Hi,
I have read a lot of threads (and posted a few) trying to resolve my confusion regarding how to get the best out of HDV...

I am filming a wedding this weekend with 2 FX1's which up to now I have only used in SD mode.

I am totally clueless about how to get the most out of HDV footage when burning to an SD DVD. (I wont even begin to work out how to burn HD DVD's yet...)

This is what I do know:

I can capture to mpeg using pp2.0 in HD
I can capture to m2t using HDVSplit

I can edit in PP2.0 using either of these

Then the tricky part starts... should I export as mpeg2 at full HD res using the adobe encoder? Would that be accepted straight into TMPGEnc DVD Author and would it downconvert the res to 720x576 for the final DVD...

I have been advised to author/burn straight from an edited m2t file but I dont know how to create this.. could i export the edited timeline to tape then capture using HDV split to create an m2t?

I am also a regular user of Mainconcept MPEG Encoder which Ive just spotted has some HD DVD presets at higher res, but does not accept m2t as an input type.

Basically as you can tell I am a total newbie to HDV and would appreciate a simple guide to how you get the most out of your HD footage when burning to DVD...!

First of all, my best DVD results actually come from outside the Premiere. While I have Premiere Pro 2, I find I have better success using doing any encoding inside of DVD Architect, which came with Vegas 7. I also use Cineforms NeoHDV to capture to an intermediate .avi format for editing. The assumption here is that you don't have access to those, so before I got Vegas, I used the following process in Premiere:

1. Captured using HDVSplit or the Premiere Capture Utility. I preferred HDVSplit, because it scened detected.

2. Edit project in HDV.

3. Render full edit to a deinterlaced Widescreen DV file.

4. Pull that into any DVD program you have to render to the DVD image. I have used Pinnacles Studio 10 for DVD production as well as DVD Architect.

Why render to DV Wide and not go straight to mpeg for DVD ? I don't know why, but for some reason, when I was using that process with Premiere, the results were inferior !

IMO using DV codec as an in-between is a horrible choice. It is highly compressed and has a non favorable chroma sampling (4:1:1) that will rob your footage. The only time ever to use DV is when that is the format that comes out of the cam. HUFFYUV would be a far better choice.

IMO using DV codec as an in-between is a horrible choice. It is highly compressed and has a non favorable chroma sampling (4:1:1) that will rob your footage. The only time ever to use DV is when that is the format that comes out of the cam. HUFFYUV would be a far better choice.

That's the point. I don't understand why Premiere transcode from HDV to SD mpeg comes out worse, but it sure seems to, in my experience. Are there other better ways, absolutely.... as I indicated....

thanks guys,

first off, Ken, what is HUFFYUV?

Also Chris, I agree with HDV split (in fact I am capturing the project using that right now which saves m2t files which premiere can edit with).

Although others say its not the best way you say you have had good results rendering out to a deinterlaced widescreen DV file. By this do you mean just do, file-export move without using the adobe encoder? As I know in SD comps this exports an avi... Would this still be at the HD resolution?

I use TMPGEnc DVD Author to edit, but I think that needs to take in mpeg files for DVDs.. I Usually convert using Mainconcept mpeg encoder but although that has a HD preset for output I am always convinced that it heavily compresses the file and loses a lot of quality.

In short, I am with you with the capturing and editing side of it, but lost when it comes to the export and authoring.

Your continued help would be much appreciated.

David:

Again, this is just my experience working inside Premiere. Yes, I would transcode or render to the deinterlaced DV wide screen standard setting. Yes this is SD. Maybe I misunderstood your original question, but yes this results in an SD resolution image. Thats all you will ever have if you are producing a standard DVD and in process I used, thats where I converted to SD, and that is maybe why I had better results, ie., Premiere converted well HDV to SD DV, and well from SD DV to SD mpeg. Although the project was shot in HDV, if it is going to standard DVD, it is going to be SD output.

Chris,

I see.. I was under the impression that I could create a normal dvd on a single layer disc that uses full HD resolution, I just would not benefit from the same quality as if it was burned as a HD DVD.

In your experience will I see be able to see the superior quality of this HD to SD DVD, when compared to SD footage on an SD DVD?

SD is SD. I haven't stepped into Blue Ray or HD production. Any HD I do is viewed on my cheap HD monitor, through my DVI out cable on my system. It is obvious even there, though, that there is a tremendous difference between DV SD and HD finished files

My experience is anecdotal, of course. I can't prove it by showing scientific numbers, and tests. Others have attempted to do so shooting charts and tests. In the end what you have is a DV image. We all know that the DV image out of a low end consumer cam is not as good as out of a GL 2 or VX2100-- and I think the experiece is the same here. HD and HDV are advances in imaging quality, or else, why would we be doing it.

