VideoHelp Forum




+ Reply to Thread
Page 2 of 2
FirstFirst 1 2
Results 31 to 49 of 49
  1. Simple, about the same burriness as _Al_:
    Code:
    BilinearResize(24, 12)
    Blur(1.0)
    Spline36Resize(1920,1080)
    Adjust saturation and contrast too if you want.
    Quote Quote  
  2. Ohhhhh wow thanks!!
    Quote Quote  
  3. I was about to suggest something similar to jagabo, but as he beat me to it I thought I'd just add....
    If you want the background more symmetrical, which might be less distracting, try flipping the video over and blending it into itself first.

    Just a frame from a video I happen to have open.

    BilinearResize(24, 12)
    Blur(1.0)
    Spline36Resize(1280,720)

    Image
    [Attachment 84962 - Click to enlarge]


    V = last
    Overlay(V, FlipHorizontal(V), Opacity=0.5)
    BilinearResize(24, 12)
    Blur(1.0)
    Spline36Resize(1280,720)

    Image
    [Attachment 84963 - Click to enlarge]
    Quote Quote  
  4. Originally Posted by _Al_ View Post
    It worked, only I had to disable that AddGrain line for myself, because I do not have plugin, but you might have it.
    mm I think have some different versions

    Image
    [Attachment 84968 - Click to enlarge]


    however it seems works well
    Quote Quote  
  5. so finally

    this is me

    Image
    [Attachment 84969 - Click to enlarge]


    ----------------------------------------------------------------------------------------------

    using:


    Code:
    LoadPlugin("v:\automazioneclip\AviSynth\LSMASH_AGO_2023\x64\LSMASHSource.dll")
    
    LWLibavVideoSource("C:\Users\Administrator\Desktop\Nuova cartella (6)\catLearn.jpg")
    
    function QGaussBlur(clip C, float radx, float rady)
    
    
    {
        return C.BilinearResize(
        \           Max(16, _mod(2, Float(C.Width)/radx)), 
        \           Max(16, _mod(2, FLoat(C.Height)/rady)))
        \       .GaussResize(
        \           C.Width, C.Height, p=19)
    }
    
    function _mod(int m, float f)
    {
        x  = (m>0) ? -1 : 1
        m  = Max(1, Abs(m))
        sg = Sign(f)
        i  = Round(Abs(f))
        return sg * Max(0, i + x * (i % m))
    }
    
    function f2c(float f) {
        return Round((f - 1.0) * 256.0) 
    }
    
    #LoadPlugin("C:\ ...   path   ... \AddGrain.dll")
    #LWLibavVideoSource("catLearn.jpg")
    GaussResize(1920, 1080, p=19)
    blur = 100
    gain = 0.9
    blue = 0.2
    noise = 3
    sat = 0.5
    QGaussBlur(blur, blur).QGaussBlur(blur, blur)
    ColorYUV(gain_y=f2c(gain), off_u=blue, cont_u=f2c(sat), cont_v=f2c(sat))
    AddGrain(Float(noise), 0.22, 0.22, constant=1>0)
    this is the final result 1

    Image
    [Attachment 84970 - Click to enlarge]


    ----------------------------------------------------------------------------------------------

    using:


    Code:
    LoadPlugin("v:\automazioneclip\AviSynth\LSMASH_AGO_2023\x64\LSMASHSource.dll")
    
    LWLibavVideoSource("C:\Users\Administrator\Desktop\Nuova cartella (6)\catLearn.jpg")
    
    
    
    crop(0,height/4, 0, -height/4)
    bicubicresize(24,12)
    bicubicresize(1920,1080)
    this is the result 2

    Image
    [Attachment 84971 - Click to enlarge]


    ----------------------------------------------------------------------------------------------

    using:

    Code:
    LoadPlugin("v:\automazioneclip\AviSynth\LSMASH_AGO_2023\x64\LSMASHSource.dll")
    
    LWLibavVideoSource("C:\Users\Administrator\Desktop\Nuova cartella (6)\catLearn.jpg")
    
    
    
    V = last
    Overlay(V, FlipHorizontal(V), Opacity=0.5)
    BilinearResize(24, 12)
    Blur(1.0)
    Spline36Resize(1920,1080)
    this is the result 3

    Image
    [Attachment 84972 - Click to enlarge]


    ----------------------------------------------------------------------------------------------

    the cat thanks everything and asks: considering that this should be the background to "animate" a possible photo of the cat in the center of the screen could there be a better alternative? otherwise it seems to me that in some way these results could also be fine
    Quote Quote  
  6. Oh, yes, thanks for that correction, I did not notice that "C" at the end and also I put it into plugins+ folder, not plugins64+, now it is working
    Quote Quote  
  7. Originally Posted by marcorocchini View Post
    considering that this should be the background to "animate" a possible photo of the cat in the center of the screen could there be a better alternative? otherwise it seems to me that in some way these results could also be fine
    I do not know what you mean,

    but btw. that script using "FuzzyBorders" could be used like that, just importing all functions needed by importing that FuzzyBorders.avsi:
    Code:
    import("C:\Program Files (x86)\AviSynth+\plugins64+\FuzzyBorders.avsi")
    blur = 100
    gain = 0.9
    blue = 0.2
    noise = 3
    sat = 0.5
    LWLibavVideoSource("catLearn.jpg")
    GaussResize(1920, 1080, p=19)
    QGaussBlur(blur, blur).QGaussBlur(blur, blur)
    ColorYUV(gain_y=f2c(gain), off_u=blue, cont_u=f2c(sat), cont_v=f2c(sat))
    AddGrain(Float(noise), 0.22, 0.22, constant=1>0)
    or if that FuzzyBorder.avsi is in that plugins64+ folder, it loads automatically, without using import
    Last edited by _Al_; 18th Jan 2025 at 17:22.
    Quote Quote  
  8. for example: the script simply has to output the blurred image without anything else. Subsequently I could, with some good video editing program, generate a video using only the background image and moving only the photo
    Quote Quote  
  9. Code:
    import("v:\automazioneclip\AviSynth\plugins\FuzzyBorders.avsi")
    
