where the hell is the sharpness i get in 720x480 ? ( SVCD no

[LordVader]

ASUS Geforce 3 V8200 Deluxe ( TV in )
Pentium 4 - 1.7 Ghz
256 PC800 RAM
20 GB (DMA Enable) running at 15 MB/sec
Windows 98 SE
also i have a WinTV and a VideoHighway Xtreme (bt84

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

i'm from venezuela ( my english is not to good , so try to understand )

i know you guys are always helping everyone in the forum with problems similiar to mine but , i read ,and read , and read , and i don't understand anything because you give a diferent opinion every 2 seconds .

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

Right now i'm capturing realtime MPEG in 704x480 at 8000Kbits ( i have directv ) Interlaced captured image.

the son of a bitchs in my country i dont know what the hell they are doing to the system but every day directv looks worst ( macro blocks , blurry image , to much brightness , etc )

so the little of sharpness i'm getting in my captures , i'm losing it when recomressing my videos to SVCD with tmpgenc.

of course the image is interlaced ,and i have to apply De-Interlace filter ( Double ).

Now my question is : what i need to do to preserve the quality of my videos , because my videos in 704x480 looks worst than to much svcds out there .

for example the videos of telejunkie : i mean i saw the video of britney (SVCD), i'm a slave 4u captured from VHS and i have to deal with a worst quality in 704x480.

and some videos in vcd has much more quality than mine , how is this possible ? , people capturing from vhs in 352x240 getting more quality than me in 704x480 ? i don't believe it .

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

that's why i need answers now :

I'm not planning to capture with virtualdub , i have that program and i always drop frames , even with the system i have .
i don't know why. I think the chip of the geforce don't have hardware compression.

Anyway , if i need to start capturing in another format and use virtualdub , i will do it .

Now what should i do (STEP by STEP) to preserve the image quality when recompressing to SVCD or VCD . Adding de-interlace filters in this softwares will make me lose sharpness , so please , be creative .

I want to hear something new .

[p®è§+ïgé]

quite honestly, your computer isn't fast enough to do realtime mpeg2 encoding. i'm also curious as to what you're using right now to do RT capturing? there's not too many mpeg2 codecs that aren't integrated into programs.

asus implements a hardware motion jpeg capture chip on their deluxe cards. use their hardware to capture to avi format using asus' asv2 codec included in the latest drivers. capture at 720x480 at the desired quality setting, then re-encode the avi to an SVCD compliant stream using tmpegenc or any other program of your choice. although mjpeg is a lossy format, with noise and deinterlace filters (yes, a deinterlace filter SHOULD BE USED - an interlaced image is basically at half resolution since only one field is displayed), an acceptable video stream can be acheived. i would recommend that you use avi_io to capture if you have problems with sound synchronization or if you capture large files that FAT32 file systems can't handle.

re-encoding a lossy format like mpeg2 will always result in further loss of quality. proper realtime mpeg2 capture would require professional equipment - i do not recommend that anyone capture RT mpeg2 without the proper hardware. a properly encoded svcd within svcd specifications (2.6Mbps max bitrate) from a lossless format will still look better than your realtime capture at 8Mbps. The more time your computer has to encode video, the better your final result will be - as an analogy, realtime encoding is like printing something in draft mode. you will still get a picture, but the quality would be a lot worse than if you printed in high quality mode, where the printer takes time to properly position the ink droplets.

the absolute best way to capture (if your system can handle it) would be to use a lossless codec like huffYUV at 720x480 (or even 480x480) and then re-encode using a multi-pass variable bitrate svcd template. the only problem with this would be that you might need a LOT of hardrive space and a fast hardrive(faster than any IDE drive is).

i think the best option for you would be to capture using asus' hardware codec and then re-encode to an svcd stream, considering your limited amount of hardrive space.

[LordVader]

OK , wait , wait , wait ... you're confused

i'v never mention MPEG-2 , read my post again man. hehe

i'm using Ulead Video Studio 5.0 with a cracked DLL that allows to me capture in MPEG-1 at high bitrate.

i do this everytime , also , before i buy the P4 , i had a pentium iii 450Mhz 64RAM , and i always was capturing my videos in 640x480 ( preview mode ) 6000 Kbits without 1 droped frame .

