Frage von eLcoMANdaNtE:
hello their clever heads ...
I did not know in which category my question fit, so I present them here.
I am looking for a possibility the chrominance and luminance from wmv and mpeg files "read" them. I need these values for each frame and as an average over all frames of a clip?
how come I get? gibts da ne small software or how could such a software work?
is really urgent ...
mfg eL
Antwort von TheBubble:
That would be a decompression of the same film in which only a final Farbraumtranformation to RGB is performed.
If there is only on the information on each pixel in the YUV colorspace important is it for you may be easier to just extracted Picture in RGB color space and take it back to a YUV color space for further analysis to transform.
Antwort von eLcoMANdaNtE:
thanks for the reply.
I am still unclear how to proceed? Which decoder should I use as s.besten. I want the luminance and chrominance s.besten in matrix form, in order to zb with matlab to use next.
is a bit tricky, I think ... but I am really puzzled at the moment ...
Here again the summary of my problems:
I want the quality of wmv streams objectively assessed. I will use the fritzframalyzer to me PSNR and MSE Euclidean distance of a reference to the clip-clip streamed output. to make sure that the correct analysis program delivers results, I must by a measurement examples verify. I think this works on s.besten luminance and chrominance. if I'm wrong, please clarify me on. who has an idea to solve this problem, please sign!
is really urgent
mfg eL
Antwort von TheBubble:
I am still unclear how to proceed?
You take an uncompressed picture in RGB Frabraum and walking it in the YUV color space, if you get the picture in this space look like. Caution: There are several variants of YUV.
You should bear in mind that the data in the YUV color space, as they are usually from the codec are stored in the chrominance components possessing a lower resolution. The way components sampled and later reconstructed, it can be between different codecs also differ (eg, MPEG-1 to MPEG-2). Either you take into account all of this (if you have the YUV data directly from a tap decompressor want), or you're going to step on an RGB picture and take any rounding errors in converting multiple purchase.
Synonymous You could use an external program in order to be investigated is the video file is stored in lossless (RGB)-frames to decompose (eg, PNG or BMP file format).
I think this works on s.besten luminance and chrominance. if I'm wrong, please clarify me on.
Monitors work in RGB colorspace. Cameras are synonymous RGB components. So why not in this color berurteilen? The MSE between two images, you can synonymous in the RGB color space calculated.
The MSE is suited to the difference between two images to determine (and minimize, if necessary), but the results do not always correspond to the perceived differences. A viewer may have a different perception.
who has an idea to solve this problem, please sign!
What do you need it?
Antwort von eLcoMANdaNtE:
I investigate the quality of streaming offered. I would like the fritz framalyzer use. deliver me for the wmv files mse, PSNR and eukl. distance. therefore I can objectively review the design.
I have this high-quality reference material and the corresponding streamed files.
but the results in order to be able to verify, do I have a small measurement s.werte come, with whom I obtained the PSNR and MSE can confirm.
I just thought it s.besten on the chrominance and luminance and their changes compared with the reference materials would run? AviSynth seemed as skripsprache s.geeignetesten to s.die worthy to come ...
because you have a better proposal? the measurement must be independent from the analytical tool to work ...
mfg eL
Antwort von TheBubble:
AviSynth seemed as skripsprache s.geeignetesten to s.die worthy to come ...
Can you take. You need then have a filter that you like the values (MSE, etc.) is calculated. So something can be programmed.
Ultimately you have two independent measurement, the results of which you can compare (and hopefully the same).
Antwort von eLcoMANdaNtE:
you do not know how to do it randomly? I have not yet much of unknowing AviSynth.
but I wanted to clarify whether times previously since it does not already have a plugin or similar There, around the wheel twice to invent.
I thought a plugin reads the appropriate values and matlab, etc. then can calculate mse etc. ... or located since I wrong?
