View Full Version : Maximizing HD to SD Quality


Pages : 1 2 3 4 [5]

Brian Luce
March 3rd, 2011, 10:52 AM
What's you reason for going from Cineform to Vegas? I'd either pick on or the other and then import to VD. Seems like that's an extra step and you risk a generation loss, and at the least, you're making work.
Also, does your Cineform software come with a down scaler? Cineform makes a good one.

Arunas Jocius
March 3rd, 2011, 05:34 PM
I would stay in HDV, edit as normal in a 1440x1080I project. Export a HDV file in the same format and then use TMPGenc V5 to encode for SD DVD with its preset. A lot easier and best downconvert and encode that I have experienced both for HDV and AVCHD. You could even use TMPGenc Authorworks and do the whole thing including authoring in the one program. I would normal do my video in Edius V6.02 where I would export a Canopus HQ file then use TMPGenc. Cineform would be another alternative but for one track of HDV it isn't worth going to an intermediate with modern PC's.

TMPGEnc - Products: TMPGEnc Video Mastering Works 5 (http://tmpgenc.pegasys-inc.com/en/product/tvmw5.html)

TMPGEnc - Products: TMPGEnc Authoring Works 4 (http://tmpgenc.pegasys-inc.com/en/product/taw4.html)

For Bluray just export from Vegas with the Bluray preset for authoring in DVDArchitect.

Ron Evans

Thank you Ron, I will try TMPGenc, but would be great to use the programs that I already have. Plus a new learning curve... I am a bit familiar with Vegas already and would like to do editing with this software. I didn't mention video editing above in my steps, since for test purposes I only straight convert 1 minute of video. Once I get algorythm, I will start working for longer video files and will do some editing, of course.

Arunas Jocius
March 3rd, 2011, 05:41 PM
What's you reason for going from Cineform to Vegas? I'd either pick on or the other and then import to VD. Seems like that's an extra step and you risk a generation loss, and at the least, you're making work.
Also, does your Cineform software come with a down scaler? Cineform makes a good one.


Well, after I understand the whole process and find out parameters I need to use in every step of my procedure, I will move one from my 1 minute test video file, to longer video files and need to edit them, before final reendering. I have used vegas few times and liked it, would like to use it for my future video editing. Cineform ConnectHD does only capturing, no editing.

you say you would pick one or another, but this is where I do not understand, why so many Vegas users mention "Cineform avi" as their source file. This was one of my questions above. Also m2t files seem like a lot smaller than avi file.

This thread repeats number of times, that best down scaler is VD with Lanzos, this is why I decided to try that route.

Ron Evans
March 3rd, 2011, 07:04 PM
A few years ago both PC's and the software were not powerful enough to decode and edit HDV ( in fact it goes back to DV originally) So intermediate formats like Cineform were introduced. Canopus used in Edius intent was the same. They trade off file size for CPU power. The load on the CPU is less but the files sizes are bigger since they in the main decode a long GOP file format to a format that encodes on a per frame basis. This is easier for the CPU to manipulate in an NLE but requires a greater hard disc throughput hence the adoption of RAID controllers to get a greater disc throughput.

Things have changed. CPU's are now a lot more powerful and the NLE decode algorithms have got fast enough to edit most files native.

I have both Vegas 10 and Edius V6 both will easily edit HDV and AVCHD native with no need for intermediate files.

On the Edius forum there were also lots of fans of the VD approach but the general opinion now is that TMPGenc is easier and better. In my tests I find TMPGenc easier and better on a number of accounts. The VDUb output tends to loose some detail and colour needing to come back into an NLE to apply some sharpening and colour boost. Neither is necessary with TMPGenc. I go straight from TMPGenc to authoring. TMPGenc uses Lanzos too

Ron Evans

Arunas Jocius
March 4th, 2011, 03:53 AM
Thanks, this clarifies a lot. My PC is rather old (4-5 years), on the other hand, so I might try both options:direct NLE capture and editing or using Cineform intermediate.

Capturing with NLE seems straight forward. Select 1440 X 1080 50i project template (to be the same format as HDV on cassette, start capture. It will create number of m2t files. I can then do edit and once I have done with editing, I should be able to reender to avi file and then move on to TMPGenc?

