1. I've been producing a couple of music videos with the HD101, doing both the filming and editing. I shot the first in HDV 24p, and the second in HDV 25p. During editing I noticed in both videos, that dark backgrounds (flat and larger areas) become blocky. I also noticed that on closeups, human skin lost some texture. All in all it looked like the footage was compressed heavily. I am a Avid 5.2 user, and since Avid does not support the HD100-series (yet), I captured the footage with Ulead Mediastudio 8 (into mpeg2) and converted that into uncompressed quicktime with the dv codec using ProCoder 2. I then imported the uncompressed clips into Avid.

I thought it might have been my workflow of capturing/converting the footage, BUT, when I connected my cam straight to a external monitor, the footage still had some compressed artifacts in it, not as bad as the edited footage, but just doesn't look as clear as lets say dv.

Now, I'm pretty new at using the HD101 camera, so I thought maybe I am doing something wrong on the settings of the cam. F.ex does skin detection, low f.stop, low lights,, shutter speed, zoom etc cause lower quality on the footage? Or is it just the mpeg2 conversion that causes this? I thought HDV would regardless be clearer than DV, specially after downconversion.

2. When capturing HDV, the resolution is 1280x720, so by converting the footage to 720x576 (in PAL land), I have converted it to SD, right?

3. On the HD101 you can choose to shoot between HDV-HD25p and HDV-SD50p. I understand one shoots at 50p the other 25p, but why are they both HDV, when the 50p says SD?

Sorry if I ask wierd questions, but I'm a fairly new HD100/101 user.

- Nima

BUT, when I connected my cam straight to a external monitor, the footage still had some compressed artifacts in it, not as bad as the edited footage, but just doesn't look as clear as lets say dv.
- Nima


Not as clear as DV?
You must have a defective camera, although I've never heard of this problem before.

You just bought this camera, right?
Send it back for another.

In reply to 2 and 3:

2: Yes, your footage is now SD.

3: HDV-HD25P is 1280x720 25 progressive frames per second encoded in HDV (mpeg 2). HDV-SD50P is 720x576 50 progressive frames per second encoded in HDV (mpeg 2). This in comparison to DV: 720x576 50i - 50 fields per second. Obviously, with DV encoding and 50 fields per second, you can encode progressively, but only at a rate of 25 frames per second. In order to fit the extra 25 frames per second of HDV-SD50P onto tape, it has to be compressed more, so it is compressed in HDV and not DV.

Tim:

Thanks for the answers.

On subject 2. Is there any difference in quality when converting hdv to sd with computer software rather than downconverting internally on a camera (f.ex Sony) or via the HD100 Component/Composite output?

subject 3. So if you wanted to f.ex film a footage that you wanted to do slow-mo in post, the HDV-50p would be a better choice? Also, would the HDV50p almost compare to Canon XL1's 50 full frame mode (but whereas Canon's is dv)

- Nima

>Is there any difference in quality when converting
>hdv to sd with computer software rather than
>downconverting internally on a camera

Most definitely. When you downconvert to SD in camera (which you can't do with the HD100), you downconvert to DV, which means losing a lot of color resolution.

As far as the second question, I am unfamiliar with the XL1's 50 frame mode. If you were shooting a standard def project and needed slow motion, shooting in HDV-SD50p would be an excellent idea. Just conform it to 25 fps and you have 50% slomo. If you were shooting for an HD project, HDV-SD50p would have to be uprezzed and things might get a little soft. Not sure how acceptable this would be since I haven't really tried it.

For HD slomo, the Panasonic HVX is a better tool.

So by downconverting from hdv to dv uncompressed quicktime, with computer software (Procoder/Sorensen Squeeze etc) you don't lose quality?

of course you can lose quality plus you re-compressing the footage into another compression codec (mp2) before going to uncompressed.

You should capture the footage properly into the computer then convert to the highest quality codec you can edit with. The best solution would be to capture, convert to uncompressed, downsample to the size you want then edit in a codec that your sytem can handle. Remember that when you size something down, not all conversions are equal. You need to use a high quality code for this, if you do it with more re-compressing you are just going compound any artifacts that exist.

Listen to Tim.

And remember SD and DV are not the same thing.

Tim Gray: SD and DV - not equal. Got it.

