i will try it tomorrow when i have cam in hand but if find it redicualous that we have to set the cam up a certain way so our sd dvd's look good. we shouldnt have to adjust the cam. the problem here is that your saying its good enough, most people say, pretty good, okay, not bad, good enough, does anyone out there say great or better than originally shot sd? i let you know what i think, thnks

how was it downconverted do you know, im hoping someone has a recipe that "really" works.

update :

i just compressed basic default dvd 90 setting in compressor to make an m2v and burned it in toast as opposed to dvdsp and it looks awesome, im running more dvdsp tests, there might be something wrong with my dvdsp, hopefully. this makes me happy, ill keep you posted

HD to SD PAL Liquid or Fusion
 

Screens from Liquid 7.2 and EYEON Fusion HD to SD (PAL)
In Liquid usable unsharp mask also
Render to Uncompress and export to m2v
Most important thing - selection resize filter and matrix for filtration

http://www.dvcampro.ru/video/LiquidD...vert2_copy.jpg
http://www.dvcampro.ru/video/LiquidDownconvert_copy.jpg
http://www.dvcampro.ru/video/DFdownconvert_copy.jpg

man i wish i could implement this but im on fcp on mac, the settings are nothing like this.

I wonder why this topic has not come up before?

I am reading a lot about this on the Canopus forums as well.

HDV cameras have been out for a few years now and the general concensus was that SD DVDs looked better shot on HD sources.

Why the change?

The change is become someone 'noticed', and then others looked for it saw it too. It's apparent in most HD (or HDV) footage downressed to SD at some point, with some shots.

This is a problem only if you make it a problem.

This is one of those things no one saw before because they weren't looking for it. Like, vignetting wasn't noticed till it was pointed out, or Chromatic Aberration, or 1440 vs 1920 resolution, or going back far enough, 4:2:0 vs 4:2:2. Now that people can afford near 1000 lines of resolution, now they're seeing something the pro-pros have seen for years.

I'm surprised that the EX1's holy grail of resolution seems to be being treated as a fault!

This all said, shooting interlaced with a camera with this kind of resolution and then wishing to output SD is asking for trouble, IMO. I'd shoot progressive.

Grigory,

I too am getting very good results using Liquid 7.2. In fact I didn't even know there was an issue with downconvert until I read this thread. Liquid is a very nice editor - a shame that it is poorly supported by Avid.

Also shooting 1920x1080 -- not EX1 -- and the key to avoiding aliasing (and line-twitter) is to filter the HD to match the bandwidth of SD.

Alas, each NLE has a different set of filters that could be used. And, there are dozens of math that can be used to implement any single filter.

With Apple products you have access to a FILTER in the QT dialog box. I use BLUR = 2 or = 3, and it works. But, this is a crude way of softening edges.

Adding to the problem -- interlace inherently has line twitter. And, Sony in the past has under low-pass filtered it's CCD output so as to increase rez. measures. This, of course, allows aliasing to appear in HD recordings. So you can see aliasing in HD. And, it gets passed down to SD. Which adds it own. (Then one plays back SD DVDs on an HDTV it can look bad.)

The reason high-end HD cameras cost so much is because they have far more sophisticated amplifiers and filters. So, one expects them to record a very clean image.

He's using Unsharp Mask. Of course that'll fix it and the cost of blurring the lines. You can certainly duplicate your sequence and apply a very slight blur in one direction. Compressor doesn't seem to have an equivalent filter.

The concern people have is that they DO NOT want to lose resolution to solve the issue which I'm not sure Grigory's workflow is really a fix. It seem more of a band aid that we are trying to avoid.

okay im starting to get the down convert to look acceptable on tube tvs (not as good as my sony dsr390 sd dvcam) but on lcd hdtv, especially the bigger ones, man sd still looks like crap, tube sd blows the doors of sd on hdtv, am i missing something or are we just screwed till all our clients have bluray players?

my dsr390 on lcd looks better than my ex1 or fx1 downconvrt sd on lcd

these are the screens ive tested.

sony bravia 40" 1080p HDTV
Westinghouse 32" 720p HDTV
LG 20" 720p HDTV

the picture on all tests looks the best on the LG ($400 tv) - the bravia surprisingly looked the worse (and yes i have the settings on the dvd player and tv set optimal - actually spent 8 hrs tweaking settings to make sure).
i guess the big tvs will never play sd well (except for hollywood movies). does anyone have 30 second m2v's they could link (if happy with your encodes on lcd) so i can burn and see for myself?

I'm uploading some files to a web server for people to play around with.

One is a .m2t file, set up as a 23.976P 16x9 SD MPEG2 render from original 1920x1080 footage. I'm also including a .jpg screen grab showing the settings I used in Premiere Pro. Another is an H.264 Blu-Ray file @ 1920x1080, so you can see how the original HD looks.

The third is the same footage, except instead of being rendered in Premiere Pro, I took into After Effects, scaled it, and then created my MPEG file. I also experimented with After Effects color space conversions, but didn't notice a difference using that. This render has no effects applied, just scaling.

The footage from multiple sources: After Effects compositions (created from a client-supplied PDF file with vector graphics), EX1 video, and even two 8MP stills shot on a Canon Digital Rebel XT. The reason I sent you mixed footage is so you can see the issues of resolution and scaling aren't EX1 issues, they're issues inherent to the limited resolution of SD.

You'll see that the SD down-res looks like... well... SD. When you're taking a 720x480px image and spreading it across 42" or more of LCD screen, it's not going to look like HD. My clients have been happy with the output of our EX1 so far.

Just for comparison's sake, the results I'm getting from my MPEG2 renders are almost identical to results achieved using the hardware scaler built into my Matrxo Axio system.

After Effects scaling does seem to create better results (marginally) than Adobe Media Encoder. I'm not sure if the additional hassle and rendering time is worth it for most projects.

I haven't shot a whole lot of stuff interlaced. Maybe someone else can upload some of that.

Footage can be downloaded from:

http://www.actvonline.com/HD_SD_Test_Files.zip

I don't know how long I can leave the footage up. If it starts getting a lot of hits and eating into our bandwidth (the file is 214MB), I'll have to take it down.

Please let everyone know how these compare to your own renders. Any chance you can post some files from your workflow for people to look at?

@Jon
I think the AfterEffects Resize looks much worse than the one made with your hardware.

But the hardware resize has the wrong aspect ratio and doesn't use the full resolution.
It would look even better if it has the right A/R.

The HD-source (which I had to remux to an avi to work with) looks stunning :)



Edit: compared them again.
The hardware encoded one has the right A/R (but black boarder on the sides) and has many artifacts...

regards Dennis

@Jon:

I did a convertion myself.
It looks incredible good!

I used Virtualdubs' smart resizer with avisynth and Hcenc for M2V encoding.


I will upload the resulting file (with proper pulldown) in a few minutes.
If anyone is interested then I will do a howto.


regards Dennis

I'm curious to see how it compares to Premiere Pro or After Effects.

Looking forward to it!