Hi all. I'm trying to encode a movie in mpeg4, using lavc. I'ld like to convert it to YV12. What ever I do, it never seems to work. Typically I tried: mencoder -oac copy -ovc lavc -lavcopts vbitrate=1000 input.avi -o output.avi -vf format=yv12,scale mencoder -oac copy -ovc lavc -lavcopts vbitrate=1000 input.avi -o output.avi -vf format=yv12,scale=640:-11 mencoder -oac copy -ovc lavc -lavcopts format=yv12:vbitrate=1000 input.avi -o output.avi -vf format=yv12,scale Whatever I do, when I play output.avi with mplayer, I got: VIDEO: [FMP4] 640x320 24bpp 29.970 fps 737.7 kbps (90.0 kbyte/s) The documentation says that lavc can encode in yv12. What am I doing wrong ? Thanks for help.
On Tuesday, 20 September 2005 at 19:51, Julien Lancien wrote:
Hi all.
I'm trying to encode a movie in mpeg4, using lavc. I'ld like to convert it to YV12. What ever I do, it never seems to work.
Typically I tried:
mencoder -oac copy -ovc lavc -lavcopts vbitrate=1000 input.avi -o output.avi -vf format=yv12,scale
mencoder -oac copy -ovc lavc -lavcopts vbitrate=1000 input.avi -o output.avi -vf format=yv12,scale=640:-11
mencoder -oac copy -ovc lavc -lavcopts format=yv12:vbitrate=1000 input.avi -o output.avi -vf format=yv12,scale
Whatever I do, when I play output.avi with mplayer, I got:
VIDEO: [FMP4] 640x320 24bpp 29.970 fps 737.7 kbps (90.0 kbyte/s)
The documentation says that lavc can encode in yv12.
What am I doing wrong ?
Thinking that you need to convert the colorspace. That output might be a bit misleading, but the video really is in YV12. R. -- MPlayer RPMs maintainer: http://rpm.greysector.net/mplayer/ "I am Grey. I stand between the candle and the star. We are Grey. We stand between the darkness ... and the light." -- Delenn in Grey Council in Babylon 5:"Babylon Squared"
On Tue, 20 Sep 2005 13:51:20 -0400 Julien Lancien <calexicoz@gmail.com> wrote:
Whatever I do, when I play output.avi with mplayer, I got:
VIDEO: [FMP4] 640x320 24bpp 29.970 fps 737.7 kbps (90.0 kbyte/s)
The documentation says that lavc can encode in yv12.
What in that line do you think is listing the colorspace?
Am Mittwoch, 21. September 2005 06:21 schrieb RC:
On Tue, 20 Sep 2005 13:51:20 -0400
Julien Lancien <calexicoz@gmail.com> wrote:
Whatever I do, when I play output.avi with mplayer, I got:
VIDEO: [FMP4] 640x320 24bpp 29.970 fps 737.7 kbps (90.0 kbyte/s)
The documentation says that lavc can encode in yv12.
What in that line do you think is listing the colorspace?
24bpp If you use -vo xv mplayer probably never converts the video to 24bpp, nevertheless it prints out "24bpp" which is very misleading. Matthias
In fact, I was asking the question because I was wondering if there was any quality impact of the colorspace when compressing in MPEG4. I'm almost sure that this is not the case, but I was thinking something like : 1000 kbits/s in YV12 is like 2000 kbits/s in RGB24. So is there any influence of the colorspace ? I also got another question, which I guess is much debated: which do you prefer of xvid and libavcodec ? I must precise that I don't care about time, only about quality at bitrates between 1000 and 2000 for DVD resolution. Thanks for all the help.
24bpp
If you use -vo xv mplayer probably never converts the video to 24bpp, nevertheless it prints out "24bpp" which is very misleading.
Matthias
On Wednesday 21 September 2005 16:37, Julien Lancien wrote:
In fact, I was asking the question because I was wondering if there was any quality impact of the colorspace when compressing in MPEG4. I'm almost sure that this is not the case, but I was thinking something like : 1000 kbits/s in YV12 is like 2000 kbits/s in RGB24.
So is there any influence of the colorspace ?
It always gets converted to YV12 for MPEG4 anyway.
