Mailing List Archive

I don't have a Tivo either, but I've seen comments on the list implying that
a good Myth setup has a better picture than Tivo. Depends on your hardware
& capture settings, of course.

-JAC

> -----Original Message-----
> From: mythtv-users-bounces@snowman.net
> [mailto:mythtv-users-bounces@snowman.net]On Behalf Of Andres Valenciano
> Sent: Friday, May 16, 2003 1:32 PM
> To: Discussion about mythtv
> Subject: [mythtv-users] Tivo output vs MythTV-PC output question
>
>
> Hello,
> I have been reading this list for a while, and I am wondering what has
> Tivo that makes it a better option in terms of video output quality
> compare to a DIY PC setup. I don't have access to a Tivo so I really no
> idea about its quality.

Andres Valenciano wrote:

> Hello,
> I have been reading this list for a while, and I am wondering what has
> Tivo that makes it a better option in terms of video output quality
> compare to a DIY PC setup.



This is debatable. I suspect that since tivos are all the same and most
everyone is pretty satisfied with the tv-out quality you don't hear many
complaints. With tv-out from a computer people are using various video
cards, various drivers, various distros, or scan converters. I suspect
that many of them are happy (like me) and do get very good output
quality, but since there are many more failure points you hear
complaints from a number of people who are not satisfied with their
tv-out quaility.

cedar

Andres Valenciano wrote:
> Hello,
> I have been reading this list for a while, and I am wondering what has
> Tivo that makes it a better option in terms of video output quality
> compare to a DIY PC setup. I don't have access to a Tivo so I really no
> idea about its quality.
>
> If it is better... then, why can't we reach the same quality in our pcs?
> Or is just a matter of $?
>
> Your opinions?

Dredging up some points when this was previously discussed ;-),

Jitter: If you watch the crawler on CNN, it is smooth on
commercial products but has nervous twitches on myth. The
frame buffer should be updated during vertical synch but
myth hasn't had access to this timing. However, there is
work in progress to address this for video cards that report
vblank timing from their driver. Also, the frame buffer
is updated about every 1/30 of a second but because the
update is subject to a multitasking scheduler, the timing is
subject to when the process get the CPU. The "jitter reduction"
option can improve the timing but if the update straddles
the vertical refresh, there will be two updates during one
display frame and no update on the next causing this twitch.
You'll often see several seconds of smooth crawler followed
by a few seconds of twitches. Dedicated hardware can know the
refresh timing and not be subject to a process scheduler.

Deinterlace: Again using a crawler, when myth is paused the
characters look smooth but are jagged in motion. On commercial
products the characters are smooth in motion but have 'stair
steps' when paused. I believe this is because they are doing
something smart about updating the two interlaced fields so
that scan lines are updated like the original interlaced signal.
Myth has a frame of both fields updated once per full frame.
The default linear blend deinterlace code waters down the data
so it is neither a true reproduction of the interlace fields
nor is it as crisp as the original sampling.

Color reproduction: In a side by side comparison, my myth
picture doesn't look as life-like because the color and
intensity ranges are distorted by the bttv device while
recording. One would hope that the default contrast and
color settings would be 'line level' but apparently they
are not. Brightness and hue are close to neutral but contrast
and color at 50% are clearly distorted on each bttv card I've
tried. For contrast, white and anything near white it is fully
saturated while medium intensities are dull. For color, when
solids are bright but not over-driven, medium tones look washed
out.

Setting the v4l contrast to something below 40% (usually 36-38%)
greatly improves the picture. Color has greater variation
from one card to the next but is usually best between 25-40%.
Even after making these adjustments and adjusting the TV set
to match, the color and intensity is not as close to the
direct TV image as the commercial products.

Tvtime displays more vivid and realistic color in side by
side comparison with MythTV. This may be a result of their
luma correction code. If so, something like this may be
incorporated into MythTV also.

-- bjm

I was wondering if any of these issues are fixed by switching from a "dumb"
analog capture card to the pvr-250 card...

The "scheduler jitter" is my only reason I have not retired my tivo box now
that I have the remote/ir blaster situation at bay. I would assume that the
pvr-250 captures an interlaced picture as well and the playback might have some
jitter introduced ?? Anyone with pvr-250 chime in on this.

-mike

Previous post [ bjm@lvcm.com and others ]:

Jitter: If you watch the crawler on CNN, it is smooth on
commercial products but has nervous twitches on myth. The
frame buffer should be updated during vertical synch but
myth hasn't had access to this timing. However, there is
work in progress to address this for video cards that report
vblank timing from their driver. Also, the frame buffer
is updated about every 1/30 of a second but because the
update is subject to a multitasking scheduler, the timing is
subject to when the process get the CPU. The "jitter reduction"
option can improve the timing but if the update straddles
the vertical refresh, there will be two updates during one
display frame and no update on the next causing this twitch.
You'll often see several seconds of smooth crawler followed
by a few seconds of twitches. Dedicated hardware can know the
refresh timing and not be subject to a process scheduler.

