Log in

View Full Version : Experts: Is encoding the same clip twice like this safe??


Pentajet
12th October 2007, 23:26
Since I am a total newbie to avisynth (meaning that I use it but have no clue as to how it works) I would like to know if this procedure that I am using to encode my videos is safe or if I am losing quality by doing this.

I am encoding videos by first loading my script into VIRTUAL DUB then saving an UNCOMPRESSED (very huge 40Gb) file using Direct Stream Copy to avoid color space conversions. The file is saved as a YV12 file.

Then I take the huge 40Gb file and load it using avisource in a separate script and use Blockbuster(method="noise") and ConverttoYUY2() on the big file. Then I load this script into CCE2 and make the final encode using a 4 pass.

The advantage of doing this over loading the file directly into CCE is that CCE alone takes about 13 hours to process one episode. With this method I do the episode in just 2.5 hours.

Thanks for you help.:)

Dark Shikari
12th October 2007, 23:36
Its perfectly fine, though I'd recommend you use a lossless compressor like HuffYUV for the clip rather than uncompressed, as uncompressed means you're disk-limited rather than CPU-limited, slowing down CCE.

Pentajet
12th October 2007, 23:55
Thanks for the quick reply! :)

stickboy
13th October 2007, 07:56
Uhm, it seems unnecessarily time-consuming to process the AviSynth script four times. Why bother separating the AVISource() and the Blockbuster() + ConvertToYUY2() scripts?

Combine them into one script, open it in VirtualDub, save to Huffyuv, and then have CCE do the four passes on the resulting AVI file.

Pentajet
13th October 2007, 13:07
Thanks for the advice Stickyboy. The grain produced by blockbuster is so fine that probably HuffyYUV would miss it anyway (I tested this). I said that I do it uncompressed to make sure I don't lose one grain of quality but somehow I felt uncomfortable using a subtle effect like Blockbuster beforehand (yes, that is stupid and doesn't make sense at all if there is no compression at all involved).

As for ConverttoYUY2 I do that because CCE needs that as input (right?) and even saving with DirectStreamCopy in VD and putting ConvertToYUY2 in the original script will still give a YV12 output somehow. Again I tested this. Maybe it's just me?

squid_80
13th October 2007, 14:40
The grain produced by blockbuster is so fine that probably HuffyYUV would miss it anyway (I tested this). Test again. HuffYUV is lossless - it's the exact same quality as uncompressed.

Adub
14th October 2007, 04:34
You can set the color output of VirtualDub in the Color Depth options under the Video tab.

Oh and how do you save a uncompressed file with Virtualdub using Direct Stream Copy? Using Direct Stream Copy, the file should be exactly the same size as your input, thus doing nothing.

However, if your are in fact using fast recompress, then I second the fact that you should be using Huffyuv, as it is lossless so there should be no difference between your input file and your output file interm of quality. Plus there is no colorspace conversion in fast recompress mode.

foxyshadis
14th October 2007, 07:16
Direct Stream Copy of an avisynth script is the uncompressed YUV/RGB that the script feeds vdub.

Adub
14th October 2007, 10:25
OHH!!!
Right, I forgot he was feeding via an avisynth script. Sorry.

AVIL
14th October 2007, 12:20
Hi,

I prefer lagarith lossless codec to make intermediate files. IMHO, size is more reduced, decode speed more high (alas encode speed not)