I still have my questions 1 and 2 that are not clear in my head:
1. Do I want to keep video interlaced all the way, or do I need to think on deinterlacing. If I watch it on LCD and if it is progressive, should I leave that job to dvd player or would I be better off deinterlacing video material? Please don’t kill me on this if LCD’s actually are ok with interlaced material and there is no need to worry about it…
2. Aspect Ratio is different in source video, DVD standard and TV used to watch that video. My thinking is that changing aspect ratio will decrease video quality, but something needs to be done here. What route do I go? Should I resize in Virtual Dub into 720 X 540 to keep same aspect ratio as source? Or to 720 X 405 to keep same ratio as TV? If I do so, will DVD authoring software (DVD Architect) understand selected aspect ration and will not automatically resize back to 720 X 576 ???
3. .....................Got the answer to this question
4. .....................Got the answer to this question, understand it is the same quality of capture.
5. .....................Got the answer to this question

Going through Cineform ConnectHD created me some challenges, not sure why Vegas does not recognize avi file created by Cineform ConnectHD.

"Challenges along the way:
1. First challenge came while capturing… I have several options to choose and suggested file formats are m2t or avi. Many users talk about Cineform avi in this thread so I tried to capture into avi. Captured file was huge. 1 min 49 sec of capture ended up in 1 314 MB size avi file. I then try to drop captured file into vegas timeline and get error message “CFHD is not supported”… I then try to capture into Cineform intermediate and MPEG2-TS format. This creates two files: avi (same size as in first attempt) and m2t of 295 MB (4,5 times smaller!!!). Dropping onto Sony Vegas was not successful with avi again, but m2t file went fine into vegas timeline…"

Ron Evans
March 4th, 2011, 07:31 AM
If you are capturing from HDV then capture native files which are m2t. These will then be a copy of the data on the tape with no conversion. Place on the timeline and see if your PC can edit native. If you can then do not bother with an intermediate file conversion. Leave interlace and do not bother with aspect ratio changes as the presets in TMPGenc will do all this for you. You can edit in HDV, export a HDV file and then encode in TMPGenc as a very easy way to try everything out. Once you have that sorted out then it may improve the quality to export a higher quality file from Vegas and there are several to choose from but all will create very large files and you may decide that it isn't worth it over a 2 pass high quality HDV export. Commercial DVD's are mainly interlace so what everyone sees on their TV is the result of the DVD player and LCD scaling and deinterlacing. The higher the quality of this scaling and deinterlacing the better the viewed picture.

Ron Evans

Arunas Jocius
March 4th, 2011, 08:17 AM
thanks again, ready for testing, downloading TMPGenc trial.

Jeff Harper
March 4th, 2011, 08:47 AM
Aruna, please report back, I haven't downloaded trial yet, I don't have time now to test it now, I'm focusing on selecting lenses right now, and editing my current projects.

Jeremy Dallek
March 13th, 2011, 01:10 PM
Use YV12 (both as output from Virtual Dub to the compressor and as the Lagarith compression mode) if you want to export 4:2:0 from Virtual Dub (DVD video is 4:2:0).

If the the designated format is DVD, how important is choosing YV12 here (vs YUY 4:2:2), rather than letting vegas do this when rendering the mpeg2 stream? Thanks!!

Randall Leong
March 13th, 2011, 02:29 PM
To export 4:2:2 YUV encoded with Lagarith, you need to tell VirtualDub to export YUY2 to the compressor and set Lagarith to encode as YUY2 (defaults for both are RGB).

To set VirtualDub to outputting YUY2: From "Video" on the menu bar, select "Color Depth..." and under "Output format to compressor/display", select "4:2:2 YCbCr (YUY2)".

To set Lagarith to encoding as YUY2: From "Video" on the menu bar, select "Compression..." then select "Lagarith lossless codec" then click the "Configure" button, and under mode select "YUY2".

Use YV12 (both as output from Virtual Dub to the compressor and as the Lagarith compression mode) if you want to export 4:2:0 from Virtual Dub (DVD video is 4:2:0).

True if you are using a direct encoder to make the final output. However, if the lossless file is to be imported back into an NLE that natively works only in RGB, then leave that as RGB.

If the the designated format is DVD, how important is choosing YV12 here (vs YUY 4:2:2), rather than letting vegas do this when rendering the mpeg2 stream? Thanks!!

