HDV --> SD DVD Workflow (DebugMode FrameServer, AviSynth)

[Dan Isaacs]

While the old thread contains a wealth of great information, I've decided to break away from it. It's too old, too long, too confusing... In summary:

Many users are aware that Premiere/AME/Encore does a "less than stellar" job of converting 1080i HDV to interlaced SD DVD. I (and others) have tried every conceivable combination of options and found none that provided "professional quality" conversions. This is largely due to the way that Premiere handles scaling of interlaced material ... Example #1 ... Example #2 ... and Premiere's failure to convert from the HDV color matrix (Rec.709) to the SD specification (Rec.601).

In an effort to solve these problems, I developed a workflow (using several third-party freeware tools) that achieves results far superior to that of Premiere and/or Encore by themselves.

I have written a guide for this workflow, including step-by-step instructions and links to download all the required tools.

This page also contains a link to a more advanced option for doing these conversions (using mostly the same tools). Please do not attempt the advanced version unless you have already implemented my "basic" workflow!

The page and linked files will be updated frequently. It is a work in progress, but should already provide excellent quality. I look forward to comments and suggestions from this community (as always).

Enjoy! -- Dan

[Jeff Bellune]

Pr/AME with the GPU-accelerated MPE scales much better than CS4.  Software-only MPE is better, too.  I still keep the hd2sd tools around just in case.

Check out my tutorials for general CS5 exporting, hd2sd, and CS4 HD to SD here:

BDVS Tutorials

-Jeff

[Huntrex]

I have been experimenting with the scaling in Premiere Pro CS5. I encoded the same file (DVCPROHD 1080i 60i) in both CS5 and CS4 (fully updated) going from HD to SD using the exact same settings. After viewing them side by side in After Effects, I came to the conclusion that CS4 actually has higher scaling quality than CS5. The CS5 version was noticeably softer.

Has anyone else compared, side by side, the scaling quality differences between CS5 and CS4?

[Jeff Bellune]

Has anyone else compared, side by side, the scaling quality differences between CS5 and CS4?

Yes.  I got the exact opposite results that you did.  Using a burned DVD-RW disc in a hardware DVD player connected to 3 different types of TVs: LCD HDTV, CRT SDTV, DLP HDTV.

-Jeff

[arlessio]

hi jeff, my name is Alessio from italy

i have a project premiere cs4 with a matrox rtx2 HD (1280x720 25p) and i must export a file matrox avi mpeg2-I-frame 100 mbt (1280x720 25p)  for make a dvd with a hd-sd conversion dan isaac method. please, can you give me a script for my avisynth for my progressive file 25 ftp? i use hc compress for encoding.          thank for all

[Jeff Bellune]

Hi Alessio.  Welcome to the forum.

Follow the steps in my TUTORIAL using this script to be loaded into HCEnc:

SetMTMode(2,0)

MPEG2Source("Alessio's720p25clip.d2v")

hd2sd(interlaced=false, OutputColorSpace="YV12")

-Jeff

[arlessio]

hi jeff, thank for your script

sorry, but i think that my workflow to converte my file matrox.avi is

differente. i use dan isaacs plug in "avisynth 2.5, right? i don't

understand what is "SetMTMode(2,0)"

besides, my file Matrox avi is not MPEG Source (matrox's mpeg2-I-frame 100

mbt had a extension .avi, i think

if you can, look mi worxflow:

- to timeline premiere 1280x720 25 p (matrox project) i export my Matrox

avi 720 25p

- my avisynth' script is hd2sd ("my Matrox avi 720 25p.avi,

interlaced=false, OutputColorSpace="YV12")

- open virtualdub and export "lags_sd_yv12"

- with my "lags_sd_yv12" i make new avisynth's script, hc_source.avs

("lags_sd_yv12.avi") and export my mpeg2 file for dvd

it may be that my workflow is old?

sorry for my questions, tyhank for all bye alessio

[Jeff Bellune]

Did you watch my tutorial?

-Jeff

[Guest]

I am using Premiere Pro CS4 and have exported my video as MPEG2 I-frame, using Dan's presets.

I then created my script:

hd2sd("Sequence 02.m2v", OutputColorSpace="YUY2", OutputBFF=true)

Each time I tried to open this in Virtual Dub it would freeze up ("Program Not Responding")

So I did a little more research and tried this script:

MPEG2Source("Sequence 02.m2v")
hd2sd("Sequence 02.m2v", OutputColorSpace="YUY2", OutputBFF=true)

Now I get an error that reads:

Avisynth open failure:

Script error: there is no function named "MPEG2Source"

I have saved the script in the same folder as my video (per Jeff's great tutorial). But I just can't seem to make it work. My script is obviously wrong, but I don't know what else to do. How should my script read? Is there something else I'm over looking?

