Hello everybody,
I've been using MSP7 quite a bit and I'm very satisfied with the results, allthough I'm very demanding on the picture quality side, which imo is never good enough, especially for large screen tv's.
Recently, while browsing on various forums, I noticed that the Mpeg.now encoder is from Mainconcept. I went on to their site and downloaded their DVC pro codec (which is used by (semi)professionnals). Here is their advertisement :
The MainConcept DVCPro 25/50 DV Codec is our new product for our amateur and professional customers working with high-end Video. DVCPRO50 delivers considerably higher picture quality since it uses less compression than DV. The Codec comes with a full-featured DirectShow filter, making it compatible with any program which supports the standard DirectShow interface.
Since MSP uses directshow filters, I installed it, but MSP does not list it in its directshow filter/plugins list.
Does anyone know how to have DVC pro and MSP work together to capture DV ?
Thanks a lot in advance
Best regards
jacky serpenti
Mainconcept DVC Pro Directshow
The first question must be what are you using for your input files? If you are using ordinary DV or any other equally compressed lossless format, there is no point in going to DV50: you will never eliminate the artefacts from the earlier compression. GIGO.
The answer to your question is to use a frameserver to render the timeline to a 3rd party encoder. IMHO, it is unlikely that there will be a direct iinterface of such an encoder in MSP8, because it would involve a tonne of work (the standard encoder is already much modified from the MC one to make it "fit", even though it uses the MC algorithms).
The answer to your question is to use a frameserver to render the timeline to a 3rd party encoder. IMHO, it is unlikely that there will be a direct iinterface of such an encoder in MSP8, because it would involve a tonne of work (the standard encoder is already much modified from the MC one to make it "fit", even though it uses the MC algorithms).
[b][i][color=red]Devil[/color][/i][/b]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
-
jacky.serpenti
Thanks for your reply Devil.
As for the input files : DV type1 captured by MSP Video capture 7.3, through a firewire port.
I do not agree with you on the point of lossless compression : some time ago I gave one of my miniDV footage to a professionnal. He wanted to have the tape, not the captured file. He raised the point that once captured the footage is compressed; whereas, when captured with another codec directly from the tape, it can be less compressed (he used Avid Xpress and some sort of DVCAM codec). Afterwards I compared my captured footage and his : the difference was obvious. And when compressed to mpeg2, the difference remains : his encoded footage gave a significantly better picture than mine (mpeg.now setting the same).
So I still would want to know how to use the DVC pro filter at the capture stage.
On my system, I have two directshow filters to choose from in videocapture : panasonic msdv and winfast 2000XP (since I have a TV board in my PC)
Allthough I have installed the DVC Pro filter with the provided install procedure, it is not recognized by MSP 7.3 (or ignored)
I suppose there is some manual "plumbing" to do to have them both work together
Best regards
jacky serpenti
As for the input files : DV type1 captured by MSP Video capture 7.3, through a firewire port.
I do not agree with you on the point of lossless compression : some time ago I gave one of my miniDV footage to a professionnal. He wanted to have the tape, not the captured file. He raised the point that once captured the footage is compressed; whereas, when captured with another codec directly from the tape, it can be less compressed (he used Avid Xpress and some sort of DVCAM codec). Afterwards I compared my captured footage and his : the difference was obvious. And when compressed to mpeg2, the difference remains : his encoded footage gave a significantly better picture than mine (mpeg.now setting the same).
So I still would want to know how to use the DVC pro filter at the capture stage.
On my system, I have two directshow filters to choose from in videocapture : panasonic msdv and winfast 2000XP (since I have a TV board in my PC)
Allthough I have installed the DVC Pro filter with the provided install procedure, it is not recognized by MSP 7.3 (or ignored)
I suppose there is some manual "plumbing" to do to have them both work together
Best regards
jacky serpenti
-
Terry Stetler
- Posts: 973
- Joined: Thu Mar 17, 2005 3:34 pm
- Location: Westland, Michigan USA
That logic holds for analog captures, which are YUV data on the tape but get compressed to whatever codec you've chosen by the capture software.
IMO the only valid reason for him to want the tape is if he's editing on a MAC and you've captured with a PC. Here the file wrapper would differ: AVI vs. Quicktime. It's also possible that the software is doing an auto level adjustment during the file write.
DV-whatever is compressed in the camera by its firmware. What many euphemistically call a DV "capture" is actually nothing more than a file transfer. miniDV, DVCAM whatever....it doesn't matter.
All the capture software does is place this already compressed DV data in an *.avi or *.qt/*.mov (in the case of Macs) container so it can be written to the HDD.
Where an alternative codec becomes useful is when exporting projects or when doing analog captures to DV. Here they can deliver much higher quality results than with the default Microsft DV codec.