But the question remains with the FX1, if my output is DV, should we edit in HDV, and render to DV. I think so. That may be because this camera has native 16:9 output, while the older cameras had to do some electronic manipulations to lay down a 16:9 image.

I have never shot anything other than HDV with my Sony FX1. I determined early on that getting the highest definition on tape is important, and then I would have the choice to capture HDV or SD. In the end, I felt that although I was going out to SD on a project, I still got a better result rendering to DV after editing in HDV- and now, that I have it, capturing with NeoHDV, and editing to whatever output I desire. Certainly, it put more strain on the editing system, but in the end, it just seem a bit deeper in color and feel.

I have, in rush situations, captured in DV, and certainly, the end result, which I mixed with footage from my VX2000, was still an improvement. But I have concluded that the best result come from the HDV edit.

Thanks a lot Chris. This weekend I have a shoot in DV using the FX1 so i can compare footage from last weeks HDV footage.

So what I intend to do is export 720x576 widescreen avi and convert using mainconcept mpeg encoder as I usually do. The only issue with this is that the file size of the entire DVD is below half the disc capacity. I'm not sure if there is any software out there that will render to higher quality in order to fill the space available?

Mainconcept even has HD Mpeg output options so I will play around this and see if it burns/plays back on a normal single layer disc. Then it seems a bit strange that I'd go from HDV - SD - HDV...

If only there was a simple answer......!

Hi all

Like Chris I have always shot HDV on my Z1. This I import into PPro 2.0 for editing..
For a long time I just used the camera to downrez.. as I was not at all happy with the Premiere down convertion.

More recently I have got hold of After Effects 7.0. I have lately been importing and editing HDV in PPro2 and then importing an uncompressed HDV avi file into AE.

This is easy to downscale using After Effects and I output a SD avi which can be encoded to DVD (mpeg2) in either Premiere, Tmpgenc etc... I have found this give a sharper overall result, as the footage goes from HDV to SD uncompressed avi as opposed to the normal DV codec...

I must say though that once you play back on your average TV set the difference becomes unnoticeable.

cheers
Gareth

My understanding of DVD mpeg is that once you reach the maximum bitrate it operates under, you cannot push it more because it will not play on standard DVD players. In my experience, I can force up to two hours on a single layer disk, getting decent though reduced SD resolution for a standard TV, and at the highest conforming bit rates I can only get around 60 minutes. So with your shorter than one hour disks, you will have some unused space..

first off, Ken, what is HUFFYUV?
HuffYUV is a lossless codec. You can download it here:

http://neuron2.net/www.math.berkeley.edu/benrg/huffyuv.html.

In general, a huffYUV file will be about 1/2 the size of the equivalent uncompressed *.avi file. The advantage to using it is that there is no loss in image quality due to compression.

In PAL land, where DVD and the DV codec are both 720x576 4:2:0, you may not see a huge difference between using an intermediate HuffYUV and DV version, since you're only introducing 1 generation of DV DCT compression.

In NTSC land, DVD is 4:2:0 and DV is 4:1:1. This means when you create a DVD from the DV file, you end up with essentially a 4:1:0 colourspace. Not only would you get extra DCT compression, but also the colourspace hit. If your intermediate was HuffYUV, rather than feeding a 4:1:1 file to the compressor, you'd be feeding a 4:4:4 file (from the HDV downconvert), which has significantly more information in it.

-Steve

Steven:

Thanks for that. Thats my tech lesson for the day. I'll give that a try. Do work flow would be:

HDV edit --> huffYUV --- to DVD mpeg. Can you just import the huffyuv file into a DVD program like DVD Architect for instance, and transcode without a hitch ?

Transcoding should be hitchless provided the codec is installed properly. Obviously you'd have to test the application to find out (preferably with a small easy to export file first). If the HuffYUV file doesn't work, but you've got tons of hard drive space, you could also export to uncompressed.

-Steve

David, as someone who travelled the same path, I perfectly understand your frustration. You would think that Premiere should be your one-stop solution, but unfortunately it's not. I understand that some of the issues have been fixed at least to some degree in PPRO 3 but we both have 2. After a lot of testing this is my way of doing things, you may want to give it a try.

