GH2 Any reason not to shoot in SD if the target is DVD?

D

dishan

Well-known member
I recently recorded a speech that was to be delivered to the client on DVD.
I set the camera to record in 1080p, not only because I'm used to it, but also because I'm coming from the world of Canon DSLRs, where the lower resolution resulted in more moire and aliasing.

After I shot it and dealt with the rendering times downsampling the video to 720x480 (and even considered the lanczos algorithm for better looking resizing), it occurred to me that perhaps this isn't necessary with the GH2.

Since the GH2 resamples the sensor instead of line skipping, virtually eliminating moire and aliasing, is there any reason not to shoot standard definition if I know the product is going straight to DVD?
 
General consensus I have always heard is that it's much better for you to control the down sampling from HD to SD rather than letting the camera do it.
 
Funny, in the world of full sized video cameras, I've heard the opposite- all the talk I've found about lanczos algorithm and the horrors of bicubic resizing seemed to imply that we would be better off shooting with SD in the first place. If you didn't, then go through the song and dance or resizing with lanczos, usually requiring quite a few extra steps and intermediate files in post.

Is this general consensus that you speak of regarding specifically DSLR shooting?
 
If the downscaling is done correctly and proper codec is used for editing, the final result will be SD 4:2:2, whereas from the camera one would get only SD 4:2:0. The original 1080p 4:2:0 chroma has resolution of 480 * 270 (1920/4 * 1080/4) whereas a SD 4:2:2 has chroma resolution of 360 * 240 (720/2 * 480/2). If one uses a codec like Blackmagic design MJPEG, one will be editing and grading in Digi Beta SD quality.
 
Ultimate answer...
Test it on your camera. Set up, same lighting, same everything, shoot it in SD, then shoot it in HD and render the HD as a DVD mpeg. See which looks better to you.

Also, It will be much easier to color correct and otherwise process the HD then render that SD rather than trying to do those processes with the limited SD info.
 
gonzo_entertainment said:
Ultimate answer...
Test it on your camera. Set up, same lighting, same everything, shoot it in SD, then shoot it in HD and render the HD as a DVD mpeg. See which looks better to you.

Also, It will be much easier to color correct and otherwise process the HD then render that SD rather than trying to do those processes with the limited SD info.
Click to expand...

This.


I would definitely shoot HD and scale it down, but test it and see what works best for you.
 
Fair enough. In general, does the GH2's IQ get worse the way Canon's does as you downsample more in camera?
I'm under the impression that the GH2's resample system doesn't create the artifacts that Canon does (why folks hate shooting in 720 despite needing it sometimes for slow motion).
 
Aki_Hartikainen said:
If the downscaling is done correctly and proper codec is used for editing, the final result will be SD 4:2:2, whereas from the camera one would get only SD 4:2:0. The original 1080p 4:2:0 chroma has resolution of 480 * 270 (1920/4 * 1080/4) whereas a SD 4:2:2 has chroma resolution of 360 * 240 (720/2 * 480/2). If one uses a codec like Blackmagic design MJPEG, one will be editing and grading in Digi Beta SD quality.
Click to expand...

Would that make it better, then if greenscreening to get it down to SD4:2:2 first before keying, or are the extra pixels more helpful in the key process?
 
If you are shooting SD that is 4:3
Trying to wrangle a 16:9 size from an SD source
is asking for trouble.
I have not seen any 4:3 TV's in use at
any of my clients anymore.
As a matter of fact - most of them
are asking for Blu-Ray or a file on a USB drive.
Many HDTV's can play H.264 files from
a USB thumb drive.
And what will you do when the client asks
for an HD version of the show a few years
from now?
 
If you are using CS5 or higher you can benefit from their amazing resampling algorithms if CUDA is enabled. For one, this will allow you to use more or less lossless pan and zoom. There are several other reasons, the pan and zoom one is the one I use the most. Basically, no reason to use SD, lots of reasons to use HD.
 
HVXguy said:
If you are shooting SD that is 4:3
Trying to wrangle a 16:9 size from an SD source
is asking for trouble.
I have not seen any 4:3 TV's in use at
any of my clients anymore.
Click to expand...

