Darksoul71
7th August 2003, 17:55
Hi board,
Iīve owned a lot of capture adapters over the past years:
Voodooģ 3500TV, Elsa Erazor, Hauppauge WinTV, a Zoran based Hardware MJPEG Board,
Pinnacle Bungee MPEG and now a Canopus AVDC100.
So I think I really know a lot of cards and Iīm able to judge the quality I get.
My encodings are mainly TV shows first recorded to VHS and afterwards captured and converted
to DivX. Beside my ADVC100 I also own a MSI GeForce4 4400Ti with VIVO.
Over the last weeks I did some interesting testing (using different sources as VHS tape, VCR tuner, DVD player):
I captured with both capture devices using different combinations:
1) ADVC100 via S-VHS connection (Mainconcept DV-Codec @ 720x576)
2) ADVC100 via Composite connection (Mainconcept DV-Codec @ 720x576)
3) GeForce4 via S-VHS connection (Huffyuv @ 720x576)
4) GeForce4 via Composite connection (Huffyuv @ 720x576)
5) GeForce4 via S-VHS connection (Mainconcept MJPEG @ 720x576)
6) GeForce4 via Composite connection (Mainconcept MJPEG @ 720x576)
Quality:
1) S-VHS with ADVC100 looks VERY good ! A lot of details and vidid colors. A little bit grainy.
2) Composite with ADVC100 looks still good. Only a little loss in sharpness compared to 1)
3) Somehow the SVHS-connection of my GF4 has a problem. Looks very grainy compared to 1). Somewhat like the "Film effect" of the DivX playback filter at maximum.
4) Canīt really see a big difference to 2). Colors are sometimes a little less vidid but not that bad.
5) as 3)
6) as 4)
Compressibility:
As DivX5 somewhat smoothes the image I dislike the use of too strong denoise filters when encoding from analogue sources. I generally aim for 2 CDīs for a 2h movie so I can put 3 movies to a DVD-R. In my point of view 1 CD movies from analogue sources are crap as long as you donīt choose very low resolutions (<= CIF). XVid keeps a lot of details and has lower compressibility values. So IMHO you have two options if you go for MPEG4 when you archive analogue material:
a) use DivX5 and not so strong denoising
b) use XVid and some stronger denoising
Too low compressibility/quality values (in GKnot) result in compression artefacts all over. Letīs head back to my combinations:
1) Pain in the a$$ to compress. Needs either strong filtering or higher bitrates. For me this will not be an option unless I have some very special movie.
2) Better to compress. You loose some details compared to 1) but in general like this much more than 1).
3) Even more worse compressible than 1) and looks somewhat shitty because of the "grain" effect.
4) Best to compress
5) as 3)
6) as 4)
Conclusion:
This is just my point of view. I judged quality with my eyes when looking at the encoded movie and raw captured material on my 15" TFT display (which has good contrast and a sharp image). On TV itīs hard for me to see differences in quality. So here are my conclusions:
1) ADVC vs. GeForce4:
Although I can tell differences between these two capture solutions with raw material I canīt tell a big difference when I compare encoded material (either MPEG2, DivX or XVid) after being filtered (Denoise, Deinterlace, etc) and resized. If you pick one of my DVD-Rīs I produced from old VHS tapes or TV broadcasts in most cases I wanīt be able to tell you with which device Iīve captured it.
2) MJPEG vs. Huffyuv:
Here I couldnīt tell a difference between both codecs when I captured with my GF4. Only at very low bitrates (like 1 MB/s) I can clearly see some compression artefacts. I must agree that Huffyuv is better for recompression when you do some heavy filtering (esp. on old VHS tapes) than MJPEG but in general I donīt see any visual difference at all.
3) Composite Video vs. S-VHS:
Here I can clearly see a big difference between composite and S-VHS when I capture with my ADVC100.
Final words: I agree to anyone that S-VHS feed into a good capture device gives you the best results you can get (beside direct digital stream conversion).
Satellite->S-VHS->any capture card of your choise with Huffyuv will give best results but is still analogue. Unless youīre willing to spend a lot of bitrate for each movie (something like 4 MBit) and donīt use any deinterlace, you will always have a loss in details due to filtering, resize and compression algorithms.
What I really ask myself and also ask you:
Does it really make sense to buy expensive capture hardware with high bandwith that keeps a maximum of details if you have to reduce this details significant in order to compress this movie to your target size without compression artefact ?
Or wouldnīt it be better to use cheaper, lower bandwith capture hardware (TV card, graphic adapter with VIVO) that doesnīt keep all details but where you can compress the movie without having to use too much filtering ?