Hi, On 9/21/05, Julien Lancien <calexicoz@gmail.com> wrote:
In fact, I was asking the question because I was wondering if there was any quality impact of the colorspace when compressing in MPEG4. I'm almost sure that this is not the case, but I was thinking something like : 1000 kbits/s in YV12 is like 2000 kbits/s in RGB24.
So is there any influence of the colorspace ?
I also got another question, which I guess is much debated: which do you prefer of xvid and libavcodec ?
lavc is very fast and of good quality if you tune it properly, xvid is fast and good quality with almost no tuning, and x264 is slow (compared to xvid and lavc) and yields excellent quality with no tuning needed. Guillaume -- Reading doesn't hurt, really! -- Dominik 'Rathann' Mierzejewski
On Wed, 21 Sep 2005 09:37:40 -0400 Julien Lancien <calexicoz@gmail.com> wrote:
I also got another question, which I guess is much debated: which do you prefer of xvid and libavcodec ?
I must precise that I don't care about time, only about quality at bitrates between 1000 and 2000 for DVD resolution.
Every time I find a video that shows artifacts with lavc, I try encoding with Xvid, and the Xvid results always look worse. I have no idea why some people are so dedicated to Xvid.
Every time I find a video that shows artifacts with lavc, I try encoding with Xvid, and the Xvid results always look worse.
I have no idea why some people are so dedicated to Xvid.
It's fun: I got exactly the opposite result. I always had better results, on many different kind of materials, with xvid (I use quite low bitrates though). Anyway, I'm gonna try x264 to see what it's worth.
Am Mittwoch, 21. September 2005 20:02 schrieb RC:
Every time I find a video that shows artifacts with lavc, I try encoding with Xvid, and the Xvid results always look worse.
I have no idea why some people are so dedicated to Xvid.
Xvid often looks better if you forget to use a postprocessing filter at playback time. Matthias
Hi On Wed, Sep 21, 2005 at 09:35:41PM +0200, Matthias Wieser wrote:
Am Mittwoch, 21. September 2005 20:02 schrieb RC:
Every time I find a video that shows artifacts with lavc, I try encoding with Xvid, and the Xvid results always look worse.
I have no idea why some people are so dedicated to Xvid.
Xvid often looks better if you forget to use a postprocessing filter at playback time.
do you maybe have some short sample which shows this effect? just curious, i suspect its easy to make lavc behave that way too optionally, maybe simply tuning some ME parameters is enough ... [...] -- Michael
Am Mittwoch, 21. September 2005 22:18 schrieb Michael Niedermayer:
Xvid often looks better if you forget to use a postprocessing filter at playback time.
do you maybe have some short sample which shows this effect? just curious, i suspect its easy to make lavc behave that way too optionally, maybe simply tuning some ME parameters is enough ...
Unfortunately not, because I most often use lavc + pp=ac. But some time ago I have seen a few samples where low bitrate encodings with xvid were slightly unsharp, while libavcodec showed more block artefacts. I don't know if this behaviour still can be observed with recent versions. Regards, Matthias
On Wed, 21 Sep 2005 21:35:41 +0200 Matthias Wieser <mwieser@gmx.de> wrote:
Xvid often looks better if you forget to use a postprocessing filter at playback time.
I don't use postprocessing at all, and I still never see cases where Xvid looks better.
Hi RC! On 2005.09.22 at 12:52:12 -0700, RC wrote next:
Xvid often looks better if you forget to use a postprocessing filter at playback time.
I don't use postprocessing at all, and I still never see cases where Xvid looks better.
That's interesting, because since I switched to xvid I almost haven't seen cases where lavc is better than xvid. I'm talking about high bitrate (1.5-2 mbit/s) encoding that looks just like dvd where postprocessing only destroys details. Maybe you used wrong options in xvid encoding, like forgot to use b-frames (unlike lavc, they don't suck and decrease bitrate without impact on quality), mpeg quantization or 2-pass encoding? Try again, if you watch without postprocessing, there is a great chance that you'll like it. Also, in my opinion xvid is always better for real life video, but it performs worse on animation. Lavc can catch up with xvid here sometimes. -- Vladimir
Am Donnerstag, 22. September 2005 22:58 schrieb Vladimir Mosgalin:
Hi RC!