Wait a minute- why are you assuming that SD is only 4:3? Before I started shooting HD, I shot 16:9 480p for DVDs all the time. Even my old DVX100 could do that. You do know that standard definition was widescreen as well, right?

Unless the GH2 has some weird limitation that only records in SD 4:3, but that would be really surprising. They wouldn't change the aspect ratio for that one mode, would they? (I don't have it next to me to check).

As a matter of fact - most of them
are asking for Blu-Ray or a file on a USB drive.
Many HDTV's can play H.264 files from
a USB thumb drive.
And what will you do when the client asks
for an HD version of the show a few years
from now?
Click to expand...

Yes, which is why I almost always shoot in HD. Except for cheap projects like this- I do a lot of work with fundraisers who want to mail out DVDs to potential donors. I'm recording lectures, award acceptance speeches, etc., and then delivering a DVD for them to send out for duplication. Next year, they will want to shoot another one as the material is time-dependent.
The key here is to get something out as quickly as possible that looks good. For something like this, I could totally justify shooting in native 16x9 SD as long as I wouldn't be somehow introducing IQ problems and artifacts.
 
Last edited:
DrDave said:
If you are using CS5 or higher you can benefit from their amazing resampling algorithms if CUDA is enabled. For one, this will allow you to use more or less lossless pan and zoom. There are several other reasons, the pan and zoom one is the one I use the most. Basically, no reason to use SD, lots of reasons to use HD.
Click to expand...

That's cool- I've actually read a lot of about CUDA, I do own CS5 but currently have an ATI Radeon card, so nothing doing yet. I'm actually a little bit annoyed that they only enabled those features for cards that offer the extra processing power. Why can't they just let my CPU handle the bulk of the processing like every other application does? Give me the option to turn it on even if it will be slower (there's nothing about CUDA that makes this possible versus not having it, its just that they figured the rendering and processing times would be too slow for the average CPU to handle on its own).

I'm looking into new hardware at the moment anyway, so that will definitely be on my list.
 
The same quandary happened in audio. I routinely record 96KHz audio now, even though it ends up on a CD. While I still do originate many projects in SD for DVD, I'm starting to do more in HD because it occurred to me that some of my SD footage in the library is now quite old, and still handy - so when I'm doing ONLY HD work, it will be difficult to bring in SD without it being obvious, so why not originate in the best format you can to future proof it?
 
greg sage said:
Would that make it better, then if greenscreening to get it down to SD4:2:2 first before keying, or are the extra pixels more helpful in the key process?
Click to expand...

That is a good question. If the delivery format is SD, then having less suppressed color resolution would help in keying, but if the keying was done for the HD 4:2:0 with colors upsampled to 4:2:2 and finally scaled down to SD, it is difficult to tell without testing and side by side comparison. In theory one will have a bit more color pixels for 1080p 4:2:0 than 480p 4:2:2, but only for uncompressed material. Compression will eat color pixels too.
 
Just buy a GT 240 with 1gb for $45 and edit the info in the approved cards list and you are good to go with CUDA. It is a must.
 
I would film in HD, just to allow for reframing, image stabilization and pan and zoom flexibility. Same reasons why it would be nice to be able to film in 4K
 
But why increase the work and processing time in post when you can frame, pan and zoom in camera?

And I say this as a post guy.
 
DrDave said:
Just buy a GT 240 with 1gb for $45 and edit the info in the approved cards list and you are good to go with CUDA. It is a must.
Click to expand...

I've been considering that. Seems hard to find a 240 with DDR5 and 1GB in that price range tho. And once I'm spending more money, I start looking at other cards with more CUDA cores and my shopping starts going around in circles until I decide I need to buy a new editing workstation anyway. :)
I guess I could go with a cheap 240 + DDR3... It would definitely help my rendering time, and enable CS5's lanczos resizing engine so this wouldn't be nearly as much of an issue anymore.

Still, I'm curious if shooting in lower resolution on the GH2 would somehow degrade the image the way it does on Canon. I'll probably always shoot HD anyway for most of the reasons stated here, but I should test it out and see when I have some free time.
 
You must log in or register to reply here.
Back
Top