The only question is how such a plugin / tool created. as I said, yes it is used in principle only the results of hedging.
mfg eL
Antwort von TheBubble:
you do not know how to do it randomly? I have not yet much of unknowing AviSynth. You can create filters for Avisynth. Whether (and if so how) to filter for Avisynth can create, the two video data streams as input to accept, I can try to find out.
but I wanted to clarify whether times previously since it does not already have a plugin or similar There, around the wheel twice to invent.
Must stop to look. Presumably it will have similar type. Whether your desires are fulfilled, you need to find out.
how such a plugin / tool created.
Plugins (eg for Avisynth) are usually written in C / C + +.
If there is only the results for individual frames, then export it to comparative frames from the streams with appropriate software, do this with Matlab and calculate the values.
Antwort von eLcoMANdaNtE:
Whether (and if so how) to filter for Avisynth can create, the two video data streams as input to accept, I can try to find out.
that would be super! so that would be very helpful to me.
for suitable plugins I have in various AviSynth plugin collections compatriots, but not really found. This could of course synonymous s.der sometimes very confusing name is * g *
it makes sense to go through the extraction of individual images to go?
there would be but synonymous sure to ask how individual values s.die comes so matlab able to deal with ...
mfg eL
Antwort von eLcoMANdaNtE:
I have found a plugin grad. it means "show pixel values" and what could be.
could be good for s.die yuv and rgb color spaces come.
I try through the times.
for instance-syntax, however, speaks of *. avi which is then actually synonymous for wmv?
LoadPlugin ( "ShowPixelValues.dll")
AVISource ( "test.avi")
Show pixel values ()
# Show the pixel values s.the top left of the frame mfg eL
Antwort von TheBubble:
That gives you just the color of a pixel? (I know the filter is not.)
If you MSE, and so reasonably efficiently calculate, then you create yourself a search or filter, the two clips and accepted as input for each frame the values calculated and output.
Alternatively, convert both to individual clips and see these paired with a program.
Antwort von eLcoMANdaNtE:
you're right. but it comes s.die but ran YUV values. I thought that would be needed in order to calculate mse, etc.? I'm on the wrong boat?
If you MSE, and so reasonably efficiently calculate, then you create yourself a search or filter, the two clips and accepted as input for each frame the values calculated and output.
because the dog is buried. I do not have a glimmer as I create filter itself ... as I probably need your help.
mfg eL
Antwort von TheBubble:
you're right. but it comes s.die but ran YUV values. I thought that would be needed in order to calculate mse, etc.?
The mean square error, you can both pixel data in YUV color space, as synonymous on pixel data in RGB color space calculated.
What exactly you want to compare, you know. If you are working with YUV, you need two things in mind:
- A designated as the YUV format is not always the expected YUV format
- The quality of the displayed image depends RGB synonymous of the kind of upsampling the U and V components, if these low dissolved saved.
A few things I would be interested in:
What need to leave a Comparison between two video files, so in what context do you want to compare video clips, and what are your previous experience in the field?
Antwort von eLcoMANdaNtE:
I want the quality of streaming offered compare.
ie I have the source material and would now like the quality of the streams of several providers to compare objectively.
So I know how the material originally looks (available on dvd as mpeg2 before) and I have the material as a stream (usually wmv).
I would now streaming all outcomes against the original material for a proper comparison to be able to make the best quality of the stream is. I want to with the framalyzer mse & co as quality dimensions can be calculated.
but these results must be hedged. this is about "my" Messprogramm happen.
how accurately the analytical results can be verified, I am trying to figure out grad. I thought it works on the change of chrominance and luminance and their changes compared with the reference material. therefore YUV colorspace.
Previous experience is not very extensive. I try to incorporate grad.
Please let us communicate via email. I would prefer over the forum ... it is quite cumbersome.
my mail: dj-mos (at) gmx.de
Thank you
mfg eL
Antwort von TheBubble:
I understand that you're streaming video look like. However, it is not entirely clear whether it is real streams involved (which may dynamically depending on network load between levels of quality verschiedneen change) or prepared video files that are easy during the download already be (I suspect the latter).
To view the results of the tool to verify you will not much more left than to try the same values with the same calculation rules with another tool to identify.