On 2005.09.22 at 12:52:12 -0700, RC wrote next:
Xvid often looks better if you forget to use a postprocessing filter at playback time.
I don't use postprocessing at all, and I still never see cases where Xvid looks better.
That's interesting, because since I switched to xvid I almost haven't seen cases where lavc is better than xvid. I'm talking about high bitrate (1.5-2 mbit/s) encoding that looks just like dvd where postprocessing only destroys details.
If the postprocessing is done right it does not destroy any real details, only compression artefacts. Matthias
Hi Matthias Wieser! On 2005.09.23 at 10:45:14 +0200, Matthias Wieser wrote next:
On 2005.09.22 at 12:52:12 -0700, RC wrote next:
Xvid often looks better if you forget to use a postprocessing filter at playback time.
I don't use postprocessing at all, and I still never see cases where Xvid looks better.
That's interesting, because since I switched to xvid I almost haven't seen cases where lavc is better than xvid. I'm talking about high bitrate (1.5-2 mbit/s) encoding that looks just like dvd where postprocessing only destroys details.
If the postprocessing is done right it does not destroy any real details, only compression artefacts.
What can I try then, besides pp=hb/vb/dr, pp=ac and [f]spp? They all destroy details even on video with qp=2. -- Vladimir
Am Freitag, 23. September 2005 18:04 schrieb Vladimir Mosgalin:
Hi Matthias Wieser!
On 2005.09.23 at 10:45:14 +0200, Matthias Wieser wrote next:
On 2005.09.22 at 12:52:12 -0700, RC wrote next:
Xvid often looks better if you forget to use a postprocessing filter at playback time.
I don't use postprocessing at all, and I still never see cases where Xvid looks better.
That's interesting, because since I switched to xvid I almost haven't seen cases where lavc is better than xvid. I'm talking about high bitrate (1.5-2 mbit/s) encoding that looks just like dvd where postprocessing only destroys details.
If the postprocessing is done right it does not destroy any real details, only compression artefacts.
What can I try then, besides pp=hb/vb/dr, pp=ac and [f]spp? They all destroy details even on video with qp=2.
You could try to adjust the thresholds of the deblocking filters. Instead of pp=ac (== ha:a:128:7,va:a,dr:a) maybe ha:a:64:9,va:a,dr:a
On Fri, 23 Sep 2005 00:58:16 +0400 Vladimir Mosgalin <mosgalin@VM10124.spb.edu> wrote:
I'm talking about high bitrate (1.5-2 mbit/s) encoding that looks just like dvd where postprocessing only destroys details.
I use bitrates of around 1000.
Maybe you used wrong options in xvid encoding, like forgot to use b-frames (unlike lavc, they don't suck and decrease bitrate without impact on quality), mpeg quantization or 2-pass encoding?
B-frames are even enabled by default, and I use them with lavc as well much of the time (fade in/out looks terrible otherwise). Same for mpeg_quant, and 2-pass.
Try again, if you watch without postprocessing, there is a great chance that you'll like it.
Believe me, I've tried. I'd really love to find a video codec that compresses/looks better than lavc, even if encoding does take 10x longer. Unfortunately, every source I've tested (mostly NTSC DVDs) shows the opposite.
Also, in my opinion xvid is always better for real life video, but it performs worse on animation. Lavc can catch up with xvid here sometimes.
lavc is absolutely incredible on animation. *cmp=2 qns=3 2bframes, etc., and you can get great quality animation at very, very low bitrates.
Hi RC! On 2005.09.25 at 13:54:41 -0700, RC wrote next:
I'm talking about high bitrate (1.5-2 mbit/s) encoding that looks just like dvd where postprocessing only destroys details.
I use bitrates of around 1000.
Maybe that's where the difference lies. IMHO 1 mbit/s isn't enough for picture to be good enough, or it will require downscaling, which hurts quality too. I downscale only to nearest divisible by 16 size - downscaling by 2-14 pixels isn't noticeable at all. If you don't downscale and don't use postprocessing, I don't really understand why you don't see artifacts.