Tim Holtermann: I usually capture HDV with CapDVHS, and use Procoder to convert and downsize the HDV into Quicktime Movie uncompressed quicktime 720x576. I've been testing the Quicktime Target with the Avid DNxHD codec (720p/25 8bit) aswell, but it seems like the Quicktime Movie uncompressed comes out best. Does this make sense, or is another codec usually used for optimal results (for quality, not saving hdd space)?

You really can't do better than uncompressed, but editing it isn't always easy. I'd say you may want to convert to uncompressed at full HD before downsizing. Not sure what happens when you downconvert directly from HD to Uncompressed DV at the same time. It would take some experimenting.

>Is there any difference in quality when converting
>hdv to sd with computer software rather than
>downconverting internally on a camera

Most definitely. When you downconvert to SD in camera (which you can't do with the HD100), you downconvert to DV, which means losing a lot of color resolution.


Hi Tim. There will obviously be a loss of spatial resolution going from HDV to DV, but I thought that for a PAL system the colour resolution is not going to suffer since they are both 4:2:0?

Richard

I'm not really familiar with PAL that much. In fact I am just sorting out this color business myself. From the numbers they sound the same. 4:2:0 is 4:2:0, right? I think (correct me if I'm wrong) that the PAL 4:2:0 is different from the MPEG2 4:2:0. I don't really know.

Maybe one of the PAL guys can chime in.

Tim Gray: SD and DV - not equal. Got it.

Tim Holtermann: I usually capture HDV with CapDVHS, and use Procoder to convert and downsize the HDV into Quicktime Movie uncompressed quicktime 720x576. I've been testing the Quicktime Target with the Avid DNxHD codec (720p/25 8bit) aswell, but it seems like the Quicktime Movie uncompressed comes out best. Does this make sense, or is another codec usually used for optimal results (for quality, not saving hdd space)?


Nima the problems you are experiencing almost certainly relate back to your workflow. As you have noted Avid does not fully support 720P25 yet so you need to go through an intermediate codec. Everything here is a compromise: Even if Avid did support it, editing in HDV uses MPEG 2, so working directly with HDV is CPU intensive as each block of 6 frames has to be decoded to and re-encoded to make a simple cut. Downcoverting to SD relies on the quality of your conversion software.

I believe the best workflow would be to convert the .ts file from CapVHS into another HD codec - in Avid's case DnxHD. Avid then allows you to drop clips or even edited timelines into a an SD project timeline. That way you have an HD master and an SD master. You should also save time on your conversion as there is no spatial conversion.

The bad news - this won't work completely on Xpress Pro HD yet. The DnxHD 720P25 codec is the 60Mb 8bit version which isn't supported on Xpress Pro. However 720P24 which is available should work in the 60 and 90Mbit versions (8 and 10bit). Your only other oprion on the Avid is to use DVCProHD codec. If you want to go down the Vegas or PremierePro route then Cineform offers an excellent solution at a reasonable price.

Re: your original question about MPEG artifacts, I would suspect your monitor before anything else - monitors usually pre-process the signal unless they can display 1280 x 720 natively - most of the low-end processors built into monitors introduce there own artifacts when they scale and compress the original signal. The HD100 series camera have the most robust MPEG compression scheme of all 3 hi-end HDV camera's on the market although you can still get errors, they should be less noticeable on this cam.

John

When I capture with CapDVHS, I can only save the file as .mpg and not .ts. Do the different file formats make any difference? I'm just trying to capture as close as native quality. Maybe there are some settings I overlooked that can makes CapDVHS save files in .ts... Also, do you think CapDVHS is a appropiate tool to use for capture for professional products that will be broadcasted? If no, what other tools would be appropiate?

As with the conversion of the dnxhd codecs, is it as simple as using lets say Canopus Procoder, selecting the target Quicktime Movie, selecting Avid dnxhd codec with the 720p 23.976 90 10bit (since Avid doesn't support 720p 25)?
1. does the 23.976 on the codec effect the framerate in any way (since my original footage is 25fps)?
2. will 720p 23.976 90 10bit give better image quality than the 720p 23.976 90/60 8bit?
3. because the end result will have to be SD, should I do the HD-SD downconversion during this same process of Avid dnxhd codec conversion? (downing from 1280x720 to 720x576)

Would you say this workflow is sufficient for achieving broadcast quality?

About the mpeg artifacts/monitor issue, I also think it's the monitor at blame, but I think lowering the detail level on the cam during the next shoot might help some too, as I understood after reading some posts.