    LoadPlugin("v:\automazioneclip\AviSynth\LSMASH_AGO_2023\x64\LSMASHSource.dll")
    
    LWLibavVideoSource("C:\Users\Administrator\Desktop\Nuova cartella (6)\catLearn.jpg")
    
    blur = 100
    gain = 0.9
    blue = 0.2
    noise = 3
    sat = 0.5
    
    GaussResize(1920, 1080, p=19)
    QGaussBlur(blur, blur).QGaussBlur(blur, blur)
    ColorYUV(gain_y=f2c(gain), off_u=blue, cont_u=f2c(sat), cont_v=f2c(sat))
    AddGrain(Float(noise), 0.22, 0.22, constant=1>0)
    out this:

    Image
    [Attachment 84977 - Click to enlarge]




    it seems beauty, or not?

    there a little "noise" (a little "dithering"?)which I don't know if it can be removed or improved, but perhaps it can also be accepted as it is
    Quote Quote  
  10. Originally Posted by hello_hello View Post
    For FrostyBorders you need to specify the output dimensions rather than the aspect ratio

    even though FrostyBorders is not designed to get the only blurred background, would it be possible to know if it can still be used to do this?
    Quote Quote  
  11. hi

    my moving cat picture attached
    Image Attached Files
    Quote Quote  
  12. Oh Ken Burns effects? I start to understand more (again late) your intentions. In avisynth? There is this one avisynth Ken_Burns_Effect. Not sure how well it works.

    Not using NLE, you need an algorithm, that crops automatically, in certain speed, to a given starting coordinate. Interestingly enough, I made that algorithm myself, but in python, not avisynth, for zooming in on a current mouse cursor, using a mouse scroll button for opencv previewer. The tricky part is not to allow zooming to wiggle too much. For this precise effect, not sure. I used it in comparator app, before posted somewhere, times ago.
    Quote Quote  
  13. There are some flickering artifacts on the left edge of "catMovingPicture.mp4". Almost as if 2 layers of the moving cat were used, but with 1 offset incorrectly. Or maybe your editor has some layer/alpha channel interpretation problems with odd width ?

    When using lower quality / small dimension assets - you should upscale separately outside of the editor - because you can usually get better quality with more advanced external upscaling algorithms, producing fewer aliasing artifacts.
    Quote Quote  
  14. Originally Posted by poisondeathray View Post
    When using lower quality / small dimension assets - you should upscale separately outside of the editor - because you can usually get better quality with more advanced external upscaling algorithms, producing fewer aliasing artifacts.
    oh my

    Code:
    ffmpegOCT2024.exe -sn -an -hide_banner -v 32 -stats -y -i "%%a" -vf zscale=w=-2:h=6000:f=spline36:r=full:d=ordered -pix_fmt bgra -frames:v 1 -f image2 -update 1 -c:v targa -rle 0 "%OUTDIR1%\%%~na.tga"
    think the pandy's advise is true



    https://www.swisstransfer.com/d/c760292d-093f-4499-96e2-15ad96e82531
    Last edited by marcorocchini; 23rd Jan 2025 at 10:01.
    Quote Quote  
  15. so my final question would be: is there a script for avisynth, in addition to those mentioned, to generate a nice blurry background?
    Quote Quote  
  16. Originally Posted by _Al_ View Post
    Oh Ken Burns effects? I start to understand more (again late) your intentions. In avisynth? There is this one avisynth Ken_Burns_Effect. Not sure how well it works.

    Not using NLE, you need an algorithm, that crops automatically, in certain speed, to a given starting coordinate. Interestingly enough, I made that algorithm myself, but in python, not avisynth, for zooming in on a current mouse cursor, using a mouse scroll button for opencv previewer. The tricky part is not to allow zooming to wiggle too much. For this precise effect, not sure. I used it in comparator app, before posted somewhere, times ago.
    is there a way to use the KenBurnEffect script to get the only baclground to the output script?
    Quote Quote  
  17. KenBurnEffect is a script. You can edit it to make it do whatever you want.
    Quote Quote  
  18. Originally Posted by marcorocchini View Post
    is there a way to use the KenBurnEffect script to get the only background to the output script?
    Ken Burns was a guy who got famous making American history documentaries. There is no video or film footage's available of course for those times. So his videos consists mostly on presenting a photographs with a narrative. But those were not static or just simply zoomed in. He was animating or making camera move, not sure what, but he could start zoomed in on a detail and sloooowly zooming out, or maybe vice versa, or just panning over those photograph and perhaps zooming at the same time. Those moves were very slow and not intrusive, almost like a video footage revealing a scenes.

    So that filter is most likely using a zoom in or zoom out, not sure about pans, but perhaps yes, because for an algorithm it is all the same.

    It has nothing to do with blur background. That my response was reacting on your video when zooming in on a cat.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!