Maybe you used wrong options in xvid encoding, like forgot to use b-frames (unlike lavc, they don't suck and decrease bitrate without impact on quality), mpeg quantization or 2-pass encoding?
B-frames are even enabled by default, and I use them with lavc as well much of the time (fade in/out looks terrible otherwise). Same for mpeg_quant, and 2-pass.
fade in/out looks terrible? Heh, I get opposite results, at least sometimes. If you've got enough bitrate, p-frames only sequence can produce nice looking fade in, but b-frames almost always make it worse. But really, it depends.. When I use lavc, I don't use b-frames, because without them lavc produces good results. On xvid, you have to use them, they always help a lot (bitrate-wise). However, on some material they just don't cut (no matter what kind of encoder you use). This is the case where lavc (without b-frames) can perform better than xvid (without b-frames).
Try again, if you watch without postprocessing, there is a great chance that you'll like it.
Believe me, I've tried. I'd really love to find a video codec that compresses/looks better than lavc, even if encoding does take 10x longer. Unfortunately, every source I've tested (mostly NTSC DVDs) shows the opposite.
Out of curiosity, have you tried some of this new stuff, like snow, x264 or something else h264-based? I haven't really, but I've seen a few really good wmv3 samples. They all promise a lot, however their promises mostly lie in land "more quality at very low/low/middle bitrate". While I need "the least possible bitrate for encode that my eye can't distinguish from dvd". So far, xvid + no pp is the champion (though for low and middle bitrate encodes I prefer lavc + pp)
Also, in my opinion xvid is always better for real life video, but it performs worse on animation. Lavc can catch up with xvid here sometimes.
lavc is absolutely incredible on animation. *cmp=2 qns=3 2bframes, etc., and you can get great quality animation at very, very low bitrates.
I haven't tried qns=3, and turning on b-frames never (at least I don't remember) helped me on animation when encoding with lavc, but I wouldn't say that it is that much better than xvid.. It is better in some cases, but not in the most of them. -- Vladimir
On Mon, 26 Sep 2005 01:42:45 +0400 Vladimir Mosgalin <mosgalin@VM10124.spb.edu> wrote:
If you don't downscale and don't use postprocessing, I don't really understand why you don't see artifacts.
I don't use Xvid ;-) To be fair, I do see artifacts on occasion, but as I've said, Xvid is worse at the same bitrates. It's usually a case of the source DVD being blocky (very low bitrates) anyhow, with lots and lots of noise that helps cover the blocks.
sometimes. If you've got enough bitrate, p-frames only sequence can produce nice looking fade in, but b-frames almost always make it worse. But really, it depends..
Well if you have a very high bitrate, Lavc will just make every frame of a fade a keyframe, so it will look perfect. I've never seen a good-looking fade in/out with p-frames.
Out of curiosity, have you tried some of this new stuff, like snow, x264 or something else h264-based? I haven't really, but I've seen a few really good wmv3 samples.
Haven't gotten around to trying x264 yet. Snow looks good, and I'm sure it will be quite impressive as soon as it stabilizes. I've looked at WMV3 samples, and I'm not impressed. Pretty much on-par with MPEG-4. Most samples use strong denoising to squeeze the bitrate down as much as possible and still look decent, but you can get better results with lavc and nr=~500.
While I need "the least possible bitrate for encode that my eye can't distinguish from dvd".
I'm beginning to suspect you're someone who needs noise in the video. Leave out all other postprocessing, and just use -vf noise on a lavc MPEG-4 video and see if that's the reason. If so you can try lavc encoding with nsse. (I find sse in general to not be very good, but that's the only lavc *cmp function that preserves noise)
I haven't tried qns=3, and turning on b-frames never (at least I don't remember) helped me on animation when encoding with lavc, but I wouldn't say that it is that much better than xvid.. It is better in some cases, but not in the most of them.
I'd love to see a sample of good looking animation with Xvid at 720x480/24fps with bitrates of ~200.
Hi RC! On 2005.09.25 at 17:24:50 -0700, RC wrote next:
sometimes. If you've got enough bitrate, p-frames only sequence can produce nice looking fade in, but b-frames almost always make it worse. But really, it depends..