Would you say this workflow is sufficient for achieving broadcast quality?
The term "broadcast quality" has lost its meaning a long time ago...:-) Just remember those cell phone videos aired after the London subway bombing. "Broadcast quality" is directly related to content... other than that, the term has little meaning.

my bad

what I meant by "broadcast quality", is a product which will be broadcasted on national tv stations... is there another term for that that I can use?

my bad

what I meant by "broadcast quality", is a product which will be broadcasted on national tv stations... is there another term for that that I can use?
Don't worry about it...it was meant 'tongue-in-cheek'.

As for the video quality, it all depends on the station airing it. Most networks have their own guidelines for "technical specifications" they expect/demand of procured material. Check with your broadcaster.

> When I capture with CapDVHS, I can only save the file as .mpg and not .ts. Do the different file formats make any difference?

No visual difference at all - same resolution, color depth etc. Transport streams (.ts or .m2t) differ from program streams (.mpg) in how the same digital data is packaged, and in some metadata that .ts carrys on camera settings.

BTW, remuxts.exe is a fast and handy way to convert transport streams to program streams, for software than can't handle the former.

John

When I capture with CapDVHS, I can only save the file as .mpg and not .ts. Do the different file formats make any difference? I'm just trying to capture as close as native quality. Maybe there are some settings I overlooked that can makes CapDVHS save files in .ts... Also, do you think CapDVHS is a appropiate tool to use for capture for professional products that will be broadcasted? If no, what other tools would be appropiate?

As with the conversion of the dnxhd codecs, is it as simple as using lets say Canopus Procoder, selecting the target Quicktime Movie, selecting Avid dnxhd codec with the 720p 23.976 90 10bit (since Avid doesn't support 720p 25)?
1. does the 23.976 on the codec effect the framerate in any way (since my original footage is 25fps)?
2. will 720p 23.976 90 10bit give better image quality than the 720p 23.976 90/60 8bit?
3. because the end result will have to be SD, should I do the HD-SD downconversion during this same process of Avid dnxhd codec conversion? (downing from 1280x720 to 720x576)

Would you say this workflow is sufficient for achieving broadcast quality?

About the mpeg artifacts/monitor issue, I also think it's the monitor at blame, but I think lowering the detail level on the cam during the next shoot might help some too, as I understood after reading some posts.

1. Yes - I imagine ProCoder will attempt to interpolate the 25 down to 23.976 - that would be a bad thing. I don't think there is anyway of telling it to just run the footage slower. Maybe if you could force interpretation of your source to 23.976 that would work. Unless you are really experienced at working between film and tape rates I would not go down this path - I would leave the 23.976 frame rate codecs for material you shoot at that frame rate. We really need Avid to pull their finger out and support the 25P frame rates.

2. Not really - the 10bit codec is there if you plan to do a lot of post effects - it will hold up better than the 8bit footage. Your original footage is only 8bit but each time you render an 8bit codec to an 8 bit codec you introduce mathematical errors.

3. Avid's workflow is to just drop the footage into an SD comp and let Avid handle the conversion.

I've looked at the available options in Xpress Pro and unfortunately I just don't have time right now to run any tests for you. I would advise testing the following workflows to see if you can get an acceptable result:

1: a] CapDVHS 1280 x 720/25P Mpeg to AvidDnxHD 1080P/25 codec
b] Work in 1080P/25 Avid project and bring final timeline into PAL project
Some users have reported quite good results uprezzing.
2: a] CapDVHS 1280 x 720/25P Mpeg to Avid 1:1 PAL codec (will only work if you have Mojo) but should give a much better result than converting Mpeg to DV
3: a] Cineform Intermediate capture (trial version available) to .avi file.
b] Convert to Avid DV 25 codec from .avi file
4: Use a hardware converter to capture directly into Xpress Pro in PAL resolution.

Sorry - not much help I know, but I have mentioned ot Avid (a number of times) that they are very slow in implementing 25 support.

I have run some tests uprezzing to 1080P then dropping the edited timeline into a 25P SD project and it seems to work, although Iwas only using a test clip downloaded from the web.

I can see some typical 8 bit errors in a blue sky section of the video when monitoring in SD of Mojo on a CRT - I'm unsure whether these errors are attributable to ProCoder, Avid or the live display through Mojo.

I only used the 1080P 8bit codec so the 10bit codec might work better. But the process works and didn't seem to be introducing any really nasty artifacts like you seemed to get.

John

Thank you for the feedback and taking time to do the testing, I really appreciate it. I will try the options you gave me and get back on how it went when I done.


- Nima