Need better quality than "DVD Best Quality 90"

I'm producing a DVD of short documentaries and in the past have used the compressor setting DVD Best Quality 90 to convert our HDV 1080/i60 footage into a SD DVD. However, we have found that when played on the large screen- the beautiful shot (previously HDV) material looks highly interlaced, soft, and the subtitles looked jagged.


Previously I have sent the HDV sequence directly to Compressor and set it at "DVD Best Quality 90." Is there a better workflow that I can use, or different compression settings in order to produce video quality that I can be proud of?

Thanks for your suggestions,
Megumi

Render to ProRes in FCP and in Compressor turn on Frame Controls and set resizing to "best".



www.strypesinpost.com

Hey Thanks for that Strypes. Unfortunately for my team the quality doesn't cut it.
What would a post house that specializes in DVD encoding do differently?

>What would a post house that specializes in DVD encoding do differently?

They may use better encoding engines like the stuff available from digigami or omni cinema craft, either hardware or software based. Not to mention, the down scaling would be done with an Alchemist or Teranex.


>However, we have found that when played on the large screen- the beautiful shot (previously
>HDV) material looks highly interlaced, soft, and the subtitles looked jagged.

But what kind of quality are you guys looking for? There's also that point about quality in, quality out, and quality doesn't happen if you shot on a very compressed format and encode to a very compressed end user format. If you are trying to go for good fidelity on large screen playback, you would do much better to encode to an HD format- HDCAM SR, or if you need a consumer HD playback device, BluRay.

Also, you need to de-interlace if you intend to screen this.



www.strypesinpost.com

Quote

we have found that when played on the large screen

Can you elaborate on this? What kind of screen are we talking about? SD material never looks as good as HD material when blown up to the same physical size, even under the best of circumstances. But if you're (for example) pumping that SD material through an off-res projector, it'll degrade more than if you were looking at it on a proper interlaced CRT. Same for watching SD on an HD LCD, though less severely.

Quote

looks highly interlaced, soft, and the subtitles looked jagged

I'm not sure what "highly" interlaced means; either it's interlaced or it's not, and in your case it is, because your original material was interlaced. DVDs can either be interlaced at 60i or progressive-scan at 24p; those are your only options. Since your material is 60i, your DVD will be interlaced.

As for the subtitles, are they burned-in text overlays, or are they generated by the DVD player from a subtitle track? DVD players are notorious for putting jagged subtitles up.

Quote

They may use better encoding engines

True, but there are pretty specific circumstances where that difference shows up. A better encoder will do a better job of handling situations where lower-end encoders will macroblock, but that doesn't sound like what's being described here exactly.

Quote

Not to mention, the down scaling would be done with an Alchemist or Teranex.

Yes, that's definitely true. But in my experience, an HD-to-SD conversion with Compressor's "best" settings doesn't look drastically different from an HD-to-SD real-time hardware conversion. Note that this is not true of SD-to-HD upconversion. But going from a higher-resolution format to a lower-resolution format is inherently easier in many respects than vice versa, so Compressor doesn't produce radically different results. It sounds like a drastic difference is what's being described here, but maybe I'm just misreading it.

Maybe I'm just missing something fundamental, but to me this sounds exactly like what you'd get out of any 60i HD-to-SD downconversion. The result will be obviously interlaced and soft. That's the nature of the medium.

>But in my experience, an HD-to-SD conversion with Compressor's "best" settings doesn't look
>drastically different from an HD-to-SD real-time hardware conversion.

I would think otherwise. Yes, down conversion is inherently easier than up conversion, in almost any regards. However, interlaced formats is always trickier to down convert than progressive formats, also between different software/hardware converters, you have the issue of moving between color/gamma spaces, as well as artifacts and aliasing that come about from different down conversion algorithms. And between 2 hardware based down converters (a Teranex and the HDCAM deck's in built down converter), I couldn't see the same results on the same calibrated SD broadcast monitor. I didn't do a back to back comparison with Compressor then, so I would base this theory on assumption.


For the OP, I think I initially missed the point that he's comparing the results based on playback on a large screen, and addressed the issue of down scaling. Yes, he needs to toggle the resize settings to best, as well as deinterlace the footage in frame controls, and see if he gets results which he deems satisfactory.

However, he may be unable to produce satisfactory results (depending on what he considers satisfactory), due to it firstly being a heavily compressed SD format, and secondly, that he didn't shoot on a very high resolution acquisition format to start with.

A higher end encoder and encoding path would result in less noise, less macro blocking, and less compression artifacts by optimizing GOP patterns and using more efficient filters, all of which Compressor is unable to do/customize. But I was assuming he's trying to target the best quality output on DVD, which was his initial question. So yes, deinterlacing it, and down scaling it through hardware converters, and encoding with the best encoders money can buy, will produce the best possible result on a DVD. The cheaper solution for this case is to go through Compressor.

On the other hand, a DVD (or HDV acquisition) will never produce satisfactory results required for a theatrical projection.



www.strypesinpost.com

I finally found someone who sees what I'm seeing!!!! This is EXACTLY the problem I have!!!

What I have been doing with my HDV1080i60 footage is to lay the edit back to HDV tape, then take the output of my Sony Z1U through firewire into a Panasonic set-top DVD recorder. I finalize the DVD and bring it back in the Mac. I use MPEG Streamclip to demux to M2V and AC3 files and then author with DVD Studio Pro.

I don't see it as much with video as I do with graphics...but when I encode HDV to DVD with Compressor, there are jaggies EVERYWHERE, but when I do my DVD recorder method...everything is sharp and clear.

I'm still trying to figure out what the problem is. I had a friend take identical footage into his computer and use the Compressor preset for DVD, just to see if I had some setting wrong somewhere, but it happened on his too....and you wouldn't see it as readily unless you've seen it side by side, then you'll see the big difference.

When I use the same footage captured and edited as ProRes, and then go to Compressor to encode the DVD, it looks great.

Once I tried converting my HDV edit to ProRes, and then exported to Compressor, but it still looked the same with the jaggies.

I've posted threads like this in the past, and both Strypes and Jeff have replied, and there has been lots of talk about how my DVD recorder method shouldn't (from a technical standpoint) look better than encoding with Compressor...as well as the benefits to why I should be using ProRes (it takes me twice as long to render stuff, though Jeff says it shouldn't...still not sure why that is), but as of now, it's been impractical for what I'm doing.

In an ideal world, I would like to be using ProRes for stuff, as well as encoding with Compressor, but these problems have forced me to use these unconventional methods to get the results I need (not to mention, on time and on budget).

I still need help with this one...at least I'm not the only one seeing this now.

Casey

With interlaced HD formats, you need frame controls set to "best" to do down scaling. I think there was a post a few months ago and someone mentioned they can't get interlace to downscale properly, and the workaround was to set resizing to "best", which does take substantially longer. Alternatively, you could try down scaling in After Effects or with Mpeg Streamclip as they are faster than Compressor for this purpose.



www.strypesinpost.com

I just tried this and it barely looked better than before.

The trouble comes whenever HDV is in the mix...I guess the ultimate solution would be to just never use HDV. You'd think it would work.

Maybe it works better with the new Final Cut Studio? I'm still on Studio 2.


Casey