Well if you have a very high bitrate, Lavc will just make every frame of a fade a keyframe, so it will look perfect. I've never seen a good-looking fade in/out with p-frames.
Hmm.. maybe, the point is that b-frames destroy perfectness. Disabling them help. I don't remember this for sure, but you are probably right.
While I need "the least possible bitrate for encode that my eye can't distinguish from dvd".
I'm beginning to suspect you're someone who needs noise in the video. Leave out all other postprocessing, and just use -vf noise on a lavc MPEG-4 video and see if that's the reason. If so you can try lavc encoding with nsse. (I find sse in general to not be very good, but that's the only lavc *cmp function that preserves noise)
I need noise? Yes, that's right, I always use -vf noise=10uah:3uah on playback. It hides a lot of artifacts. However, I denoise original content before encoding, it always helps.
I haven't tried qns=3, and turning on b-frames never (at least I don't remember) helped me on animation when encoding with lavc, but I wouldn't say that it is that much better than xvid.. It is better in some cases, but not in the most of them.
I'd love to see a sample of good looking animation with Xvid at 720x480/24fps with bitrates of ~200.
I already said that I'm not interested in such bitrates.. No codec will produce good-looking animation at any resolution with bitrate of 500 kbit/s. Maybe it won't swarm with artifacts, but then low resolution and aliasing will destroy the quality. And no small details at all.. Depending on the kind of animation, sometimes 1300-1500 is enough for 704x480. Sometimes you have to go as high as 2200-2500 for 704x360, or you'll see a lot of artifacts in fast scenes. -- Vladimir
On Tue, 27 Sep 2005 00:06:43 +0400 Vladimir Mosgalin <mosgalin@VM10124.spb.edu> wrote:
I already said that I'm not interested in such bitrates.. No codec will produce good-looking animation at any resolution with bitrate of 500 kbit/s. Maybe it won't swarm with artifacts, but then low resolution and aliasing will destroy the quality. And no small details at all..
No, at 500k lavc will look practically perfect on animation. I don't know why you're mentioning aliasing, that has nothing to do with bitrate/artifacts.
Depending on the kind of animation, sometimes 1300-1500 is enough for 704x480. Sometimes you have to go as high as 2200-2500 for 704x360, or you'll see a lot of artifacts in fast scenes.
I don't think you could even force lavc to use a bitrate of 2200 for 720x480 animation. It would be absolutely perfect long before you'd get that high.
On Mon, Sep 26, 2005 at 04:39:59PM -0700, RC wrote:
On Tue, 27 Sep 2005 00:06:43 +0400 Vladimir Mosgalin <mosgalin@VM10124.spb.edu> wrote:
I already said that I'm not interested in such bitrates.. No codec will produce good-looking animation at any resolution with bitrate of 500 kbit/s. Maybe it won't swarm with artifacts, but then low resolution and aliasing will destroy the quality. And no small details at all..
No, at 500k lavc will look practically perfect on animation. I don't
What kind of shitty animation are you talking about? Anyone encoding anime knows that 1000-1100 kbit/sec is the minimum acceptable rate, and 1200-1500 are not uncommon.
know why you're mentioning aliasing, that has nothing to do with bitrate/artifacts.
Apparently talking about downscaling with a bad resampler...
Depending on the kind of animation, sometimes 1300-1500 is enough for 704x480. Sometimes you have to go as high as 2200-2500 for 704x360, or you'll see a lot of artifacts in fast scenes.
I don't think you could even force lavc to use a bitrate of 2200 for 720x480 animation. It would be absolutely perfect long before you'd get that high.
No way. Constant q=2 should give around 1700-2500 depending on the content, iirc. Rich
On Mon, 26 Sep 2005 22:37:06 -0400 Rich Felker <dalias@aerifal.cx> wrote:
What kind of shitty animation are you talking about? Anyone encoding anime knows that 1000-1100 kbit/sec is the minimum acceptable rate, and 1200-1500 are not uncommon.
Well certainly not Anime. Conventional animation ala. Futurama, Family Guy, etc.
Hi Rich Felker! On 2005.09.26 at 22:37:06 -0400, Rich Felker wrote next:
know why you're mentioning aliasing, that has nothing to do with bitrate/artifacts.