I am fairly new to the video world and all this is a little overwhelming. Any help would be great. Thanks.

[Jeff Bellune]

1. How long did you wait before deciding that VirtualDub had hung?  hd2sd is CPU intensive so it may take a while for it to show up in VirtualDub.
2. MPEG2Source can only be used with .d2v files and the DGIndex AviSynth plug-in.

-Jeff

[Guest]

Thanks for the quick response!

After about 10-15 seconds it freezes ("Program Not Responding"). After that I would give it at least 2-3 minutes before calling in the task manager.

I was also looking at the VirtualDub Frequently Asked Questions page and found the following:

Can VirtualDub read QuickTime, MPEG-2, RealMedia, or ASF files?

No, and installing codecs won't help.

Is my MPEG2 I-frame (.m2v) not an MPEG-2 file? Can VirtualDub even read my file?

How should my script read?

Thanks.

[Guest]

I was going back throught your tutorial and found that I needed to download the DGMPGDec.

Anytime I try to extract (w/ 7-zip) to any folder in my program file (x86) folder I get the following error:

cannot open output file C:\Program Files (x86)\AviSynth 2.5\plugins\DGDecode.dll

I got the same error message when I was trying to extract Dan's hd2sd_sd2hd folder. So I just extracted them to my desktop and copied them into the AviSynth plugins folder.

[Jeff Bellune]

Your .m2v file is MPEG2, but the output of your AviSynth script is seen by VirtualDub as an AVI file.

Let us know how the script works using MPEG2Source and the .d2v file from DGIndex.

-Jeff

[Huntrex]

Update to my first post:

After several more downscaling tests in Premiere Pro CS5, I found that the quality equaled or exceeded that of CS4. Plus the "Use Maximum Render Quality" option is much faster.

[lin2m3]

Jeff, I just want to say that using you tutorial and taking step by step instructions I was able to wake up my 8 cores to run at 80-90%. I tried that before using only the Dan's script but I achieved only 8% core load and 9,4 fps. After implementing your script which includes Dan's, I achieved 22-27fps, which is a huge improvement. The missing part from the chain was the DGindex. I'm on PAL land with 6Gb, I7930, Windows 7, 64

When exporting the file from VirtualDUB for Encore  what outputcolorspace should be used, RGB24 or YUY2? Dan said in the hd2sd help file it shoud be RGB24. You say in your tutorial it should be YUY2. Now, I don't know for what purpose you set it to YUY2. Maybe you can clarify a little bit. Thanks.

[Jeff Bellune]

Welcome to the forum.

I've used YUY2 Lagarith AVI files in my testing with Squeeze and the AME, and been happy with the results.  I've found YUY2 to process faster in VirtualDub than RGB using Lagarith as an output codec.  Try a clip with YUY2 and again with RGB24.  Use the one that looks best to you.

-Jeff

[Guest]

Jeff, after going back and revising my steps, I finally got through VirtualDub using the following script:

SetMTMode(2,0)
MPEG2Source("Bankson_Final_DGI.d2v")
hd2sd(OutputColorSpace="YUY2", OutputBFF=true)

However, my aspect ratio seems to be off (It's "squeezed' horizontally which makes everyting appear tall and skinny). I also have no audio.

I checked the details of the .avi that came from VirtualDub and it said it was 720x480 @ 23 frames per second. 720x480 is correct but i'm not sure about the frames per second?

As I said before, I am using Premiere Pro CS4 and have exported my video as MPEG2 I-frame, using Dan's presets for 1440x1080 @ 24p (which matches my source video). The exported video from AME looks great and sounds great. (aspect ratio looks fine)

The quality of the video from VirtualDub looks really great aside from the squeezed image.

Is there still something wrong with my script? I'm not real sure where the "squeeze" is happening in my process. Hope you can help. Thanks for all you've done so far.

[Jeff Bellune]

Technically, AVI files don't disclose the PAR to the viewing program.  So all you have to do is interpret the footage to the correct PAR when you import the AVI file into your editor or encoding/authoring program, and any distortions should vanish.