Like I stated above, Vegas and most other NLEs work natively in RGB. YV12 gets converted to RGB in the NLE itself and then re-converted back to YV12 during the re-encoding to MPEG-2. You want to avoid multiple re-conversions, in this case.

Brian Luce
March 17th, 2011, 04:13 AM
I just did my first test with TMPGEnc and the results for and HD to SD DVD are far superior to what I was getting with the standard Vegas MPEG2 Template+AC3 double render. Sharper, cleaner, brighter and fewer jaggies. Render times seem a lot faster than Virtual Dub. The TMPGEnc software is easy to use as well. Simple interface, and no double video/audio render as with Vegas. I don't even know if you can author from TMPGEnc, I'm just using DVDA for that. WIN.

Jim Snow
March 17th, 2011, 08:58 PM
It's a bit ironic. TMPEGEnc is a consumer level editor that just happens to have an amazingly good resizing function. Frankly it does a much better job of resizing than any of the 'Pro' editors. Seems backwards doesn't it? I never have understood why the Pro editor developers haven't dealt with this.

Jeff Harper
March 17th, 2011, 09:19 PM
Brian, I tried it last night, and it made me crazy. Which codec do I select, etc. There were other settings that I had no idea of how they would affect things, and I quit. Coming from HD link trial, I didn't like it, and the consumerish interface was a bit of a turn off.

On the other hand, I have footage piling up that I'll need to do something with, so I'll give it a go again soon. At $100, it is a bargain.

Part of the problem is just me...I'm stressed, and every little complication pisses me off right now; that can't be blamed on Tmpgenc.

Brian Luce
March 18th, 2011, 03:20 AM
Jeff, the interface is a bit consumerish, something like Premiere Elements where you click a tab that walks you through the process. I find it intuitive and easy. I'm not sure if I'm using it correctly because I haven't read the manual.
Here's what I've been doing with 1080p AVCHD files.
-From Vegas 10 render out a avi file with the Lagarith codec (also tried this with the 1080p hdv codec)
Set the audio in the Custom tab. I've been using WAV for audio.
-import to TMPGEnc which guides you through the encode process. I change the "Clip Settings" to progressive and 1:1 pixel.
-in Format set it to 16x9 and linear PCM for audio
-click encode
-take the TMP created avi and import it to DVDA and burn away.


Have no idea if this is the best method, but for sure it looks a lot better than staying in Vegas for your encode.

The software also has some sort of timeline based editing system, haven't played with that yet, not so impressive looking though.

Mark Howells
March 18th, 2011, 07:36 AM
Can I ask, which TMPGenc software is required for this encoding as there appears to be a number, all at different prices.
ie Video Mastering Works 5, Moviestyle, TMPGenc Plus 2.5, Authoring Works 4.

Jeff Harper
March 18th, 2011, 07:41 AM
Brian, giving an avi file to DVDA means it will have to re-encode the avi to convert it to mpeg 2...wouldn't you want to render out an mpeg 2 for DVDA? Or am I misunderstanding something here?

Garrett Low
March 18th, 2011, 07:57 AM
Jeff,

Get VirtualDub to resize. It's free and the steps are the same as what you'd have to go through with TMPGEnc:

1. Vegas edit then output HD AVI
2. Take HD AVI and convert using VDub (I typically use Lanczos3 algorithm) to SD AVI
3. Take SD AVI into Vegas to create DVD compliant mpg and AC3 files
4. Take SD mpg into DVDA to author DVD.

That way DVDA does not re-rendering of the edited material. The results I get are noticeably better than using Vegas to resize and are as good as TMPGEnc which I have tried. And as I mentioned, VirtualDub is free. I've used this work flow for footage from a variety of HD codec from HDV on up to XDCAM HD footage.

If I were using TMPGEnc I would just substitute it where I use VDub. The advantage you get with TMPGEnc over VDub is that you could do some editing in TMPGEnc. For me that does not justify the $100 expense.

-Garrett

Brian Luce
March 18th, 2011, 09:11 AM
Hi Garret, I agree VD is great, but somehow I'm getting better and faster results with TMP. I still use VD for the stabilizer.

Jeff Harper
March 18th, 2011, 09:20 AM
Garrett, thanks for the tip on Virtual Dub. I am familiar (vaquely) with it seemed a bit complicated to me when I tried it, and a bit slow.

