>\nbounding box for the indicated frame range.</div>"}
    }
    Input {
    inputs 0
    name Input1
    xpos 0
    }
    Transform {
    center {965 495}
    shutteroffset centred
    name TransformGenericOptBbox
    xpos 0
    ypos 139
    }
    Output {
    name Output1
    xpos 0
    ypos 300
    }
    CurveTool {
    inputs 0
    ROI {480 270 1440 810}
    autocropdata {480 270 1440 810}
    intensitydata {{curve} {curve} {curve} 0.3067211802 <tel:3067211802>}
    name CurveToolGenericOptBbox
    xpos -179
    ypos 122
    }
    end_group


    2016-04-29 17:40 GMT+01:00 Brian Willard <br...@radarstudios.com
    <mailto:br...@radarstudios.com>>:

        I built a version of this the the other day using the
        curveTool hack to sum the bbox across a range of frames. It
        worked ok, but I noticed it didn't return the same values for
        the bbox consistently.

        A much faster method seems to be to shuffle black into the
        rgba, then frameblend the range to get the bbox sum. After
        that, increasing the scale in a reformat by 0.1 until the
        summed bbox fits into the reformat's width & height. I'm
        using a 2nd reformat expression linked to the 1st to return
        to the original format size.

        let me know if this is what you we're thinking of, or if you
        have any suggestions!

        https://gist.github.com/mybikeislost/3a28fbc5253b73f3db52160d655fe742

        Brian


        On Tue, Apr 26, 2016 at 3:37 AM, Justin GD
        <j.grosde...@gmail.com <mailto:j.grosde...@gmail.com>> wrote:

            Yep, thanks for your answers; I'll post it and will have
            a look at it when I have more time !

            Cheers,
            J

            2016-04-26 5:17 GMT+01:00 Frank Rueter|OHUfx
            <fr...@ohufx.com <mailto:fr...@ohufx.com>>:

                I reckon that's one for the nukepedia wishlist
                <http://www.nukepedia.com/uncategorized/tool-wish-list>.




                On 04/26/2016 02:46 PM, Mike Frank wrote:
                What would really be magic is if a script knew the
                optimal center point to scale from too.

                On Apr 25, 2016 3:14 PM, "Frank Rueter|OHUfx"
                <fr...@ohufx.com <mailto:fr...@ohufx.com>> wrote:

                    I have wanted to look into this for ever and
                    never had the time.
                    I reckon it should be totally possible with a
                    bit of math magic.

                    On 04/26/2016 01:42 AM, Justin GD wrote:
                    Hi all,

                    Is there a way to scale up a plate to the
                    optimal scale after stabilisation ?
                    What I want is the bbox to always be bigger
                    than my format. I can scale up by hand but I
                    was just wondering if there was an automatic
                    way of doing this (expressions?).

                    Cheers,
                    Justin


                    _______________________________________________
                    Nuke-users mailing list
                    Nuke-users@support.thefoundry.co.uk
                    
<mailto:Nuke-users@support.thefoundry.co.uk>,http://forums.thefoundry.co.uk/
                    
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
                    _______________________________________________
                    Nuke-users mailing list
                    Nuke-users@support.thefoundry.co.uk
                    <mailto:Nuke-users@support.thefoundry.co.uk>,
                    http://forums.thefoundry.co.uk/
                    
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users



                _______________________________________________
                Nuke-users mailing list
                Nuke-users@support.thefoundry.co.uk
                
<mailto:Nuke-users@support.thefoundry.co.uk>,http://forums.thefoundry.co.uk/
                
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
                _______________________________________________
                Nuke-users mailing list
                Nuke-users@support.thefoundry.co.uk
                <mailto:Nuke-users@support.thefoundry.co.uk>,
                http://forums.thefoundry.co.uk/
                
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users



            _______________________________________________
            Nuke-users mailing list
            Nuke-users@support.thefoundry.co.uk
            <mailto:Nuke-users@support.thefoundry.co.uk>,
            http://forums.thefoundry.co.uk/
            http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users



        _______________________________________________
        Nuke-users mailing list
        Nuke-users@support.thefoundry.co.uk
        <mailto:Nuke-users@support.thefoundry.co.uk>,
        http://forums.thefoundry.co.uk/
        http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users




    _______________________________________________
    Nuke-users mailing list
    Nuke-users@support.thefoundry.co.uk
    <mailto:Nuke-users@support.thefoundry.co.uk>,http://forums.thefoundry.co.uk/
    http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
    _______________________________________________
    Nuke-users mailing list
    Nuke-users@support.thefoundry.co.uk
    <mailto:Nuke-users@support.thefoundry.co.uk>,
    http://forums.thefoundry.co.uk/
    http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users




_______________________________________________
Nuke-users mailing list
Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users


--
ohufxLogo 50x50 <http://www.ohufx.com>    

*vfx for storytellers <http://www.ohufx.com>*

*vfx compositing <http://ohufx.com/index.php/vfx-compositing> |*workflow customisation & consulting<http://ohufx.com/index.php/vfx-customising>**


*W E L L I N G T O N    |    N E W   Z E A L A N D *

_______________________________________________
Nuke-users mailing list
Nuke-users@support.thefoundry.co.uk, http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users

Re: [Nuke-users] BBox after stabilisation

Reply via email to