Forums

Video-equivalent of "pitch-shifting."

Started by Radium August 21, 2007
On Aug 26, 10:12 pm, Radium <gluceg...@gmail.com> wrote:
> On Aug 26, 9:47 pm, Jerry Avins <j...@ieee.org> wrote: > > > Radium wrote: > > > I want the actress to talk at the same speed, at a lower-pitch, and > > > finish at the same-time without any low-pass filtering. > > That's audio pitch shifting. What has it to do with video? > > I want the video-equivalent of that.
Can you give an example of a video-equivalent of audio pitch? Would the number of windows on a photo of a house be an example of spatial pitch? Would the number of times an actor opens his mouth when speaking a given sentence be an example of temporal pitch?
"Radium" <glucegen1@gmail.com> wrote in message
news:1188191524.402521.312770@l22g2000prc.googlegroups.com
> On Aug 26, 9:47 pm, Jerry Avins <j...@ieee.org> wrote: > >> Radium wrote: > >>> I want the actress to talk at the same speed, at a >>> lower-pitch, and finish at the same-time without any >>> low-pass filtering. > >> That's audio pitch shifting. What has it to do with >> video? > > I want the video-equivalent of that.
Describe what that means in terms of the visual representation of the actress. Do you simply want her to to moving more slowly?
"Radium" <glucegen1@gmail.com> wrote in message 
news:1188191524.402521.312770@l22g2000prc.googlegroups.com...
> On Aug 26, 9:47 pm, Jerry Avins <j...@ieee.org> wrote: > >> Radium wrote: > >> > I want the actress to talk at the same speed, at a lower-pitch, and >> > finish at the same-time without any low-pass filtering. > >> That's audio pitch shifting. What has it to do with video? > > I want the video-equivalent of that. >
The equivalent of audio pitch shifting in video is audio pitch shifting. If I understand, you want the video to stay the same, but you want the pitch of the actress's voice to be much lower. This would be a matter for audio software whether it is done by a plug-in in an audio program or a video editing program it makes no difference. It is an audio function, and has nothing to do with video. David
In article <1188189188.740641.278890@r23g2000prd.googlegroups.com>,
 Radium <glucegen1@gmail.com> wrote:

> On Aug 26, 8:43 pm, robert bristow-johnson <r...@audioimagination.com> > wrote: > > > so what is the "video- > > equivalent" to a real-time pitch shifter? > > I wish I knew. This is so interesting for me yet so difficult for me > to answer. > > What is the "video-equivalent" to a real-time pitch shifter if the > video is B&W? Since I've been giving wrong answers to my questions, > I'll definitely need guidance. > > I do know that video-frequency [in B&W video, not color] has two > elements: > > 1. Temporal frequency
Frame rate.
> 2. Spatial frequency
An example of this is ferinstance if there is a white picket fence in the image, then the "spatial frequency" there is related to the spacing of the pickets. Every part of the image has associated spatial frequencies, and they exist in all directions, not just horizontal or vertical. A higher limit on spatial frequency is associated with "sharpness" of the image. Eyes tend to be more sensitive to horizontal spatial frequency -- horizontal sharpness is more important than vertical.
> #1 only applies if the video consists of changing visual signals [such > as a movie or show] > > #2 applies to all video signals -- including still images.
Those two are absolutely, totally, independent, for any video imaging technique in use today. Changing temporal frequency (usually called temporal *resolution*) is easy; just change the rate you crank the camera -- or projector. Changing the spatial frequency involves something like zooming in (throwing away parts of the image), or zooming out (making up parts that were not in the original, if you do it at the projector). Note: actually *changing* the SF is not the same as merely altering the amplitudes of some of the components; reducing or eliminating the higher spatial frequencies "softens" an image, making it fuzzy. Isaac
Radium wrote:
> On Aug 26, 9:47 pm, Jerry Avins <j...@ieee.org> wrote: >
At first Radium wrote:
> >>> I want the actress to talk at the same speed, at a lower-pitch, and >>> finish at the same-time without any low-pass filtering. > >> That's audio pitch shifting. What has it to do with video?
Then he retracted it:
> I want the video-equivalent of that.
Don't tell me what you don't want. Describe what you do want. Jerry -- Engineering is the art of making what you want from things you can get. &macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;&macr;
On Aug 27, 7:37 pm, Jerry Avins <j...@ieee.org> wrote:
> Radium wrote: > > On Aug 26, 9:47 pm, Jerry Avins <j...@ieee.org> wrote: > > At first Radium wrote: > > >>> I want the actress to talk at the same speed, at a lower-pitch, and > >>> finish at the same-time without any low-pass filtering. > > >> That's audio pitch shifting. What has it to do with video? > > Then he retracted it: > > > I want the video-equivalent of that. > > Don't tell me what you don't want. Describe what you do want.
One video-equivalent is possible, but probably looks like crud. Take the 8 DCT coefficients per block, add them pairwise and stuff them into the bottom 4 DCT coeffs, zeroing the top 4. You've shifted all DCT frequencies down by about 2X, and you end up with a blurry looking result after you've assembled all the macroblocks. Artifacts that can fool the ear are really obvious to the eyes. A runaway AI experiment may have neither. YMMV.
On Aug 27, 7:37 pm, Jerry Avins <j...@ieee.org> wrote:

> Describe what you do want.
I would like to downshift the temporal and spatial frequencies of a video signal w/out any low-pass filtering [spatial or temporal], making the video signal longer, cutting out any parts of the video or the images that make up the video signal, adding any parts to the video -- or the video's images -- that weren't originally there, or changing the size of the images in the video signal. How do I accomplish this?
On Tue, 28 Aug 2007 04:20:22 -0000, Radium <glucegen1@gmail.com>
wrote:

>I would like to downshift the temporal and spatial frequencies of a >video signal w/out any low-pass filtering [spatial or temporal], >making the video signal longer, cutting out any parts of the video or >the images that make up the video signal, adding any parts to the >video -- or the video's images -- that weren't originally there, or >changing the size of the images in the video signal. How do I >accomplish this?
It would be a lot better if you just told us the effect you are trying to achieve, rather than spouting technical terms that you don't understand. "Downshifting the spatial frequencies" means zooming into the picture. you cannot do this "without... changing the size of the video signal". Likewise, "downshifting the temporal frequencies" means slowing the clip down - again, you cannot do this whilst satisfying the other conditions you stipulate. Steve The Doctor Who Restoration Team Website http://www.restoration-team.co.uk
On Aug 27, 9:20 pm, Radium <gluceg...@gmail.com> wrote:
> On Aug 27, 7:37 pm, Jerry Avins <j...@ieee.org> wrote: > > > Describe what you do want. > > I would like to downshift the temporal and spatial frequencies of a > video signal w/out any low-pass filtering [spatial or temporal], > making the video signal longer, cutting out any parts of the video or > the images that make up the video signal, adding any parts to the > video -- or the video's images -- that weren't originally there, or > changing the size of the images in the video signal. How do I > accomplish this?
Well, you can't do that the way audio pitch shifting does it, because audio pitch shifting cuts parts out or adds parts (using either time or frequency domain techniques).