FRAMESERVING:
Duck soup in any Ulead editing product. A free frameserving plugin can be had from DebugMode;
http://www.debugmode.com/frameserver/
The frameserver appears in the MSPro export dialog as a new filetype: DebugMode Frameserver Files (*.avi)
Once the options (few) are set up and this small proxy "avi" is saved to disc everything stops until it can be loaded into virtually any encoder that supports *.avi input for processing. Set up the encoder, click GO and the timeline gets served to the encoder.
Copy the file (install folder)/VIO/dfscMSProOut.vio to the VIO folder of VideoStudio or Cool3D and you can frameserve from them as well. Handy for extending the filetype options for Cool3D.
There is another frameserver made by VideoTools.net that is commercial, but has advanced features. It can apply AVISynth scripts to the served video stream.
IMO the only valid reason for him to want the tape is if he's editing on a MAC and you've captured with a PC. Here the file wrapper would differ: AVI vs. Quicktime. It's also possible that the software is doing an auto level adjustment during the file write.
DV-whatever is compressed in the camera by its firmware. What many euphemistically call a DV "capture" is actually nothing more than a file transfer. miniDV, DVCAM whatever....it doesn't matter.
All the capture software does is place this already compressed DV data in an *.avi or *.qt/*.mov (in the case of Macs) container so it can be written to the HDD.
Where an alternative codec becomes useful is when exporting projects or when doing analog captures to DV. Here they can deliver much higher quality results than with the default Microsft DV codec.
FRAMESERVING:
Duck soup in any Ulead editing product. A free frameserving plugin can be had from DebugMode;
http://www.debugmode.com/frameserver/
The frameserver appears in the MSPro export dialog as a new filetype: DebugMode Frameserver Files (*.avi)
Once the options (few) are set up and this small proxy "avi" is saved to disc everything stops until it can be loaded into virtually any encoder that supports *.avi input for processing. Set up the encoder, click GO and the timeline gets served to the encoder.
Copy the file (install folder)/VIO/dfscMSProOut.vio to the VIO folder of VideoStudio or Cool3D and you can frameserve from them as well. Handy for extending the filetype options for Cool3D.
There is another frameserver made by VideoTools.net that is commercial, but has advanced features. It can apply AVISynth scripts to the served video stream.
Terry Stetler
I agree with Terry and I stand by my earlier statement that it is pointless re-encoding DV25 to DV50. As he said, DV "capture" into a computer is merely a transfer of data compressed within the camera before it is recorded on tape. Your "pro" clearly misled you. If his quality appeared different to yours, he had perhaps manipulated something.
[b][i][color=red]Devil[/color][/i][/b]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
-
jacky.serpenti
thanks Terry, thanks Devil
You are certainly right, my knowledge is still limited on these topics.
Anyhow thanks a lot for your advices and support.
I leave the DVC pro, and dig into the frameserver. I've been using tmpg encoder intensively and frameserve to it.
I did stop using tmpg in favour of ulead mpeg.now because of the necessity of making an avi file for the encoder. I did not know, until your help, that MSP pro could be made frameserving.
Best regards
jacky serpenti
You are certainly right, my knowledge is still limited on these topics.
Anyhow thanks a lot for your advices and support.
I leave the DVC pro, and dig into the frameserver. I've been using tmpg encoder intensively and frameserve to it.
I did stop using tmpg in favour of ulead mpeg.now because of the necessity of making an avi file for the encoder. I did not know, until your help, that MSP pro could be made frameserving.
Best regards
jacky serpenti
-
Terry Stetler
- Posts: 973
- Joined: Thu Mar 17, 2005 3:34 pm
- Location: Westland, Michigan USA
Devil;
There is ONE possible reason to transcode DV25 to DV50: it's higher colorspace resolution of 4:2:2.
PAL DV uses a colorspace of 4:2:0 and NTSC DV uses 4:1:1, both a lower quality than DV50's 4:2:2...though PAL DV's "disadvantage" is less than NTSC's.
This difference in colorspace only helps when compositing mulitple layers of overlays. I use high bitrate MJPeg in similar situations because the quality of the overlay will be better.
This makes sense for me because I'm using NTSC and am stuck with its 4:1:1 colorspace and the way it smears each color sample across 4 horizontal pixels. This can be a real pain because the foreground objects end up with a very visible halo, and with greenscreens this halo is green
In contrast MJPeg uses the broadcast standard 4:2:2 colorspace which only smears each color sample across 2 horizontal pixels. Not perfect, but believe me much easier to mask out.
Otherwise I agree that it's pointless.
There is ONE possible reason to transcode DV25 to DV50: it's higher colorspace resolution of 4:2:2.
PAL DV uses a colorspace of 4:2:0 and NTSC DV uses 4:1:1, both a lower quality than DV50's 4:2:2...though PAL DV's "disadvantage" is less than NTSC's.
This difference in colorspace only helps when compositing mulitple layers of overlays. I use high bitrate MJPeg in similar situations because the quality of the overlay will be better.