Deinterlace: Again using a crawler, when myth is paused the
characters look smooth but are jagged in motion. On commercial
products the characters are smooth in motion but have 'stair
steps' when paused. I believe this is because they are doing
something smart about updating the two interlaced fields so
that scan lines are updated like the original interlaced signal.
Myth has a frame of both fields updated once per full frame.
The default linear blend deinterlace code waters down the data
so it is neither a true reproduction of the interlace fields
nor is it as crisp as the original sampling.

Color reproduction: In a side by side comparison, my myth
picture doesn't look as life-like because the color and
intensity ranges are distorted by the bttv device while
recording. One would hope that the default contrast and
color settings would be 'line level' but apparently they
are not. Brightness and hue are close to neutral but contrast
and color at 50% are clearly distorted on each bttv card I've
tried. For contrast, white and anything near white it is fully
saturated while medium intensities are dull. For color, when
solids are bright but not over-driven, medium tones look washed
out.

Setting the v4l contrast to something below 40% (usually 36-38%)
greatly improves the picture. Color has greater variation
from one card to the next but is usually best between 25-40%.
Even after making these adjustments and adjusting the TV set
to match, the color and intensity is not as close to the
direct TV image as the commercial products.


__________________________________
Do you Yahoo!?
Yahoo! Calendar - Free online calendar with sync to Outlook(TM).
http://calendar.yahoo.com

Michael Hill wrote:
> I was wondering if any of these issues are fixed by switching from a "dumb"
> analog capture card to the pvr-250 card...

Possibly. The hardware encoder is not subject to dropping
frames when other processes are temporarily busy. Also, the
timecodes should be more exact. The timecodes for software
encoding can be a little erratic when the systems is busy
which can contribute to jitter.

...
> Previous post [ bjm@lvcm.com and others ]:
>
> Jitter: If you watch the crawler on CNN, it is smooth on
> commercial products but has nervous twitches on myth. The
> frame buffer should be updated during vertical synch but
> myth hasn't had access to this timing. However, there is
> work in progress to address this for video cards that report
> vblank timing from their driver. Also, the frame buffer
> is updated about every 1/30 of a second but because the
> update is subject to a multitasking scheduler, the timing is
> subject to when the process get the CPU. The "jitter reduction"
> option can improve the timing but if the update straddles
> the vertical refresh, there will be two updates during one
> display frame and no update on the next causing this twitch.
> You'll often see several seconds of smooth crawler followed
> by a few seconds of twitches. Dedicated hardware can know the
> refresh timing and not be subject to a process scheduler.

Update =). Isaac did find some code that can check supported
cards to see when the refresh cycle completes. I've worked
out most of the to details to get good results for all of the
configurations that I can test and this is checked into CVS.
This gives the best results for nVidia cards and is somewhat
better for other cards as well. However, Isaac and I are seeing
very different results for live TV with software encoding.
Therefore, this needs more testing and will not be the official
video timing for 0.9.

To use the new timing, go to Setup->Tv Settings->Playback then
check "Jitter reduction" and "Experimental A/V Sync".

For nVidia, I see less CPU time for playback and virtually no
jitter for prerecorded and record in progress (provided the
recording isn't damaged, of course). For live TV I see very
little jitter. Isaac sees a lot when software encoding live TV
but it is nearly perfect with his PVR-250. I set up a PVR-250
last night and playback with the new timing looked very smooth
with each frontend systems that I tried.

-- bjm

That update did a world of good for my nvidia card with a pvr250, pretty
much eliminating the scrolling problem. I thought this was going to be a
long standing issue, didn't expect to see it fixed so quickly, at least
on the pvr250/350.

>
> Update =). Isaac did find some code that can check supported
> cards to see when the refresh cycle completes. I've worked
> out most of the to details to get good results for all of the
> configurations that I can test and this is checked into CVS.
> This gives the best results for nVidia cards and is somewhat
> better for other cards as well. However, Isaac and I are seeing
> very different results for live TV with software encoding.
> Therefore, this needs more testing and will not be the official
> video timing for 0.9.
>
> To use the new timing, go to Setup->Tv Settings->Playback then
> check "Jitter reduction" and "Experimental A/V Sync".
>
> For nVidia, I see less CPU time for playback and virtually no
> jitter for prerecorded and record in progress (provided the
> recording isn't damaged, of course). For live TV I see very
> little jitter. Isaac sees a lot when software encoding live TV
> but it is nearly perfect with his PVR-250. I set up a PVR-250
> last night and playback with the new timing looked very smooth
> with each frontend systems that I tried.
>
> -- bjm
>
> _______________________________________________
> mythtv-users mailing list
> mythtv-users@snowman.net
> http://lists.snowman.net/cgi-bin/mailman/listinfo/mythtv-users
>