1. Capture from camera using HDV Split.
2. Import and edit, color grade, fix audio, etc. natively in Premiere Pro 2.0.
3. Export as MPEG2 at the highest quality.
3. Resize to uncompressed 720x480 using VirtualDub MPEG2 (free) and frame serve to TMPG Enc, or Cinemacraft for MPEG2/DVD encoding (Canopus Procoder may also work with frame servers, I'm not sure). Settings in Vdub are: method Lanczos 3 and interlaced. In your encoding software you need to specify that your source footage is upper field first, but your output has to be lower field first and 16x9 wide screen.
4. Import the resulting file or files (a/v interleaved/program stream or elementary a + v streams) into your preferred DVD authoring software (I use Adobe Encore) for DVD creation.

You may want to export to a file instead of frame serving (then ecode from the file instead of using the frame server), but count on very large file sizes in excess of 100 GB per hour. This has the advantage of having the file handy in case you need/want to export to another format. If you choose this, I suggest breaking up your project in smaller pieces (say 15-20 minutes).

The extra step comes with a huge reward in terms of sharpness and rich color; from what little I understand of colors space, resizing HDV to uncompressed 720x480 comes with close to 4:4:4 color space so the MPEG2 encoder has a lot of chroma data to extract from.

I hope this helps - the best way is always your way, so give it a try and see if it works for you. Come back with questions if you need to.

In PAL land, where DVD and the DV codec are both 720x576 4:2:0, you may not see a huge difference between using an intermediate HuffYUV and DV version, since you're only introducing 1 generation of DV DCT compression.
-Steve

From my understanding DV PAL 4:2:0 and DVD Mpeg2 4:2:0 despite the numbers do not match. I am not an expert on this but I have been told there is no advantage to PAL's 4:2:0 in this situation.

1. Capture from camera using HDV Split.
2. Import and edit, color grade, fix audio, etc. natively in Premiere Pro 2.0.
3. Export as MPEG2 at the highest quality.
3. Resize to uncompressed 720x480 using VirtualDub MPEG2 (free) and frame serve to TMPG Enc, or Cinemacraft for MPEG2/DVD encoding (Canopus Procoder may also work with frame servers, I'm not sure). Settings in Vdub are: method Lanczos 3 and interlaced. In your encoding software you need to specify that your source footage is upper field first, but your output has to be lower field first and 16x9 wide screen.
4. Import the resulting file or files (a/v interleaved/program stream or elementary a + v streams) into your preferred DVD authoring software (I use Adobe Encore) for DVD creation.

Ervin, thanks for this. I am not as up on the jargon as you and most of the people here so here are my questions..

RE: step 3 - Do i use the adobe media encoder to do this, if so do I need to up the resolution to the HD res or is that available as a preset?

RE: step 3(the second one!) - "Frameserve to TMPG Enc".. I know TMPG Enc but have not used frameserve, assuming thats a program. Could you explain what this means.

Once I have done the frameserve tmpgenc thing will i be left with an avi or an mpeg?

Many thanks again

Yes, you use media encoder, there is no preset, just select generic mpeg2 1440x1080 upper field first, interlaced, and move the quality slider to max. Our friend "down and under" posted a nice tutorial on this a while ago: see http://www.esnips.com/web/Premiere-Pro-2-Video-Tutes.

Frame serving will not leave a real video file behind - that's exactly the reason it's used for. Frame serving provides a fake file that is seen as a real file by the encoding app - if you need a real file to be used in other applications, then you will have to output to a real file, but will need to have enough hard drive space for it.

no i use TMPGEnc DVD Author and I think I'm right in saying the newer versions accept m2t files so I will give your method a try. Many thanks

Ervin,

I have exported using the full res mpeg setting (it exported an m2v file) which I have imported into Virtualdub Mpeg2. I am having trouble seeing where to resize the video in virtualdub as ive never used it before.

Do I have to go to the compression setting under the video menu and choose a specific codec?

Do you know of any similar guides for completing the virtualdub step of this process so I can try and stop bugging you so much!

Thanks :)

David

In Video > Comression set compression to uncompressed (it's the default).

Under Video > Filters go to Add and select Resize. Set to 720x576 for PAL (720x480 for NTSC), Filter mode Lanczos 3 and tick the Interlaced box. OK to save and F7 to start the resizing/decompressing.

Leave everything else at default.

Just keep bugging me, I love to help!

Ok Ervin, if you like it then I will no doubt keep them coming..

next issue.. aspect ratio of the converted file. The original is 16:9 and the one I have output is 4:3. I set the frame size to 720x576 but if i remember right there is no difference in frame resolution between 16:9 and 4:3...

Where in the program to I specify a 16:9 output?

Expect more questions once this one is answered!

Thanks

Correct, both 16x9 and 4x3 are 720x756.

You specify the aspect ratio in the MPEG2 encoding software.