Iīm not shure about this anymore.
-D$
Iīve owned a lot of capture adapters over the past years:
Voodooģ 3500TV, Elsa Erazor, Hauppauge WinTV, a Zoran based Hardware MJPEG Board,
Pinnacle Bungee MPEG and now a Canopus AVDC100.
So I think I really know a lot of cards and Iīm able to judge the quality I get.
My encodings are mainly TV shows first recorded to VHS and afterwards captured and converted
to DivX. Beside my ADVC100 I also own a MSI GeForce4 4400Ti with VIVO.
Over the last weeks I did some interesting testing (using different sources as VHS tape, VCR tuner, DVD player):
I captured with both capture devices using different combinations:
1) ADVC100 via S-VHS connection (Mainconcept DV-Codec @ 720x576)
2) ADVC100 via Composite connection (Mainconcept DV-Codec @ 720x576)
3) GeForce4 via S-VHS connection (Huffyuv @ 720x576)
4) GeForce4 via Composite connection (Huffyuv @ 720x576)
5) GeForce4 via S-VHS connection (Mainconcept MJPEG @ 720x576)
6) GeForce4 via Composite connection (Mainconcept MJPEG @ 720x576)
Quality:
1) S-VHS with ADVC100 looks VERY good ! A lot of details and vidid colors. A little bit grainy.
2) Composite with ADVC100 looks still good. Only a little loss in sharpness compared to 1)
3) Somehow the SVHS-connection of my GF4 has a problem. Looks very grainy compared to 1). Somewhat like the "Film effect" of the DivX playback filter at maximum.
4) Canīt really see a big difference to 2). Colors are sometimes a little less vidid but not that bad.
5) as 3)
6) as 4)
Compressibility:
As DivX5 somewhat smoothes the image I dislike the use of too strong denoise filters when encoding from analogue sources. I generally aim for 2 CDīs for a 2h movie so I can put 3 movies to a DVD-R. In my point of view 1 CD movies from analogue sources are crap as long as you donīt choose very low resolutions (<= CIF). XVid keeps a lot of details and has lower compressibility values. So IMHO you have two options if you go for MPEG4 when you archive analogue material:
a) use DivX5 and not so strong denoising
b) use XVid and some stronger denoising
Too low compressibility/quality values (in GKnot) result in compression artefacts all over. Letīs head back to my combinations:
1) Pain in the a$$ to compress. Needs either strong filtering or higher bitrates. For me this will not be an option unless I have some very special movie.
2) Better to compress. You loose some details compared to 1) but in general like this much more than 1).
3) Even more worse compressible than 1) and looks somewhat shitty because of the "grain" effect.
4) Best to compress
5) as 3)
6) as 4)
Conclusion:
This is just my point of view. I judged quality with my eyes when looking at the encoded movie and raw captured material on my 15" TFT display (which has good contrast and a sharp image). On TV itīs hard for me to see differences in quality. So here are my conclusions:
1) ADVC vs. GeForce4:
Although I can tell differences between these two capture solutions with raw material I canīt tell a big difference when I compare encoded material (either MPEG2, DivX or XVid) after being filtered (Denoise, Deinterlace, etc) and resized. If you pick one of my DVD-Rīs I produced from old VHS tapes or TV broadcasts in most cases I wanīt be able to tell you with which device Iīve captured it.
2) MJPEG vs. Huffyuv:
Here I couldnīt tell a difference between both codecs when I captured with my GF4. Only at very low bitrates (like 1 MB/s) I can clearly see some compression artefacts. I must agree that Huffyuv is better for recompression when you do some heavy filtering (esp. on old VHS tapes) than MJPEG but in general I donīt see any visual difference at all.
3) Composite Video vs. S-VHS:
Here I can clearly see a big difference between composite and S-VHS when I capture with my ADVC100.
Final words: I agree to anyone that S-VHS feed into a good capture device gives you the best results you can get (beside direct digital stream conversion).
Satellite->S-VHS->any capture card of your choise with Huffyuv will give best results but is still analogue. Unless youīre willing to spend a lot of bitrate for each movie (something like 4 MBit) and donīt use any deinterlace, you will always have a loss in details due to filtering, resize and compression algorithms.
What I really ask myself and also ask you:
Does it really make sense to buy expensive capture hardware with high bandwith that keeps a maximum of details if you have to reduce this details significant in order to compress this movie to your target size without compression artefact ?
Or wouldnīt it be better to use cheaper, lower bandwith capture hardware (TV card, graphic adapter with VIVO) that doesnīt keep all details but where you can compress the movie without having to use too much filtering ?
Iīm not shure about this anymore.
-D$