Apparently talking about downscaling with a bad resampler...
Why bad? 2x downscaling will always either give aliasing or produce way too smooth picture, which may be better for encoding, but worse for eye... -- Vladimir
On Tue, Sep 27, 2005 at 09:23:20AM +0400, Vladimir Mosgalin wrote:
Hi Rich Felker!
On 2005.09.26 at 22:37:06 -0400, Rich Felker wrote next:
know why you're mentioning aliasing, that has nothing to do with bitrate/artifacts.
Apparently talking about downscaling with a bad resampler...
Why bad? 2x downscaling will always either give aliasing or produce way
I get the impression you don't know what aliasing means..
too smooth picture, which may be better for encoding, but worse for eye...
Worse than aliasing? Nothing is worse than aliasing.. Rich
Hi Rich Felker! On 2005.09.27 at 12:15:04 -0400, Rich Felker wrote next:
know why you're mentioning aliasing, that has nothing to do with bitrate/artifacts.
Apparently talking about downscaling with a bad resampler...
Why bad? 2x downscaling will always either give aliasing or produce way
I get the impression you don't know what aliasing means..
Er, I mean - when you clearly see that line isn't smooth, but looks like ** ** ** and other similar effects. Either this or blurring always happen when you downscale to small resolution and then upscale. Downscaling two times is enough for sharp-looking video for this effects to show.
too smooth picture, which may be better for encoding, but worse for eye...
Worse than aliasing? Nothing is worse than aliasing..
I thought we already concluded that we measure quality of video in a different way. I still don't understand what you dislike about tearing effect in some video out drivers, for example ;) I prefer sharp picture with artifacts to smooth picture. Some of my friends prefer smoothed one. So what? -- Vladimir
On Tue, Sep 27, 2005 at 09:40:35PM +0400, Vladimir Mosgalin wrote:
Hi Rich Felker!
On 2005.09.27 at 12:15:04 -0400, Rich Felker wrote next:
know why you're mentioning aliasing, that has nothing to do with bitrate/artifacts.
Apparently talking about downscaling with a bad resampler...
Why bad? 2x downscaling will always either give aliasing or produce way
I get the impression you don't know what aliasing means..
Er, I mean - when you clearly see that line isn't smooth, but looks like
** ** **
and other similar effects. Either this or blurring always happen when you downscale to small resolution and then upscale. Downscaling two times is enough for sharp-looking video for this effects to show.
Look up aliasing, then. It has a very precise meaning.
too smooth picture, which may be better for encoding, but worse for eye...
Worse than aliasing? Nothing is worse than aliasing..
I thought we already concluded that we measure quality of video in a different way. I still don't understand what you dislike about tearing effect in some video out drivers, for example ;)
I prefer sharp picture with artifacts to smooth picture. Some of my friends prefer smoothed one. So what?
You mean to tell me a video reduced to half-height with -vf field=0 looks better than with a proper scaler? Rich
Hi Rich Felker! On 2005.09.27 at 23:21:52 -0400, Rich Felker wrote next:
Look up aliasing, then. It has a very precise meaning.
OK, I found this in mplayer's docs aliasing (only downscale) straight lines look like staircases areas of high detail look worse regular patterns turn into a funny looking mess (moire patterns) bad: fast bilinear, point, gauss with low variance acceptable: area ok: others staircases and moir patterns are exactly what I was talking about.
I prefer sharp picture with artifacts to smooth picture. Some of my friends prefer smoothed one. So what?
You mean to tell me a video reduced to half-height with -vf field=0 looks better than with a proper scaler?
No, it looks a bit worse. And aliasing effect is stronger with it (but when downscaling with bicubic scaler it appears _too_). But I don't really understand what do you want to say, that phrase of mine wasn't related to the rest of mail. Just an objection to your "nothing can be worse than <something you don't like much>". -- Vladimir
On Wed, Sep 28, 2005 at 09:15:16AM +0400, Vladimir Mosgalin wrote:
Hi Rich Felker!
On 2005.09.27 at 23:21:52 -0400, Rich Felker wrote next:
Look up aliasing, then. It has a very precise meaning.