Keep'em coming!

ah i see.. and there'll be no quality loss by exporting in virtualdub as 4:3 and then converting to wide when I burn the DVD?

So now I have the exported avi file do I import this to the dvd authoring software and let it encode or am I missin ga step?

It says in your guide I now have to frame serve to TMPGEnc... I wont lie to you... that means nothing to me...?!

Thankyou kind sir.

You won't loose a thing - the final MPEG2-DVD file will be still 720x576, but it will have information embedded to "tell" the DVD player "this is a 16x9 video".

Now that you have the uncompressed AVI file - feed that to your favorite encoder or directly to the DVD authoring software. I prefer to go in two steps, encode with TMPG Enc and then author the DVD in Adobe Encore as Encore does a poor job at encoding.

If you already have the actual uncompressed AVI file, you skipped the frame serving part; that was INSTEAD of exporting to an actual file, in order to save hard disk space.

Ervin,

Yes ive noticed that the file is very large (1.4gb per minute..) My videos should be 70 mins max so i will be able to handle it.

I do not have my authoring software with me at work but I have 2 avi files, one created using mpeg full res export and virtualdub 4:3 conversion, and one where i simply imported the m2t file into a 16:9 SD project and scaled down to 54%.

If the authoring software allows it i will burn both to one DVD and compare on my tv.. looking forward to seeing the results.

What makes you want to use TMPGEnc for the encoding specifically? I used to use mainconcept mpeg encoder to encode. I'm not sure if you are familiar with that piece of software but would that do the same as encoding with TMPGEnc? In which case I can convert them both to mpegs now at work and burn and check a disc. I am always worried that somewhere deep in the mainconcept settings it is compressing the mpeg too much and possibly losing quality to save disc space.. which with a 70 mins dvd is not generally necessary.

Thanks

I will also add that with PP2, the downrezzing of HD to SD for mpeg video is just plain busted - the quality is very poor. PP3 corrects a number of issues that PP2 has, and the quality of PP3 for MPEG video is significantly better. You will easily be able to tell the difference

Brent,

thanks for that .So if I had PP3 what would the steps be? Import the HD footage into an SD timeline and down scale? Or edit in a HD timeline and just output to SD using the imporved media encoding?

David,

If your PC can handle the processing, I'd pull it into an HD timeline, edit it in HD, then you can export the final edit as either HD or SD. that's what I usually do.

btw - to clarify, with PP2, I found that if I exported mpeg-2 as progressive, then the quality was preserved. However, exporting from HD to interlaced (where you need it for a DVD) was where the quality took a big hit - this is what was fixed in PP3.

Brent

What makes you want to use TMPGEnc for the encoding specifically?
If Mainconcept works fine for you in an all SD environment, then it might be OK for this process as well. TMPGEnc is incredibly slow but by far the best encoder in the $100-200 range. Of course, a $2000 Cinemacraft that can perform 10 passes will blow away any other encoder...

(Sorry for the delay, I was away a couple of days).

ok so the time has come to actually start exporting these HDV timelines and as ive been told in this thread as well, everyone seems to say the best method is to:

"HDV Clip import into Virtualdub, resize from HD->SD via Resize filter (method Lanczos3, Use Interlaced), export via frameserving or uncompressed temp file to Tmpgenc."

I have edited and exported the HD res mpeg, imported into VDub and resized, but I do not understand for the life of me what this frameserving process consists of. When I just saved the file from VDub as an AVI as suggested somewhere in this thread I was not too happy with the resulting quality of the DVD. It just looked SD to me. Could someone explain if this frameserving process is easy to do from VDub to TMPGEnc and if the software is free?

I'd be very greatful.

Thanks

The frame server is built into VDub and fairly easy to use; you have to first start the process by going into the auxiliary setup (auxsetup.exe) and clicking on "Install handler", then starting the actual software and choosing "Start frame server".

The resulting DVD will look SD because it is... well... SD. You have just thrown away 3/4 of you video by resizing HDV to DV. The method described above is sort of an interim solution while we already have HDV aquisition options but mostly standard definition distribution channels. You will have to go to BlueRay or HD DVD viewed on a 1920x1080 television set to get the full resolution of HDV.

Ervin,
Sorry i phrased that wrong. I'm aware the finished product is only ever going to be 720x576 but I meant it looked no better than it had of if i'd have filmed in SD mode on the same camera in the first place.. therefore making all the hard work after the editing process pointless...

Although I havent started frameserving or anything yet so I'm going to work on that today (expect some more posts unfortunately..)