I don't mind the $100 of tmpgenc. not that I have money to burn!

Steven Reid
March 18th, 2011, 11:08 AM
Garrett, thanks for the tip on Virtual Dub. I am familiar (vaquely) with it seemed a bit complicated to me when I tried it, and a bit slow.

I don't mind the $100 of tmpgenc. not that I have money to burn!

Jeff, Garrett's procedure is exactly my workflow, too, and I get stunning results on resizing HD-->SD avi files. If you're busy and stressed, then figure out VDub when you're not. It is actually quite simple to use, despite its appearance. Yes, many of the menus have a lot checkboxes and dropdowns, but so does Vegas, right? ;) I simply jumped in and let intuition take over.

As for speed, VDub actually tells you during renders how fast it is going. My source material is 24p, and on my overclocked Q9550 system, using the 64-bit version of VDub set to the fastest speed, I get ~30 fps renders on the resize. I dump the rendered avi's back into a Vegas project to then render MPEG-2 files for DVD and Blu-ray.

It's definitely not a one click operation from HD source footage in Vegas to DVD Architect-ready files, but the results are worth it!

-Steve

Jeff Harper
March 18th, 2011, 11:16 AM
Thanks for the encouragement Steven, I appreciate your outlining your workflow. I'lll come back to this soon.

Brian Luce
March 18th, 2011, 11:19 AM
Jeff, I'm not sure Lagarith is the best codec to use or not. So many codecs, so little time. But Lagarith was recommended for Virtual Dub so I'll stick with it. I also used HDV and that worked good too.Also tried HuffYUY but Vegas wouldn't see it, others have had the same issue with it.. It's lossless so I shouldn't be getting a generation loss. It makes some huge files, but I delete them after the project is finished.

Mark, I got the mastering suite. $99 but it shows up as $110 in your cart so you have to delete the CD disc to get it to $99.

Brian Luce
March 18th, 2011, 11:23 AM
Jeff,

Get VirtualDub to resize. It's free and the steps are the same as what you'd have to go through with TMPGEnc:

1. Vegas edit then output HD AVI
2. Take HD AVI and convert using VDub (I typically use Lanczos3 algorithm) to SD AVI
3. Take SD AVI into Vegas to create DVD compliant mpg and AC3 files
4. Take SD mpg into DVDA to author DVD.

That way DVDA does not re-rendering of the edited material. The results I get are noticeably better than using Vegas to resize and are as good as TMPGEnc which I have tried. And as I mentioned, VirtualDub is free. I've used this work flow for footage from a variety of HD codec from HDV on up to XDCAM HD footage.

If I were using TMPGEnc I would just substitute it where I use VDub. The advantage you get with TMPGEnc over VDub is that you could do some editing in TMPGEnc. For me that does not justify the $100 expense.

-Garrett

Garret, is step 3 necessary? won't DVDA do that just as happily? I think you can from VD to DVDA. I think that's how I did it.

Garrett Low
March 18th, 2011, 12:42 PM
Brian,

You could pull the AVI into DVDA but then DVDA would transcode the mpg2 file using it's preset settings. In many cases that may not yield the best results. By taking it back into Vegas it gives me a few advantages. First I get better control over the settings for the encoding. I can decide to use variable two pass or constant bitrates. DVDA does not allow you to control that. Also, I find it easier to set chapter points in Vegas and tell the encoder to set an I-frame at each chapter point. That way it allows me to precisely choose where the chapter points are. If you set chapters in DVDA you are limited to hitting a beginning of the GOP which means your chapter point could be as much as 7 frames off. Sounds like nothing but it's a quarter of a second and sometimes that does make things seem weird if you catch a mid word.

I believe that you might be able to get a little better quality of mpg encoding from TMPGEnc if you really want to take time and tweak the settings. One of the things that TMPGEnc allows is for you to adjust the quantization matrix. Something that is better left to those who have a lot of experience but it is an option that is not available with Vegas' built in encoder.

Brian, to get back to your quality issue, what algorithm were you using when you use VD? The default is not Lanczos3. I believe it defaults to Bicubic. You might try to check that and see if you still see a difference. Also check to see that the resolution and PAR correct for the VD encoding so that Vegas is not resizing the video when you create your widescreen mpg file. If you're using the same algorithm and same compression codec, it should look the same.

