Why does SD 16x9 (m2v) looks soft (bad) compared to ProRes422 LT

[Erwin Vanderhoydonks]

Thanks for reading this post and ik you can help me carifying this topic, thanks again...

Here is the problem...

I start from a ProRes 422 (LT or HQ) created with REDCine-X.

If I look at the ProRes422 HQ version with QuickTime or MPegStreamClip it look great. Even if I resize to SD 16x9 resolution 1024x576, it looks great.
But now comes the problem. After rendering to a m2v file with Compressor 3.5.3, it looks bad. Soft, blured, no details. Is this normal or is there something I can do...

Setting for ProRes 422 hQ --> SD 16x9 m2v.
- m2v SD DVD PAL Progressive 25 fps
- Quality 7,7 Mbps Best for motion estimation
- Frame control OFF
- No filters
- No geometry

Please can somebody help me...

 

[Jake Bastian]

Thanks for reading this post and ik you can help me carifying this topic, thanks again...

Here is the problem...

I start from a ProRes 422 (LT or HQ) created with REDCine-X.

If I look at the ProRes422 HQ version with QuickTime or MPegStreamClip it look great. Even if I resize to SD 16x9 resolution 1024x576, it looks great.
But now comes the problem. After rendering to a m2v file with Compressor 3.5.3, it looks bad. Soft, blured, no details. Is this normal or is there something I can do...

Setting for ProRes 422 hQ --> SD 16x9 m2v.
- m2v SD DVD PAL Progressive 25 fps
- Quality 7,7 Mbps Best for motion estimation
- Frame control OFF
- No filters
- No geometry

Please can somebody help me...

There should be an advanced option in there to specify data rate which sounds like it may be your problem. I think compressor defaults to variable but I'm not sure as I do this kind of thing with Episode (which IMO is much much better than Compressor which I've hated for a long time).

 

[Liam Hall]

How are you viewing? If you are viewing on your computer you might only be seeing 1 field instead of two, so your image will appear to have half the resolution hit really has.

 

[Alexander Ibrahim]

Uh ... SD Pal is like 720x576 stretched horizontally.

Each pixel is stretched horizontally from 12:11 to 16:11 aspect ratio.

You are making the mistake of using the "equivalent square pixel resolution" which isn't equivalent at all ... it just gives the right shape in square pixels with the minimum number of extra pixels.

The actual SD Pal 16:9 picture is 414720 pixels, while your preview is 589824 pixels. Your final SD output is 70.3% the resolution you are previewing at.

Yes you can increase the bitrate and some other tricks to improve the results ... but ultimately you are just throwing away a LOT of resolution.

It hurts to work in SD these days.

 

[Julian Banos]

Good point Alexander. Also, try to play in FILTERS with the aliasing settings.

 

[scott william]

DVD is just about the worst way to watch things these days. Even youtube can offer better compression. Some people still require SD DVD however and it really hurts to see your projects compressed to that.

 

[Erwin Vanderhoydonks]

Thanks for all the feedback.
I'm viewing it on my Mac with Quicktime/MPegStreamclip/... at actual pixel size and it look terrible.
I know, DVD is not the best media, but sometimes you have to do it. Many people are still asking for a DVD.
I'll some more testing and post my findings...

Alexander : Thanks for the extra technical information.

 

[Ben Brainerd]

This brings up a point I've been trying to hammer home in the post community for a while: You have to build your product for your deliverables. You can't take the same video, slap it on a BR, YouTube, and an SD DVD and expect them all to look the same.

Well, you probably could with BR and YT.

Anywho, my point is that if you know one of your final deliverables is going to be SD, you need to work around that. Much as it pains me, one of the first steps I take for SD is to crank the sharpness WAY up. Basically run it to "Oh god, my eyes" levels on the computer. That usually ends up looking about right on a TV. Your methods/tastes may vary, of course.

 

[Craig Parkes]

Why it looks bad is for multiple reasons. Firstly, SD PAL M2V's for DVD are ALWAYS interlaced, even if they are Progressive footage in an interlaced container, they will look softer on a progressive playout device without some sort of hardware/software enhancement because of this.

Secondly, in compressor you had Frame Controls off and by the sounds of things were going from either a 1024 x 576 square pixel master, or a HD or higher Prores quicktime out of Redcine X.

You either need to have frame controls on for Compressor to do a better resize, or you need to resize to a 720 x 576 master before going to compressor. I prefer doing the resize before going to compressor, because compressor is really slow with Frame Controls on, and also then if you have a video card you can actually check your master quicktime on a screen, so you can compare your ACTUAL before (SD master in native codec) to your DVD after (MPEG-2 compressed M2V).

Next, if you REALLY want to do everything you can to make it look good, you have the option to add compression markers into your quicktime BEFORE running through compressor. Add compression markers between shots with dramatic changes in lighting but are otherwise similar subject matters, or in the middle of fade outs (at the blackest point) and some types of transitions. These compression markers will help ensure you get cleaner fades and avoid bad scene transitions where the compression algorithm fails to pick up a scene change (I have actually had a video which had someone picture in picture doing sign language, combined with a wind blur type transition between two similar scenes completely break the M2V encoding in compressor so that the picture would strobe in and out between good and pixellated every 12 frames (the length of DVD's GOP or Group Of Pictures).

When you know a bit about what's going on at every step it becomes a lot more obvious what's going wrong, what's going right, but more importantly how to test it.

Ultimately, comparing a PAL encoded m2V on a computer screen against a progressive quicktime on a computer screen (even in MPEG-2 codec at the identical bitrate encoded from the exact same progressive source) could still be apples and oranges simply because of how the player deals with the progressive in an interlaced container versus true progressive nature of the two formats.

A DVD player going to an LCD TV on the other hand introduces hardware scalers and deinterlaces specifically designed to deal with this, so you really need to test in the end environment to get a feel for what you are going to see on playout.

You don't have to live with crappy DVD's just because you are going from a HD or higher source (although some issues, such as aliasing and line twitter, can be very tricky to get around when going from a high detail 4K master down to PAL resolution, without doing a reasonable amount of work on those individual problem shots after doing the resize - there is not a completely one size fits all down scaling method). In a practice you'd make compromises, but if you had the time and energy and tools you can get almost always get better results - You just have to make the choice about where to put that work in and what compromises you are willing to make.

 

[jimhare]

Hi Erwin, in addition to the advice above, I have found that adding up to 4 steps of contrast in Compressor brings my Red footage back to life.