I have been using this technique for 2 years . Ulead lets you select VFW capture devices without any problem . and the processor of the computer do all the job of compression.

if you use "I Frames only" , and a high bitrate , you will use less processor and that's why i have good perfomance at higher resolution settings . give it a try , you won't be dissapointed.

--------

anyway , i capture in MPEG-1 because i cut comercials and stuffs in MyFix , is very easy to handle.

--------

I drop frames using HuffYUV my friend , i tell you .

Huffyuv is a piece of shit , i don't have luck with that.

--------


For a second forget the format of my videos , just think that my videos in MPEG-1 are AVIs


help me to solve the problem of resolution ,and how to preserve the sharpness when reencoding to lower resolutions.

[p®è§+ïgé]

i don't think i fully understand your problem.

you are losing quality when re-encoding your captures to an svcd compliant stream. i understand that. but i still don't understand what "resolution problem" you're having. you claim that you haven't changed the way/format you capture in for 2 years. if your source has decreased in quality, then your captures will obviously reflect that. if you are trying to improve the quality compared to the source, then the only recommendation i have is to apply noise filters and the "soften block noise" option in tmpeg, or any other program that can apply these algorithms.

you are capturing in a format that is not meant to capture in realtime. the format you capture in IS important, and cannot be overlooked. even though you are using high bitrate I frame capture, mpeg (even mpeg1) cannot be properly encoded in realtime. if you are capturing I frame only mpeg1 (which is basically capturing motion jpeg), then the best way for you to capture with minimal loss of quality would be, as i mentioned in my earlier post, to use the asus hardware mjpeg codec, and then re-encode to an svcd stream. with hardware encoding, you will have minimal quality loss and a lower cpu utilization. furthermore, the hardware codec is designed to encode in realtime, unlike mpeg software codecs. the resulting avi can still be fully edited before encoding as well. also, in capturing to mpeg format, motion compensation, etc. are applied (poorly) to the video stream. when you re-encode into another mpeg format, again, motion compensation is applied, and the resulting image will loose quality compared to the first encode/capture.

[LordVader]

read my post againg , hehe

my problems is with the Interlaced image .

i mean , is ovious that to delete this SCANLINES i need to apply the de-interlace filter , right ?

well , i lose quality(SHARPNESS) when i do this .

my question is: wich method i can use to recompress my videos,.

is there any way to recompress the videos without adding de-interlace filter and delete this scanlines?

this will help to preserve sharp image.

NOTE : forget the format of my videos , just imagine that my MPEGs are AVI's .

just , spit it out .

[VidGuy]

LordVader, don't be so rude..... People are trying to help but you're not clear.

If you want to create SVCDs, why de-interlace. SVCD supports interlacing - just encode to a 25 or 30 fps interlaced signal - works great, no crappy resolution loss.

Of course, if you're capturing a movie, you can inverse telecine it back to 24 fps, but that's a whole 'nother topic.

[TurboRunner]

Ok for best image quality you should capture using uncompressed, huffyuv or mjpeg avi capture at 720x??? and edit you video to what you want then convert to svcd or vcd compliant mpeg.

But as you say you cant be stuffed recapturing your footage using those codecs and you want to convert you 720x480 mpeg file to 352x480 complait vcd or svcd with best quality while getting rid of those interlaced lines heres how
in Virtual Dub
open you movie file
Select Full processing mode under video
Select compression to uncompressed format
Then select Filters
Then add the Resize filter
and put in the following settings
(I guess you want to have an Ntsc video)
Put the new width of 352 and the new height of 240
The under filter mode choose one of the Precise Bicubic
and select ok
Then you can save the file as an avi then use tmpgenc to encode to you svcd or vcd standard.

You can frameserve if you know how.

hope this helps

TurboRunner

[p®è§+ïgé]

ah!

i finally understand your problem.. (LordVader, why didn't you say so in the first place!?) okay, i think the problem that you're having is that you apply a deinterlace filter while you're encoding. but at the same time, you're encoding to a non-interlace mpeg. what you have to do is apply the deinterlace filter, and then encode to an INTERLACE mpeg. i had the same problem before, this solved it.

[davidian]

LordVader, how do I get that cracked DLL? Do I have to have a geoforce card to use that method, or would any card do?