Can I just confirm a query with aspect ratios. Is this correct:

1) edit in 16:9
2) export in 16:9
3) Import into VDUB and resize which outputs (via frameserve) in 4:3
4) Convert to mpeg using tmpgenc or mainconcept in 4:3
5) Tell dvd authoring software (I use TMPGEnc DVD Author) to convert the 4:3 mpeg to 16:9...

I had an issue when creating some test discs that one video seemed too squashed up so I had obviously tried to 16:9 it twice or selected letterbox somewhere down the line however this has only happened once and was clearly something I did wrong.

For now I'm exporting everything as HD Mpeg as well as SD Avi, will have to start converting authoring and burning in the next couple of days.

Thanks again

David

There is no such thing as 16x9 in HDV; in other words you don't have to set that because HDV is 16x9 ONLY. In other words, when you start an HDV project in PremPro, you don't even have the option to set aspect ratio. Although 1440x1080 is 4x3, PremPro makes the correction for display purposes only, so you will see a 16x9 image while the footage itself stays 4x3. In fact, the footage stays 4x3 all the way, only in your DVD authoring software you have to specify you want a 16x9 DVD as final product. I know it sounds confusing, but once you start actually doing it, you will understand.

The frame server does not make a difference in terms of quality - it only helps you with hard drive space as you don't have to store a huge uncompressed file that you don't need after the MPEG2 file is created.

5). You don't actually "convert" the footage to 16x9 - you simply flag it as 16x9 so the DVD player will interpret it as 16x9.

Thanks Ervin, I have worked out how to use frameserver from virtualdub to TMPGEnc and its running now... I'm sure I selected 16:9 somewhere in TMPGEnc tho so I might find that the footage is too squashed again. Am I right in saying that it is in TMPGEnc that I alter the field options? When I first add the clip it is by default set to lower field first but I change that to upper. and then in the advanced options for the output file I change from upper to lower fields first? This is after I export the mpeg2 at highest quality without changing the default field options settings.

Also when using TMPGEnc on the first wizard start up screen I selected PAL DVD 4:3.. is this wrong? I have only ever used tmpgenc to create vcd's years ago.

Thanks again
David

I am not familiar with TMPG DVD authoring, I only use it for the encoding part. You have to set both properties for the input/source clip and the output/resulting clip.

I think you should set it to 16x9 from the beginning but I also think (not sure though) you can change that when actually authoring the DVD; I know I can do that in Adobe Encore. I would do a short test clip and confirm.

sorry when i said the first wizard screen of tmpgenc i did mean the encoding software not the dvd authoring software.

Ervin... i think we've cracked it. Managed to get frameserving right. Cant remember exactly what I did with aspects but it seemed fine when I watched it. Footage is obviously not near to HD but its noticeably better than it would have been had I shot in SD...

Thankyou very much for all your help with this.

David

I disagree with using HDVsplit to scene detect. Premiere Pro starts to act weirdly (consuming huge amounts of memory and opening hdv-projects slowly) when it has lots of hdv-files in the timeline.

HDVsplit does synch audio way better than Premiere does in case of dropouts though.

yeah I don't like to use scene detect either.

Ervin if you havent deserted this thread I have another question after playing around with this process quite a lot..

With regards to field orders you said that when using my encoding software (TMPGEnc) I should specify that the input field order is upper field first and the output should be lower, however when I do this and burn the DVD also with lower fields first I get horrible jerky playback.

When leaving the TMPGEnc input and output to top fields first but selecting bottom fields first in the authoring software the playback is fine except for the first second or so when it is the same as the first jerky result. Almost as if it takes a second for the field ordering to kick in upon playback.

However, when I leave everything as top fields (TMPGEnc input and output, as well as authoring software) the playback is fine. However after what you said I am worried that I am somehow losing quality by doing this.

The only combination fo settings I have not used is input upper fields first and output lower in encoding, then author the dvd with upper fields first. But that would seem pointless to me going from upper to lower and back to upper?

If you get time could you give me a brief explanation of what you thought I should change from upper to lower when encoding as this seems to cause problems when burning. Remember I am using PAL video if this makes any difference.


Thanks again
David

I am really not sure what is going on in your case but I can tell you that until I sorted it all out I also got the jerky/jumpy images. Shame on me as I am originally from Europe, but I only "think" PAL is also lower field first... I'm not sure. So the method I described should be the same for you, can you confirm that PAL is lower first?

Also, I don't quite understand where in the DVD authoring process you have to set field order, unless you encode your footage in the DVD authoring program itself. The way I told you, you should already have the video files encoded to MPEG2 when importing them into the authoring software, so a new encode is not necessary.