OK, I found this in mplayer's docs
aliasing (only downscale) straight lines look like staircases areas of high detail look worse regular patterns turn into a funny looking mess (moire patterns) bad: fast bilinear, point, gauss with low variance acceptable: area ok: others
staircases and moir patterns are exactly what I was talking about.
These are common symptoms to identify it by, not definitions.
I prefer sharp picture with artifacts to smooth picture. Some of my friends prefer smoothed one. So what?
You mean to tell me a video reduced to half-height with -vf field=0 looks better than with a proper scaler?
No, it looks a bit worse. And aliasing effect is stronger with it (but when downscaling with bicubic scaler it appears _too_). But I don't really understand what do you want to say, that phrase of mine wasn't related to the rest of mail. Just an objection to your "nothing can be worse than <something you don't like much>".
I think if you understood sampling theory my statement would make sense and not be so unbelievable.. Rich
Hi Rich Felker! On 2005.09.28 at 11:25:41 -0400, Rich Felker wrote next:
staircases and moir patterns are exactly what I was talking about.
These are common symptoms to identify it by, not definitions.
Where can I find exact definition? You know, aliasing is a generic word, and when talking about picture on computer screen, it usually means just what I said. If you know more specific definition, related strictly to video, please tell it.
No, it looks a bit worse. And aliasing effect is stronger with it (but when downscaling with bicubic scaler it appears _too_). But I don't really understand what do you want to say, that phrase of mine wasn't related to the rest of mail. Just an objection to your "nothing can be worse than <something you don't like much>".
I think if you understood sampling theory my statement would make sense and not be so unbelievable..
What should I agree to, that downscaling by throwing away half of the lines is bad and makes this effect stronger? Yes, that's right. But I know a man who doesn't hesitate to do this and they don't really worry about this. I also know some people who capture video from tv tuner at full resolution and then "deinterlace", as they say, by downscaling height by two. They are completely satisfied with results. (As for me, I don't downscale, so I don't really care - and if I do, any effect is about equally bad) So if YOU think that nothing can be worse and everyone thinks so, let me assure you, you are completely wrong. -- Vladimir
On Wed, Sep 28, 2005 at 11:00:11PM +0400, Vladimir Mosgalin wrote:
Hi Rich Felker!
On 2005.09.28 at 11:25:41 -0400, Rich Felker wrote next:
staircases and moir patterns are exactly what I was talking about.
These are common symptoms to identify it by, not definitions.
Where can I find exact definition? You know, aliasing is a generic word, and when talking about picture on computer screen, it usually means just what I said. If you know more specific definition, related strictly to video, please tell it.
Look up sampling theory, shannon's theorem, nyquist frequency, or other related terms. Wikipedia probably has good info. Essentially, aliasing is what happens when you try to reproduce a signal based on too few samples, and the higher frequencies (above the maximal frequency representable by your sampling, i.e. half the sampling rate aka the nyquist frequency) wrap down into the lower frequency bands.
No, it looks a bit worse. And aliasing effect is stronger with it (but when downscaling with bicubic scaler it appears _too_). But I don't really understand what do you want to say, that phrase of mine wasn't related to the rest of mail. Just an objection to your "nothing can be worse than <something you don't like much>".
I think if you understood sampling theory my statement would make sense and not be so unbelievable..
What should I agree to, that downscaling by throwing away half of the lines is bad and makes this effect stronger? Yes, that's right. But I know a man who doesn't hesitate to do this and they don't really worry about this. I also know some people who capture video from tv tuner at full resolution and then "deinterlace", as they say, by downscaling height by two. They are completely satisfied with results.
(As for me, I don't downscale, so I don't really care - and if I do, any effect is about equally bad)
So if YOU think that nothing can be worse and everyone thinks so, let me assure you, you are completely wrong.
Well I won't continue to argue with you, but I think these people are both being rather ridiculous and stubborn in their preferences... Rich
participants (10)
-
Dominik 'Rathann' Mierzejewski -
Guillaume POIRIER -
Jan Knutar -
Julien Lancien -
Matthias Wieser -
Michael Niedermayer -
Phil Ehrens -
RC -
Rich Felker -
Vladimir Mosgalin