What program did you use to examine the output of VirtualDub?  If you used a 24p preset out of Pr, then the frame rate of the converted file should show 23.976 in VirtualDub, MediaInfo and GSpot.

-Jeff

[Guest]

Hallelujah. Hallelujah. Hallelujah. I have a GREAT looking DVD of my project. With a little perseverance and A LOT of help from Dan, Jeff and Jon Geddes over @ www.precomposed.com.

I found the follwing over @ Jon's blog which helped a bunch:

Virtual Dub does not export pixel aspect ratio information. In Encore, you must select your imported clip in the project panel, right click and go to interpret footage, then force it to conform to the proper pixel aspect ratio. I then added the audio in encore and vuala!

Also, the MT plugin for AviSynth sped yp the VirtualDub process a ton. I was seriously amazed at the quality of the finished product.

Thanks a lot Jeff for all your help. You are an amazing resource for all of us rookies.

[Kuro_M]

robertlevigarrett wrote:

Also, the MT plugin for AviSynth sped yp the VirtualDub process a ton. I was seriously amazed at the quality of the finished product.

Congratulations!

I still use Dan's workflow for each and every project. Sometimes I even use it, when I am not downsizing/deinterlacing but just want to add decent sharpening and noisereduction to the original video. I do not use the mpeg2-I-frame variation but use Lagarith lossless codec only. The size of the files are probably very similar but I like the easyness of this codec in avi-containers.

What it the MT plugin for AviSynth, please? Sounds like "multi threading" or similar. It is not part of Dan's original workflow, I think. Can you help, please?

[Guest]

I pulled this description from the AviSynth site:

"MT is a filter that splits a frame up into smaller fragments that are processed in individual threads, allowing full utilization of multiprocessor or hyperthread enabled computers."

It works great, my quad-core CPU was running at almost full capacity the whole process. It's awesome.

Jeff talks about it and how to install it in his great tutorial. You can find more info or download it here:

http://avisynth.org/mediawiki/MT

[Guest]

Hi guys, I have a pretty basic question, but I could not find the answer in the thread.

I was using CS3 and dan isaacs original basic workflow, and i was very happy with it.

Ive not got CS5, and i read that dans original basic workflow would not work.

is there a tutorial for getting the best HDto SD in cs5?

Jeff, does your tutorial for cs4 apply to cs5 (laggrith codec etc...)

Thanks in advance

James

[Guest]

I should say Jeff i watched your tutorial for CS5, I just wondered if exporting from CS5 gave you the best results when converting 1080i footage, or if it was better to use another method like using dan isaacs for cs3

cheers

James

[Jeff Bellune]

I like CS5 for scaling HD to SD, even 1080i.  Some footage, like titles, will still look better with Dan's hd2sd scripts.  Most footage will look as good using CS5 as hd2sd.

The solution is to take a few minutes of your footage and scale it both ways.  Use what looks best to your eye when it's displayed on your intended display device.

-Jeff

[Guest]

Does Dans workflow work with CS5?

Is there a tutorial for it?

Sorry for all the questions, it just ook me forever to get dans workflow to work using CS3, just lookin for a step by step guide to using dans workflow (or equivelant) with CS5

Cheers

James

[Dan Isaacs]

Wow! I have not been on this forum for ages. Too busy with other stuff these days (such as work, child-rearing and home-ownership).

I find your conclusions very interesting, Jeff. I have not played with CS5 yet (mostly because I still lack a CUDA card). Actually, I still od most of my work in CS3 (being so incredibly familiar with its shortcomings and the added features of CS4 providing little advantage to my usual workflow).

Perhaps you could fill me in a bit on CS5. What changes have they made with regard to deinterlacing and scaling?

I am somewhat incredulous about it looking as good on normal footage and yet worse on titles. Personally, I've always thought that one of the disadvantages of hd2sd() was that it has to work on "flattened" video, where Premiere would have the advantage of compositing after scaling and deinterlacing -- so that elements such as titling and graphics would be minimally processed.

I did have an email conversation with one of the software engineers wherein we were discussing the advantages/disadvantages of different scalers. He seemed to think that Spline36 (as used in hd2sd) was best for footage but overly sharp (to his taste) on titles and graphics. Is it possible that CS5 uses a sharp scaler (such as Spline36) on video and Bicubic on the titling? That's just a guess, but maybe Jeff (like me) prefers the sharper scaling on most graphic elements as well. (???) I have not kept up with the development of Premiere lately so I really am just guessing about this.

Jeff -- If you can, perhaps email me some links to download a few comparisons.