-Garrett

Arunas Jocius
March 18th, 2011, 02:32 PM
Trying to resize 1440X1080 50i material into 720 X 540 50i (later to be used to make dvd) with TMPGenc. I have unchecked deinterlace options (Deinterlace mode - None) withing filters and only left resize filter active. Next step is Format in TMOGenc workflow where I select Lagarith codec and size 720 X540 again. There is no selection for deinterlacing here. And yet, I still get progressive output in result avi file... I can see that in file properties once I feed file back to Vegas. Whad do I do wrong??? I try to keep video interlaced during the whole project.

Brian Luce
March 18th, 2011, 11:41 PM
Brian, to get back to your quality issue, what algorithm were you using when you use VD? The default is not Lanczos3. I believe it defaults to Bicubic. You might try to check that and see if you still see a difference. Also check to see that the resolution and PAR correct for the VD encoding so that Vegas is not resizing the video when you create your widescreen mpg file. If you're using the same algorithm and same compression codec, it should look the same.

I used the Lanczos3, I always followed Perrone's lead since he understands this stuff so well. I think I did 720x480, but I didn't give VD a fair shot and most likely didn't get near the results VD is capable of and didn't have time to buckle down with it , also I didn't like the interface for some reason. But with TMPG, right out of the gate, first attempt with the trial version and BANG, dramatic improvement. So I pulled the trigger and pulled out the Visa Card. Don't tell my wife.

For sure though, VD has crazy power for a free program.

Brian

Phil Lee
May 10th, 2011, 12:23 PM
Hi

Just found an AVISynth script (via someone else that found it at Doom9) that seems to be doing a fantastic job at resizing HD to SD, and does it without any de-interlacing required avoiding some nasty artefacts that can introduce sometimes.

So thought I would share, here is the resizing script:



# Converts 1080i to 576i
clip = AVISource("Source.avi").AssumeTff()

IResize(clip,704,576)
function IResize(clip Clip, int NewWidth, int NewHeight) {

clip.SeparateFields()
Shift=(GetParity() ? -0.25 : 0.25) * (Height()/Float(NewHeight/2)-1.0)
E = SelectEven().Spline36resize(NewWidth, NewHeight/2, 0, Shift)
O = SelectOdd( ).Spline36resize(NewWidth, NewHeight/2, 0, -Shift)
Ec = SelectEven().Spline36Resize(NewWidth, NewHeight/2, 0, 2*Shift)
Oc = SelectOdd( ).Spline36Resize(NewWidth, NewHeight/2, 0, -2*shift)
Interleave(E, O)
IsYV12() ? MergeChroma(Interleave(Ec, Oc)) : Last
Weave()
}


I use it by first exporting the finished project using the Lagarith codec set to RGB at the native HD resolution. Then modify the script for the correct file name and open the AVISynth script in Virtual dub and set it to fast recompress, with the video compression set to Lagarith again, this gets out a new AVI resized using the script, that AVI file can then be dragged back to Vegas and then exported using the DVD templates.

If anyone has any test footage they would like me to try this with and post back up somewhere more than happy to. I found this virtually gets rid of interlacing artefacts and movement and camera pans are also clean, probably due to no de-interlacing. Not absolutely as sharp as some other methods, but then it compresses easier and looks great to watch.

Regards

Phil

Frans Meijer
May 12th, 2011, 11:47 AM
... without any de-interlacing required avoiding some nasty artefacts that can introduce sometimes.

I don't think you can resize without de-interlacing ... resizers typically expect pixels to be fairly evenly distributed and in interlaced material you miss every other line...

Have been using SD so far to avoid the problems with resizing interlaced video, but first tests with nnedi and lanczos look promising.

Phil Lee
May 12th, 2011, 12:06 PM
Hi

I thought the same. In the above example, it is separating out each field and resizing the field to create a new field at SD resolutions, and seems to be resizing in such away to avoid nasty resizing artefacts.

It really does work very well, perhaps not as sharp as some methods that have a de-interlacing step, but I always find pans across completed scenes becomes quite messy with a big drop in detail due to the de-interlacing, but this isn't the case with this method avoiding de-interlacing, and you are not messing around with the temporal detail as you are by interpolating up to 50 de-interlaced frames to resize and derive new fields from.

Regards

Phil