[VidGuy]

<TABLE BORDER=0 ALIGN=CENTER WIDTH=85%><TR><TD><font size=-1>Quote:</font><HR size=1 color=black></TD></TR><TR><TD><FONT SIZE=-1><BLOCKQUOTE>
On 2001-09-30 12:59:13, p®è§+ïgé wrote:
ah!

i finally understand your problem.. (LordVader, why didn't you say so in the first place!?) okay, i think the problem that you're having is that you apply a deinterlace filter while you're encoding. but at the same time, you're encoding to a non-interlace mpeg. what you have to do is apply the deinterlace filter, and then encode to an INTERLACE mpeg. i had the same problem before, this solved it.
</BLOCKQUOTE></FONT></TD></TR><TR><TD><HR size=1 color=black></TD></TR></TABLE>

If you're going to deinterlace, you encode to a non-interlaced mpeg. Otherwise, you encode to an interlaced mpeg. deinterlacing converts an interlaced picture to a non-interlaced picture. For SVCD that is not required, since one of the improvements SVCD made was to allow interlaced video. If your picture looks better in interlaced mode, it's because the deinterlacing didn't work very well. If your going to code interlaced, don't de-interlace - you just lose picture resolution that way.

[TIbrO]

Hmmm... as I read this thread I also had trouble trying to figure out what this dude wants... He is actually doing a useless step. Why deinterlace? Do not deinterlace if you are going to make an SVCD. Also, he is capturing directly to MPEG-1, which doesn't support interlacing. You should get AVI_IO and PICVIdeo MJPEG 2.0. This is the combination I use, and I get no dropped frames and excellent quality from captures. I never liked HUFFYUV because it doesn't let you control the MJPEG compression of it like PICVideo does. So you're always forced to do an almost uncompressed capture, which takes a lot of bandwidth and makes even me drop frames at full res full framerate in AVI_IO. I think PICVideo is the best out of all those MJPEG codecs, but some peolpe like Huffyuv because "it gives better quality", but I honestly think that PICVIdeo set to 18-20 looks like a huffyuv capture.

So anyway, use AVI_IO to capture, it's the best capture program. Use PICVideo MJPEG 2 as your compression, because it's so light on the bandwidth and still looks great, basically identical to your source if set from 18-20. Then use TMPGEnc's SVCD Template to encode a five minute clip of your DirecTV captures. DO NOT check in soften block noise if sharpness is what you want to keep. Soften Block noise sacrifices sharpness detail in order to minimize the pixelization and macroblocking at high motion and very detailed scenes. Personally, I rather see a softer picture with no blocks than a super sharp pixelated picture, so I always use Soften Block Noise. Buena Suerte!

[SatStorm]

Capturing 720X576 from TV or VHS source, creates interlace artifacts
When you want to convert to 480X576 SVCD, and you don't de interlace with TMPEG, you have blocks. On all motion scenes. It has nothing to do with the output source
The output source is interlaced again. But you have de-interlace first.
You see, imagine a puzzle. You love the picture of the puzzle, you want the same. But on smaller size. What is better?
Copy each puzzle part seperately and create smaller ones, or solve the original pazzle, photo it, and then create a new puzzle with the photo of the size you want?

And interlace picture is a kind of a puzzle picture you know...

I had a week ago post a question for de-intelace experts, nobody answer. I wanted to discuss that thing.

Since I de - interlace the source (odd film/adaption), I have 50% increase of the quality image and also much less blocks. And we talking for music videos here, full motion ones. With a CVD resolution (352X576) and a bitrate about 2600, you simply get PERFECT video. No blocks. I get blocks when I encode fast motion videos at a bitrate about 2350
I also like to say, that I de-interlace only with tmpeg. It does (in my eyes) doing better job than vdub.

[LordVader]

i already solve my problem , thanks for helping anyway .

for those that are curious :

i'm just capturing now in MPEG-2 instead of MPEG-1.

the aspect ratio i was using before was "Square Sample"

and for some kind of reason windows media player opens the video in 720x720 :

example: oooooooooo
o o
o o
o o
o o
o o
o o
oooooooooo
i see scanlines in that way.

, what i have done is switch the aspec ratio to 3:4 and now the video opens in it real size . 720x480

now the codec of mpeg is resizing the vertical lines from
720 to 480 , and that's why don't show scanlines.

i just go to tmpgenc i encode to svcd and everything goes ok.

hahhahahhahh my english is so bad , damn whatever guys , i'm just having fun trying to tell you my history.


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

o i forgot ... wich program is more easy to make svcd menus & chapters ?

sorry for putting this question in the wrong section