This makes sense for me because I'm using NTSC and am stuck with its 4:1:1 colorspace and the way it smears each color sample across 4 horizontal pixels. This can be a real pain because the foreground objects end up with a very visible halo, and with greenscreens this halo is green
In contrast MJPeg uses the broadcast standard 4:2:2 colorspace which only smears each color sample across 2 horizontal pixels. Not perfect, but believe me much easier to mask out.
Otherwise I agree that it's pointless.
Terry Stetler
Terry
I agree that DV colour space is marginal for colour keying (I suspect PAL is better than NTSC, in practice, because of the better horizontal colour resolution). However, I'm not sure that recompressing from DV25 to DV50 would not introduce an additional loss. Under those circumstances, if necessary, I think I would convert the clip to uncompressed 4:4:4 (RGB) in a secondary timeline then render the overlaid ensemble back to DV25 for intoduction in the main timeline as a new clip. Of course, this is personal and it would also depend on the length of the uncompressed clip (HDD space!).
That having been said, on the odd occasions I've done colour keying (not often) in PAL DV, I've not experienced undue fringeing.
I agree that DV colour space is marginal for colour keying (I suspect PAL is better than NTSC, in practice, because of the better horizontal colour resolution). However, I'm not sure that recompressing from DV25 to DV50 would not introduce an additional loss. Under those circumstances, if necessary, I think I would convert the clip to uncompressed 4:4:4 (RGB) in a secondary timeline then render the overlaid ensemble back to DV25 for intoduction in the main timeline as a new clip. Of course, this is personal and it would also depend on the length of the uncompressed clip (HDD space!).
That having been said, on the odd occasions I've done colour keying (not often) in PAL DV, I've not experienced undue fringeing.
[b][i][color=red]Devil[/color][/i][/b]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
[size=84]P4 Core 2 Duo 2.6 GHz/Elite NVidia NF650iSLIT-A/2 Gb dual channel FSB 1333 MHz/Gainward NVidia 7300/2 x 80 Gb, 1 x 300 Gb, 1 x 200 Gb/DVCAM DRV-1000P drive/ Pan NV-DX1&-DX100/MSP8/WS2/PI11/C3D etc.[/size]
-
Terry Stetler
- Posts: 973
- Joined: Thu Mar 17, 2005 3:34 pm
- Location: Westland, Michigan USA
RGB 4:4:4 is great if you can use it. On this side of the pond it's high-end and also used in devices like the RT.X100, which actually uses 4:4:4:4; RGB + an alpha channel (next system over has one 
).
Note: in some codecs you'll see a colorspace (or subsample) listing for 1:1:1. This is the same as 4:4:4. What's important is the ratio between the channels.
Another "ideal" is editing in lossless YUV space. This can encompass bitrates as high as 21 MB/S (no processing at all) to 10 MB/S (Huffman encoding only). HuffYUV uses the latter, which is just fine since Huffman encoding is lossless.
Problem is that while HuffYUV is a great editing codec it's a lousy playback codec, making previews a bit iffy when you have more than a couple of layers to composite. Unfortunately HuffYUV's developer dropped out of sight a few years ago so development has stalled
MJPeg does not have these weaknesses plus it has advantages: it's blistering fast in all directions (takes care of playback) and it's a mature codec still undergoing improvement.
At about 8-10 MB/S MJPeg becomes much like HuffYUV in that all that's done is Huffman encoding with no DCT compression. DCT's can introduce varying degrees of artifacts, which is another strike against compositing in DV as it uses DCT's extensively.
This scheme has been used in many pro hardware rigs, including the Matrox DigiSuite, for a very long time. All I'm doing is using lossless MJPeg in a software only mode when the quality needs for compositing override convenience.
).Note: in some codecs you'll see a colorspace (or subsample) listing for 1:1:1. This is the same as 4:4:4. What's important is the ratio between the channels.
Another "ideal" is editing in lossless YUV space. This can encompass bitrates as high as 21 MB/S (no processing at all) to 10 MB/S (Huffman encoding only). HuffYUV uses the latter, which is just fine since Huffman encoding is lossless.
Problem is that while HuffYUV is a great editing codec it's a lousy playback codec, making previews a bit iffy when you have more than a couple of layers to composite. Unfortunately HuffYUV's developer dropped out of sight a few years ago so development has stalled
MJPeg does not have these weaknesses plus it has advantages: it's blistering fast in all directions (takes care of playback) and it's a mature codec still undergoing improvement.
At about 8-10 MB/S MJPeg becomes much like HuffYUV in that all that's done is Huffman encoding with no DCT compression. DCT's can introduce varying degrees of artifacts, which is another strike against compositing in DV as it uses DCT's extensively.
This scheme has been used in many pro hardware rigs, including the Matrox DigiSuite, for a very long time. All I'm doing is using lossless MJPeg in a software only mode when the quality needs for compositing override convenience.
Terry Stetler